JP2003060913A - Image processing method and image processor - Google Patents

Abstract

(57) [Summary] To provide an image processing apparatus that performs compression processing at an appropriate compression ratio. An image processing apparatus determines an amount of suppression of a low-frequency component from a characteristic amount indicating a spread of a luminance distribution, and appropriately compresses a dynamic range. A high-pass filter device 11 is provided at an input unit of the image processing device 10, and a luminance variance calculation unit 17 and a compression coefficient calculation unit 18 are provided in parallel with the high-pass filter device 11. The luminance variance calculation unit 17 calculates the variance of the input image, particularly low frequency components such as light and dark. When the variance is large, that is, when the difference in brightness of the low frequency component is large, the compression coefficient calculation unit 18 increases the compression coefficient A. The high-pass filter device 11 largely suppresses low-frequency components based on the high-pass filter device.
Since the low-frequency component is suppressed based on the dispersion, even if there is a local light-dark portion, it is not affected. Compression processing is performed at an appropriate compression ratio.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing method and an image processing apparatus for compressing a dynamic range of an image. In particular, an image processing method for appropriately compressing without excessive compression based on a variance-related value related to variance, which is a statistical amount of an image in a luminance distribution, and a bimodal coefficient representing a degree of bimodality in the luminance distribution. And its equipment.

[0002]

2. Description of the Related Art When an image having a large difference in light and shade including a shadow and a sunlit area is displayed on a display device having a small dynamic range (gradation), a dark portion in the image is crushed black (black color), or Bright areas may become overexposed (white color). This makes it difficult for the viewer to accurately recognize the object. On the other hand, a method called Homomorphic Fillering for compressing the dynamic range of an image has been proposed. This is a method of focusing on the fact that the information about the illumination intensity such as the shadow of the sun and the unevenness of illumination is the low frequency component of the luminance, and removing only the low frequency component. Specifically, the luminance, which is the product of the reflectance and the illumination intensity of the object, is logarithmically converted into the sum of the logarithm of the reflectance and the logarithm of the illumination intensity, and the low-frequency component of the illumination intensity is removed. After that, the dynamic range is compressed by inverse logarithmic conversion.

In addition, there is an image dynamic range compression processing method disclosed in Japanese Patent No. 2663189. This is a method of calculating a low-frequency image by averaging the luminance within a predetermined range in the vicinity of each pixel of the original image, and processing based on the calculated low-frequency image. Specifically, it is a method of compressing the dynamic range by setting a monotonically decreasing function with the brightness of the low-frequency image as an argument at each pixel and adding the output to the brightness value of the corresponding pixel of the original image. By setting the characteristic of the monotonous decreasing function, it is possible to compress only a desired brightness range such as a bright part and a dark part.

However, the above Homomorphic Fillerin
g and the methods disclosed in Japanese Patent No. 2663189 are methods for suppressing low-frequency components of an image, and when the suppression amount is large, high-frequency components are relatively emphasized, which causes edges to be unclear. May be naturally emphasized. For example, an example of the processing process is shown in FIG. The figure is a one-dimensional luminance distribution in which the horizontal axis is the x-axis and the vertical axis is the luminance. FIG. 17A shows the luminance distribution of the input image, that is, the original image, and FIG.
Is the low frequency component, (c) is the luminance distribution when the low frequency component is suppressed to a small extent, and (d) is the luminance distribution when the low frequency component is greatly suppressed.

As shown in FIG. 17D, it can be seen that when the dynamic range is greatly compressed, large overshoots and undershoots occur at the place where the brightness changes stepwise. Therefore, it is preferable that the suppression amount of the low frequency component, that is, the compression ratio of the dynamic range is the minimum necessary. However, the above Homomo
The method disclosed in rphic Fillering and Japanese Patent No. 2663189 was unable to compress the dynamic range with the minimum necessary proper compression rate in order to process all images in a similar manner. That is, it is not possible to compress the image to a desired dynamic range for an image with a large difference in light and shade in fine weather, or for an image with a small difference in light and dark such as in cloudy weather, suppressing low-frequency components more than necessary Was sometimes overemphasized. That is, there is a problem that the image becomes unnatural. To solve these problems, for example, JP-A-6-303430 and JP-A-9-18209.
3 and JP-A-53535.

The method described in Japanese Patent Laid-Open No. 6-303430 is based on the contrast obtained from the original image, and is disclosed in Japanese Patent No. 266.
The feature of the invention of No. 3189 is to determine the signal application range and magnitude of the monotonically decreasing function. Thereby, the dynamic range compression processing is performed according to the difference between the highest luminance and the lowest luminance of the original image. In addition, JP-A-9-18
The method described in 2093 is based on IIR (Infinite Impulse Res
The dynamic range compression processing is performed based on the low frequency image created by the filter. The compression ratio of the processing is calculated from the histogram of the original image, that is, based on the difference between the maximum brightness and the minimum brightness determined by the histogram. The method described in Japanese Patent Laid-Open No. 11-53535 is to perform dynamic range compression or expansion processing on the basis of a low-frequency image created by an edge-preserving smoothing filter. The compression ratio or the expansion ratio is obtained from the histogram of the original image, that is, from the difference between the highest luminance and the lowest luminance determined by the histogram of the original image.

[0007]

However, JP-A-6-303430, JP-A-9-182093, and JP-A-1 are used.
In any of the methods described in 1-53535, the compression ratio or expansion ratio of the dynamic range is set to the maximum brightness and the minimum brightness of the original image,
Alternatively, it is determined according to the difference between the highest luminance and the lowest luminance determined by the histogram. That is, neither considers the locality of the original image. Therefore, the compression rate or the expansion rate cannot be properly and stably obtained by any of the above methods. Here, the locality means a local dark part or bright part in an image and its temporal change. For example, a local dark part or bright part is a defective pixel included in the image pickup device. Alternatively, it is a bright portion that occurs when the illumination light is partially specularly reflected on the imaging object and enters the imaging device.
Further, the change is, for example, a temporal change in the number of pixels in the local bright portion, which occurs due to a temporal change in the reflection state of the illumination light. If such a local dark part or bright part exists, the difference between the maximum value and the minimum value of the brightness becomes large even in an image of a scene with a small difference in brightness, so the dynamic range is calculated to be large, and as a result, The image is an image in which the compression ratio is set large and edges are emphasized too much.

Among the above problems, the influence of a fixed local bright portion or dark portion that does not change with time, such as a defective pixel, is the number of pixels (frequency) on the high luminance side and the low luminance side of the luminance distribution. It can be avoided by making a judgment. But,
When the number of pixels in the bright portion that is local in a moving image changes with time, it is not possible to stably obtain the difference between the maximum value and the minimum value. For example, when the number of pixels in the local bright portion changes with time, when the total number of pixels in the local bright portion exceeds a predetermined value (threshold value), the dynamic range (maximum value is And the difference between the minimum values) changes greatly. As a result, an unnatural moving image in which the degree of edge enhancement sharply changes with time is obtained.

For example, an image is captured and its luminance distribution is represented by a histogram. In the histogram, the luminance at which the cumulative luminance from the minimum is the predetermined value is the minimum luminance, and the luminance at which the cumulative luminance from the maximum is the predetermined value is the maximum luminance.
FIG. 18A shows a case where the frequency (number of pixels) of local bright areas is small, and the maximum value of the obtained brightness is D1 from the integrated value (cumulative brightness) of the shaded area. That is, the dynamic range is D1-D0. On the other hand, in the histogram of FIG. 18B, the number of pixels in the local bright portion (the total number of defective pixels and the number of genuine bright portion pixels) is large, and the maximum value of the luminance obtained from the accumulated luminance is D2. Becomes That is, the dynamic range is D2-D0. 18 (a), (b)
From this, it can be seen that the dynamic range drastically changes from (D1-D0) to (D2-D0) due to the change in the number of bright area pixels. That is, although the conventional method does not change the dynamic range of the image necessary for grasping the scene,
The dynamic range of the entire image is largely changed by the local change of the bright part or the dark part, and as a result, the compression ratio is largely changed. As a result, there are cases where the dynamic range is compressed more than necessary and edges are overemphasized, or when processing a moving image, the degree of edge emphasis changes with time, resulting in an unnatural moving image. is there. These phenomena make it difficult for the viewer to accurately grasp the scene.

The present invention has been made to solve the above problems, and its object is to determine a compression coefficient in a dynamic range compression process of an image based on a dispersion-related value related to a dispersion of a luminance distribution of the image. However, it is an object of the present invention to provide an image processing method and apparatus for appropriately compressing an image without being affected by a local bright portion or dark portion. Another object of the present invention is to provide an image processing method and apparatus for appropriately compressing an image having a large difference in brightness by determining its compression coefficient based on the bimodality in the brightness distribution of the image. Another object of the present invention is to provide an image processing apparatus which applies the image processing apparatus to a color image and enables dynamic range compression processing without affecting color information. Each of the above objects is an object achieved by each individual invention, and each invention should not be construed as achieving all the objects at the same time.

[0011]

In order to solve the above problems, an image processing method according to claim 1 is an image processing method for compressing a dynamic range by suppressing low frequency components of an input image, It is characterized in that a dynamic range is compressed by obtaining a variance related value related to the variance of the luminance distribution of the input image and suppressing the low frequency component of the input image based on the obtained variance related value. Note that the brightness distribution means a probability distribution or a density distribution of brightness, and refers to a frequency distribution of brightness of an input image or a distribution characteristic such as a histogram. That is, the brightness distribution is a distribution function indicating the existence ratio (density) of a certain brightness x in the input image,
For example, specifically, the ratio of the number of pixels having a certain brightness x to all the pixels is shown. If a rank is provided for a certain brightness x and hierarchized, the frequency distribution, the histogram, etc. are examples of the brightness distribution. The brightness distribution is used in this sense, and does not mean the spatial distribution of the brightness of the input image. Further, the dispersion means a statistical dispersion in the above luminance distribution. And the variance-related value is
It means all variables that have a predetermined functional relationship with the variance. For example, the standard deviation, which is the square root of the variance, is also the variance-related value, and any function related to the variance is also the variance-related value. Hereinafter, in the claims and the present specification, the luminance distribution and the dispersion-related value are used with their meanings.

An image processing method according to a second aspect of the invention is characterized in that, in the image processing method according to the first aspect, the greater the dispersion-related value, the greater the suppression amount of the low frequency component. The image processing method according to claim 3 is an image processing method for compressing a dynamic range by suppressing low-frequency components of an input image, wherein the luminance distribution of the input image is bimodal. The dynamic range is compressed by suppressing the low-frequency component of the input image based on the bimodal coefficient that represents. Also, claim 4
The image processing method according to the third aspect is characterized in that, in the image processing method according to the third aspect, the larger the bimodal coefficient, the larger the suppression amount of the low frequency component.

An image processing method according to a fifth aspect of the present invention is an image processing method for compressing a dynamic range by suppressing low frequency components of an input image, the low frequency image comprising the low frequency components of the input image. Is obtained, and the dynamic range is compressed by suppressing the low-frequency component of the input image based on the characteristic amount indicating the spread of the luminance distribution of the low-frequency image.

The image processing method according to claim 6 is characterized in that the characteristic amount in the image processing method according to claim 5 is a dispersion-related value related to the dispersion of the luminance distribution. The image processing method according to claim 7 is characterized in that the characteristic amount in the image processing method according to claim 5 is a bimodal coefficient representing the degree of bimodality of the luminance distribution. or,
In the image processing method according to claim 8, the characteristic amount in the image processing method according to claim 5 is the difference between the maximum value and the minimum value of the brightness in the brightness distribution, or the ratio between the maximum value and the minimum value. It is characterized by being. Further, in the image processing method according to claim 9, in the image processing method according to any one of claims 6 to 8, the suppression amount of the low frequency component is increased as the characteristic amount is increased. Is characterized by.

An image processing apparatus according to a tenth aspect of the invention is
An image processing apparatus for compressing a dynamic range by suppressing a low frequency component of an input image, comprising: a brightness variance calculating means for calculating a variance related value related to a variance of a brightness distribution of the input image; and a brightness variance calculating means. A range compression unit that compresses the dynamic range by suppressing the low-frequency component of the input image based on the calculated dispersion-related value is provided. The image processing device according to claim 11 is characterized in that the range compression means of the image processing device according to claim 10 is a device that increases the suppression amount of the low-frequency component as the dispersion-related value increases. And

The image processing apparatus according to the twelfth aspect is
An image processing device for compressing a dynamic range by suppressing a low frequency component of an input image, wherein in a luminance distribution of the input image, a bimodal coefficient calculating means for calculating a bimodal coefficient representing a degree of bimodality, A range compression unit that compresses the dynamic range by suppressing the low-frequency component of the input image based on the bimodal coefficient calculated by the bimodal coefficient calculation unit is provided. Also, claim 13
The image processing apparatus according to the present invention is characterized in that the range compression means in the image processing apparatus according to the twelfth aspect increases the suppression amount of the low frequency component as the bimodal coefficient increases.

Further, the image processing apparatus according to claim 14 is
An image processing apparatus for compressing a dynamic range by suppressing low-frequency components of an input image, the low-frequency image extracting means for extracting a low-frequency image composed of the low-frequency components of the input image from the input image; Based on the characteristic amount calculated by the characteristic amount calculating unit for calculating the characteristic amount indicating the spread of the luminance distribution of the low frequency image extracted by the image extracting unit, and the low frequency of the input image in the input image And a range compression unit that compresses the dynamic range by suppressing the component.

An image processing apparatus according to a fifteenth aspect is
The characteristic amount in the image processing apparatus according to claim 14 is a dispersion-related value related to the dispersion of the luminance distribution. An image processing apparatus according to a sixteenth aspect is characterized in that the characteristic amount in the image processing apparatus according to the fourteenth aspect is a bimodal coefficient representing the degree of bimodality of the luminance distribution. The image processing device according to claim 17 is the image processing device according to claim 1.
The characteristic amount in the image processing device described in item 4 is the difference between the maximum value and the minimum value of the brightness in the brightness distribution, or the ratio between the maximum value and the minimum value. The image processing device according to claim 18 is characterized in that the suppression amount of the low frequency component is increased as the characteristic amount of the image processing device according to any one of claims 15 to 17 is increased. And The image processing device according to claim 19 is the image processing device according to claim 1.
The image processing apparatus according to any one of claims 0 to 18, characterized in that the range compression means has an average brightness correction means for correcting the average brightness of the image after suppressing the low frequency components.

The image processing apparatus according to claim 20 is
The image processing device according to claim 10 or 11, wherein the range compression means determines at least a compression coefficient of the dynamic range based on the dispersion-related value obtained by the brightness dispersion calculation means, and an input image. The low-frequency component suppressing means for suppressing the low-frequency component and the suppression characteristic setting means for setting the characteristics of the low-frequency component suppressing means based on the compression coefficient.

The image processing apparatus according to the twenty-first aspect is:
The image processing apparatus according to claim 12 or 13, wherein the range compression means determines at least a compression coefficient of a dynamic range based on the bimodal coefficient calculated by the bimodal coefficient calculation means, and an input. It is characterized by having low-frequency component suppressing means for suppressing the low-frequency component of the image, and suppression characteristic setting means for setting the characteristics of the low-frequency component suppressing means based on the compression coefficient.

An image processing apparatus according to the twenty-second aspect is:
The image processing device according to any one of claims 14 to 18, wherein the range compression means determines at least the compression coefficient of the dynamic range based on the characteristic amount calculated by the characteristic amount calculation means. And low-frequency component suppressing means for suppressing low-frequency components of the input image, and suppression characteristic setting means for setting the characteristics of the low-frequency component suppressing means based on the compression coefficient thereof.

An image processing apparatus according to a twenty-third aspect of the invention is
The image processing device according to any one of claims 20 to 22, wherein the range compression means sets a correction addition value based on the compression coefficient and the luminance average obtained by the compression coefficient calculation means. It is characterized by further comprising an average luminance correction means comprising a means and an adder for adding the corrected addition value to all pixels after the suppression processing by the low frequency component suppression means. The image processing device according to claim 24 is the image processing device according to any one of claims 20 to 23, wherein the range compression means sets a correction multiplication value based on a compression coefficient. It is characterized by comprising a setting means and a multiplier for multiplying all the pixels after the suppression processing by the low frequency component suppressing means by the correction multiplication value.

The image processing apparatus according to the twenty-fifth aspect is:
The image processing apparatus according to any one of claims 20 to 24, characterized in that the range compression means is provided with a logarithmic conversion device as an input, and an output of the logarithmic conversion device is used as an input image. An image processing apparatus according to a twenty-sixth aspect is the image processing apparatus according to any one of the twentieth to twenty-fifth aspects, wherein the low frequency component suppressing means is a high pass filter device.

The image processing apparatus according to the twenty-seventh aspect is:
The image processing apparatus according to claim 26, wherein the high-pass filter device comprises a high-pass filter means and an image memory means. An image processing apparatus according to claim 28 is the image processing apparatus according to any one of claims 20 to 25, wherein the low-frequency component suppressing means is a low-pass filter device and a low-pass filter device after the low-pass filter processing. It is characterized by comprising a multiplier for multiplying each luminance by the suppression coefficient set by the suppression characteristic setting means, and a subtractor for subtracting the multiplication result of the multiplier from the input image.

The image processing apparatus according to claim 29 is
The image processing device according to any one of claims 20 to 25, wherein the low-frequency component suppressing unit is a low-pass filter device and a non-linear correction that non-linearly corrects each luminance after the low-pass filter processing based on the compression coefficient. And means for subtracting the non-linearly corrected image from the input image. An image processing apparatus according to claim 30 is the image processing apparatus according to any one of claims 10 to 29, characterized in that the input image is a grayscale image.

Further, the image processing apparatus according to claim 31,
The image processing device according to any one of claims 10 to 29, wherein the input image is a color image, and a dynamic range is compressed with respect to a luminance component obtained from the color image. In addition, claim 32
The image processing device according to claim 31 is the image processing device according to claim 31, wherein the range compression device decomposes the RGB data of the color image into a luminance component and corrected RGB data obtained by dividing the RGB data by the luminance component. And a decoding device for performing dynamic range compression on the luminance component, and for obtaining RGB data compressed for the dynamic range by the product of the compressed luminance component and the corrected RGB data.

[0027]

In the image processing method according to the first aspect of the present invention, first, the dispersion related value related to the dispersion of the luminance distribution of the input image is obtained. What is variance, which is a specific example of variance-related values?
It means the statistical dispersion in the above luminance distribution, represents the root mean square of the difference between each luminance and the average luminance, and represents the degree of spread of the luminance distribution. For example, in a histogram in which brightness is plotted on the horizontal axis and frequency is plotted on the vertical axis, this is an amount that represents the mean square of the average brightness of the histogram and the difference between the brightnesses, and indicates the spread of the histogram distribution. For example, when the illumination is uniform as shown in FIG. 19A, the histogram has a narrow distribution width, for example, a half-value width L ₁ . In image processing,
It becomes necessary for a person to be able to grasp an object to be recognized in an environment where there is a difference in light and shade such as shadows, uneven lighting, and the like (hereinafter, referred to as lighting by combining sun rays and artificial lighting). In general, the brightness of illumination changes spatially with a long period, and thus becomes a low-frequency component in an image. On the other hand, the contour image of the object due to the reflection from the object becomes a high frequency component because the brightness changes in a short period. The brightness of the input image is the product of the reflectance of the object and the illumination intensity. Therefore, in the luminance histogram when the illumination intensity is not uniform, the histogram of the reflectance component of the object shifts to the high luminance side in the bright part due to the illumination, and the histogram of the reflectance factor of the object shifts to the low luminance side in the dark part due to the illumination. Will be shifted to. If a histogram is created for the luminance distribution of an input image having a difference in the brightness of illumination, the bright portion due to illumination appears on the high luminance side of the histogram and the dark portion due to illumination appears on the low luminance side. That is, when there is a brightness difference between the illumination intensities, the spread of the histogram is larger than when it does not exist.
That is, the dispersion of the luminance distribution is related to the existence and the degree of the brightness difference of the illumination intensity. For example, considering the half-value width of the histogram, the half-value width L ₂ is the half-value width L ₁ when there is a brightness difference due to the illumination intensity as shown in FIG. 19B.
Will be larger than.

The present invention pays attention to this relationship and suppresses the low frequency component based on the magnitude of the dispersion related value in the luminance distribution of the input image. Suppression of low frequency components is, for example, high-pass filter processing of an input image. As a result, low frequency components are suppressed and the dynamic range is compressed. Since the above suppression is based on the magnitude of the dispersion-related value, even if there are locally high-luminance pixels or locally low-luminance pixels, their influence is mitigated. That is, the low frequency component can be suppressed without being affected by the local brightness. That is, the compression processing can be performed without causing excessive suppression or insufficient suppression as in the conventional case. Specifically, the dynamic range can be compressed by appropriately eliminating a large difference in brightness of illumination.

The image processing method according to claim 2 is characterized in that, in the image processing method according to claim 1, the larger the dispersion-related value, the larger the suppression amount of the low frequency component. The magnitude of the dispersion-related value is substantially proportional to the magnitude of the difference in brightness of the low frequency component. Therefore, the dynamic range can be effectively compressed by increasing the suppression amount of the low frequency component as the dispersion-related value increases.

Further, in the image processing method according to the third aspect, a bimodal coefficient representing the degree of bimodality is obtained in the luminance distribution of the input image. This is because, for example, when the illumination intensity (low frequency component) is uniform in the input image as shown in FIG. 20 (FIG. 20 (a)), the histogram often has a mountain shape having one peak A. When the illumination intensity is divided into two parts, a bright part and a dark part, and the difference in brightness is large (Fig. 20 (b)).
Is because the histogram has a bimodal shape having _two peaks B ₁ and B ₂ . The bimodal coefficient representing the degree of bimodality is, for example, “Pattern recognition theory and application” (pp65-6
9 Asakura Shoten 1996), which can be represented by the discrimination reference value η for the automatic threshold selection method for image binarization.

For example, the discrimination reference value η is obtained by binarizing an image with a certain threshold value k and dividing each pixel into two classes, class 1 and class 2, between-class variance σ _B and total pixel variance σ.
_As the ratio of _T ,

## EQU1 ## η = σ _B ² / σ _T ² ... (1) Where the interclass variance σ _B ² is the class
Using the mean μ _{1 of 1} and the occurrence probability ω ₁ , the mean μ _{2 of} class ₂ and the occurrence probability ω ₂ ,

## EQU2 ## σ _B ² = ω ₁ ω ₂ (μ ₁ −μ ₂ ) ² ... (2) The discriminant reference value η _max , which becomes the maximum when the binarization threshold value k is changed, is a constant representing the goodness of separation into two classes, and has a magnitude of 0 to 1. That is,
The histogram has two peaks, and the smaller and higher the two peaks, the larger the discrimination reference value η _max becomes. When the image is a binary image, the discrimination reference value η _max is 1. Moreover, the curve of the luminance distribution has one peak, and the smaller the peak is, the smaller the discrimination reference value η _max becomes. That is, the discrimination reference value η _max can be a bimodal coefficient that represents the degree of bimodality of the luminance distribution.

That is, by examining the bimodal coefficient, it is possible to estimate the spread of the distribution of the low frequency component, for example, the existence of the bright portion and the dark portion of the illumination intensity and the difference between the bright and dark portions. Therefore, based on this bimodal coefficient, for example, a low pass component of the input image is suppressed by a high pass filter. That is, on the histogram, the two peaks B ₁ and B ₂ are processed so as to approach each other as shown in FIG. Desirably, a low frequency component is suppressed so that these two peaks may overlap and form one peak A. This effectively compresses the dynamic range of the input image in which the difference between the bright and dark areas of the illumination intensity is large.

According to the image processing method of the fourth aspect, in the image processing method of the third aspect, the larger the bimodal coefficient, the larger the suppression amount of the low frequency component. The magnitude of the bimodal coefficient is also substantially proportional to the brightness difference of the illumination intensity. this is,
The larger the bimodal coefficient, the larger the difference in brightness of the low frequency component. Therefore, as the bimodal coefficient is larger, the dynamic range can be effectively compressed by increasing the suppression amount of the low frequency component.

According to the image processing method of the fifth aspect, first, a low frequency image composed of low frequency components of the input image is obtained. This is because the difference in brightness such as uneven illumination is mainly included in the low frequency component. Then, a characteristic amount indicating the spread of the luminance distribution of the low frequency image is obtained. The characteristic amount is an amount corresponding to the brightness difference of the low frequency component, for example, an amount related to the shape and size of the histogram. Since the characteristic amount corresponds (substantially proportional) to the contrast difference of the low frequency component, if the low frequency component of the input image is suppressed based on the characteristic amount, the dynamic range can be compressed more effectively. it can.

According to the image processing method of the sixth aspect, the characteristic amount in the image processing method of the fifth aspect is used as a dispersion-related value in the luminance distribution of the low frequency image. The dispersion-related value of the luminance distribution of the low-frequency image corresponds (substantially proportional) to the contrast difference of the low-frequency component. Therefore, if the low-frequency component is suppressed based on the dispersion-related value of the low-frequency component, it will not be affected by the high-frequency image containing many reflectance components of the object,
It is possible to more reliably compress the brightness difference of the low-frequency component, especially the brightness difference depending on the illumination intensity. That is, the dynamic range can be compressed more accurately.

In the image processing method according to the seventh aspect, the characteristic amount in the image processing method according to the fifth aspect is a bimodal coefficient representing the degree of bimodality of the luminance distribution. The bimodal coefficient of the luminance distribution of the low-frequency image indicates the degree of separation between the bright portion and the dark portion in the histogram, which is an example of the luminance distribution. Can be shown. Therefore, if the low-frequency component is suppressed based on the bimodal coefficient of the luminance distribution of the low-frequency image, the brightness difference of the low-frequency component, in particular, without being affected by the high-frequency image containing many reflectance components of the object, A low frequency component (brightness / darkness difference) can be more accurately suppressed from an input image having a large brightness difference due to illumination and clearly divided into a light part and a dark part. That is, it is possible to effectively compress the dynamic range of the input image having a large difference in brightness.

The image processing method according to claim 8 is characterized in that the characteristic amount in the image processing method according to claim 5 is defined as the difference between the maximum value and the minimum value of the brightness in the brightness distribution, or the maximum value and the minimum value. It is the ratio with the value. That is, the difference between the maximum value and the minimum value of the brightness in the brightness distribution of the low-frequency image or the ratio between the maximum value and the minimum value of the brightness is obtained and used as the characteristic amount. The difference between the maximum value and the minimum value of the brightness in the brightness distribution of the low-frequency image or the ratio between the maximum value and the minimum value of the brightness is a parameter that directly indicates the brightness difference of the low-frequency component. or,
At this time, since the high-frequency image such as the local difference in brightness is removed, the above-mentioned direct parameters are not affected. Therefore, the low frequency component is suppressed based on the difference between the maximum value and the minimum value of the brightness in the brightness distribution of the low frequency image or the ratio between the maximum value and the minimum value of the brightness (a parameter that directly indicates the contrast). Then, the same effect as that of the image processing method according to claim 6 or 7 can be obtained. Also, the difference between the maximum value and the minimum value of the luminance in the luminance distribution of the low frequency image,
Alternatively, the calculation of the ratio between the maximum value and the minimum value of the brightness requires a smaller amount of calculation than the dispersion-related value, the bimodal property, and the like. Therefore, the image processing time can be shortened.

An image processing method according to a ninth aspect is the image processing method according to any one of the sixth to eighth aspects, in which the larger the characteristic amount, the greater the suppression amount of the low frequency component. ing. The characteristic amount in the image processing method according to any one of claims 6 to 8 is substantially proportional to the difference in brightness of the low frequency component. That is, the larger the characteristic amount, the larger the difference in brightness of the low frequency component. This indicates that there is room for reducing the difference in brightness of the low-frequency component, for example, the difference in brightness due to the illumination intensity, as the characteristic amount increases. Therefore, if the amount of suppression of the low frequency component is increased as the characteristic amount is increased, the dynamic range can be compressed more effectively.

Further, according to the image processing apparatus of the tenth aspect, the luminance variance calculating means calculates the variance related value in the luminance distribution of the input image. The variance, which is a specific example of the variance-related value, is the mean square of the luminance and the mean square of the difference between the luminances, and indicates the spread of the luminance distribution. That is, if the luminance distribution is expressed in a distributed manner, the influence of locally high luminance pixels or locally low luminance pixels is mitigated. Then, the range compression unit suppresses the low-frequency component of the input image in the input image based on the dispersion-related value. The range compression means is, for example, a high pass filter. For example,
The low-frequency component is suppressed by the frequency characteristics such as the cutoff frequency of the high-pass filter and the characteristics such as the attenuation rate. That is, the dynamic range of the input image is compressed. Since the low frequency component is suppressed based on the dispersion-related value, the image processing apparatus can compress the image without being affected by the local (small number of pixels) brightness difference.

According to the image processing apparatus of the eleventh aspect, the range compression means in the image processing apparatus of the tenth aspect is configured such that the suppression amount of the low frequency component is increased as the dispersion-related value increases. There is. The dispersion-related value is substantially proportional to the low-frequency component, for example, the difference in brightness of the illumination intensity. This means that the larger the dispersion-related value, the more room for suppressing low-frequency components. Therefore, if the range compression means greatly suppresses the suppression amount of the low frequency component according to the dispersion-related value, the dynamic range can be compressed more.

According to the image processing device of the twelfth aspect, the bimodal coefficient calculating means calculates the bimodal coefficient representing the degree of bimodality in the luminance distribution of the input image. The bimodal coefficient is a parameter indicating the degree of separation of a bright portion and a dark portion in a luminance distribution, for example, a luminance distribution represented by a histogram. In particular, when the brightness difference of the low frequency component is large, for example, when the brightness is clearly divided into two parts, that is, the bright part and the dark part due to the intensity difference of the illumination, it is a parameter that indicates it more faithfully. Therefore, if the range compression means suppresses the low-frequency component of the input image based on the bimodal coefficient calculated by the bimodal coefficient calculating means, for example, an input having a large difference in brightness of the low-frequency component such as a brightness difference due to illumination is input. The low frequency component can be suppressed more effectively for the image. That is, the image processing apparatus effectively compresses the dynamic range of an input image having a large difference in brightness.

The image processing apparatus according to claim 13 is
The range compression unit in the image processing apparatus according to the twelfth aspect suppresses the low-frequency component to a greater extent as the bimodal coefficient increases. The bimodal coefficient is substantially proportional to the low-frequency component of the input image, for example, the brightness difference of the illumination intensity. Therefore, if the low-frequency component of the input image is suppressed to a large extent as the range compression unit has a large bimodal coefficient, the low-frequency component can be suppressed more effectively. That is, the image processing apparatus can compress the dynamic range more effectively.

According to the image processing apparatus of the fourteenth aspect, the low-frequency image extracting means extracts from the input image a low-frequency image composed of low-frequency components of the input image. Then, the characteristic amount calculating means calculates the characteristic amount indicating the spread of the luminance distribution of the extracted low frequency image. The characteristic amount is, for example, an amount corresponding to the brightness difference of the low frequency component, for example, an amount related to the shape and size of the histogram, and corresponds to the brightness difference of the low frequency component, for example, the brightness difference of the illumination intensity (substantially proportional). ) Is the amount.

Conventionally, a luminance distribution, for example, a histogram is created from an original image, the maximum luminance and the minimum luminance of the luminance distribution are calculated, and the low frequency component is suppressed based on them. In this case, the dynamic range may be excessively compressed due to the local influence of the bright portion or the dark portion. According to the present invention, the low frequency component is once obtained from the input image by the low frequency component image extracting means, and the characteristic amount indicating the spread of the luminance distribution of the low frequency component is obtained. Therefore, if the range compression unit suppresses the low frequency component of the input image based on the characteristic amount, it can be suppressed without being affected by the local bright portion or dark portion. Therefore, there is no excessive dynamic range compression. Even if a local bright part or dark part is mixed in the low-frequency image,
Statistical quantities such as variance and bimodality are not affected by the local frequency of light or dark areas.

The image processing apparatus according to claim 15 is
An image processing apparatus according to claim 14, wherein the characteristic amount is a dispersion-related value of a luminance distribution. The dispersion-related value of the luminance distribution of the low-frequency image corresponds (substantially proportional) to the contrast difference of the low-frequency component. Therefore, if the range compression means suppresses the low-frequency component based on the dispersion of the low-frequency component, without being affected by the high-frequency image containing many reflectance components of the object,
It becomes a device that compresses the dynamic range more accurately.

The image processing apparatus according to claim 16 is
An image processing apparatus according to claim 14, wherein the characteristic amount is a bimodal coefficient representing the degree of bimodality of the luminance distribution.
The bimodal coefficient indicates the degree of separation between the bright part and the dark part in the luminance distribution represented by a histogram or the like, and when the difference in light and dark of the low frequency component is large, for example, the two parts of the bright part and the dark part depending on the illumination intensity. When the area is clearly divided and the difference in brightness is large, the difference in brightness can be faithfully shown. Therefore, if the range compression unit suppresses the low-frequency component based on the bimodal coefficient, the range-compression unit can suppress the low-frequency component from the input image having a large difference in brightness and darkness without being affected by the high-frequency image containing many reflectance components of the object. The low frequency component can be suppressed accurately. That is, the image processing apparatus can more accurately compress the dynamic range of the input image having a large difference in brightness.

The image processing apparatus according to claim 17 is
The characteristic amount in the image processing apparatus according to claim 14 is the difference between the maximum value and the minimum value of the brightness in the brightness distribution, or the ratio between the maximum value and the minimum value of the brightness. The difference between the maximum value and the minimum value of the brightness in the brightness distribution of the low-frequency image or the ratio between the maximum value and the minimum value of the brightness is a parameter that directly indicates the brightness difference of the low-frequency component. Further, at this time, since the high frequency component is removed, the direct parameter is not affected by the object image (high frequency image). Therefore, the range compression means reduces the low value based on the difference between the maximum value and the minimum value of the brightness in the brightness distribution of the low-frequency image, or the ratio between the maximum value and the minimum value of the brightness (a parameter that directly indicates the brightness difference). Claim 15 or claim 16 if the frequency component is suppressed.
The image processing apparatus has the same effect as the image processing apparatus described in (1). Further, the calculation of the difference between the maximum value and the minimum value of the brightness or the ratio between the maximum value and the minimum value of the brightness requires a smaller amount of calculation than the variance, the bimodal property and the like. Therefore, the image processing apparatus is capable of high-speed compression processing.

The image processing apparatus according to claim 18 is
In the image processing apparatus according to any one of claims 15 to 17, the amount of suppression of low frequency components is increased as the characteristic amount is increased. The characteristic amount in the image processing apparatus according to any one of claims 15 to 17 is substantially proportional to the difference in brightness of the low frequency component. That is, the larger the characteristic amount, the larger the difference in brightness of the low frequency component. This indicates that the larger the characteristic amount, the more room there is for reducing the contrast difference of the low frequency component. Therefore, the larger the characteristic amount is, the larger the suppression amount of the low frequency component is, and the larger the image processing apparatus can be, which can appropriately compress the dynamic range.

The image processing apparatus according to claim 19 is
The image processing apparatus according to any one of claims 10 to 18, wherein the average luminance correction means corrects the average luminance of the image after the range compression means has suppressed the low frequency component. This is because suppressing the low frequency component by the range compression means may reduce the average brightness. Therefore, when the average brightness is substantially the same before and after the process, the average brightness correction means adds a predetermined value to all pixels. This allows the input image and the output image to have substantially the same brightness (luminance average). By using the image processing apparatus of the present invention, it is possible to obtain an image that is easily visible even after compression processing.

The image processing apparatus according to claim 20 is
The image processing device according to claim 10 or 11, wherein the range compression means determines at least a compression coefficient of a dynamic range based on the dispersion-related value obtained by the brightness dispersion calculation means, It has low frequency component suppressing means for suppressing the low frequency component of the input image, and suppression characteristic setting means for setting the characteristics of the low frequency component suppressing means based on the compression coefficient. The compression coefficient calculation means determines at least the compression coefficient of the dynamic range based on the dispersion-related value obtained by the luminance dispersion calculation means. This decision is made, for example, by the function according to FIG. The function is, for example, a monotonically decreasing function up to a predetermined value σ _{1 of} variance σ ² and a constant value above a predetermined value σ ₁ .

Then, the suppression characteristic setting means sets the characteristics of the low frequency component suppressing means based on the compression coefficient. The low frequency component suppressing means is, for example, a filter device, and its characteristics are frequency characteristics such as cutoff frequency and characteristics such as attenuation rate. This characteristic is changed. Then, the low frequency component suppressing means suppresses the input image with its characteristics. In this way, the range compression means sets the characteristic (compression rate) of the low frequency component suppression means based on the dispersion-related value that is a statistic. Therefore, according to the present invention, the image processing apparatus can compress the dynamic range at a stable compression rate without excess or deficiency of the image without being affected by the local small number of pixels.

The image processing apparatus according to the twenty-first aspect is:
The image processing device according to claim 12 or 13, wherein the range compression means determines at least a compression coefficient of a dynamic range based on the bimodal coefficients calculated by the bimodal coefficient calculation means. It has a low frequency component suppressing means for suppressing the low frequency component of the input image, and a suppression characteristic setting means for setting the characteristic of the low frequency component suppressing means based on the compression coefficient. The compression coefficient calculation means of the range compression means determines at least the compression coefficient of the dynamic range based on the bimodal coefficients obtained by the bimodal coefficient calculation means. This decision is made, for example, by the function shown in FIG. The function is, for example, a monotone decreasing function up to a predetermined value η ₁ which is a bimodal coefficient-related value (discrimination reference value) η _max , and a constant value when the predetermined value η ₁ or more.

Then, the suppressing characteristic setting means sets the characteristic of the low frequency component suppressing means based on the compression coefficient. For example, the low frequency component suppressing means is a filter device, and its characteristics are frequency characteristics such as cutoff frequency and characteristics such as attenuation rate. This characteristic is changed. Then, the low frequency component suppressing means suppresses the input image with its characteristics. In this way, the range compression means sets the characteristic (compression rate) of the low frequency component suppression means based on the bimodal coefficient. Therefore, when the range compression means of the present invention is applied to the image processing apparatus, the dynamic range can be constantly and stably compressed without excess or deficiency of the image without being affected by local bright and dark parts. In particular, since it is based on the bimodal coefficient, it is particularly effective for an input image having a large difference in brightness of low frequency components, for example, an input image in which a bright part and a dark part due to illumination are clearly divided and the brightness difference is large. The dynamic range can be compressed.

The image processing apparatus according to the twenty-second aspect is:
The image processing device according to any one of claims 14 to 18, wherein the range compression means determines at least a compression coefficient of a dynamic range based on the characteristic amount calculated by the characteristic amount calculation means. It has a calculating means, a low frequency component suppressing means for suppressing the low frequency component of the input image, and a suppressing characteristic setting means for setting the characteristic of the low frequency component suppressing means based on the compression coefficient.

The compression coefficient calculating means determines at least the dynamic range based on the characteristic amount calculated from the low frequency image by the characteristic amount calculating means of the image processing apparatus according to any one of claims 14 to 18. Determine the compression factor. Further, the suppression characteristic setting means, based on the compression coefficient,
Controls the characteristics of the low frequency component suppressing means. The characteristic control is, for example, control of the filter characteristic when the low frequency component suppressing means is a high-pass filter. Then, the low frequency component suppressing means suppresses the low frequency component of the input image. In this way, the range compression unit sets the characteristic of the low frequency component suppressing unit based on the characteristic amount calculated from the low frequency image. Therefore, when the range compression means of the present invention is applied to the image processing apparatus, the dynamic range of the image can be compressed without being affected by the high frequency image which is the reflectance component of the object and the local brightness. In particular, since the dynamic range can be compressed without being affected by a high frequency image and local light and dark, there is no excess or deficiency in the compression. The image processing device is capable of stable compression.

The image processing apparatus according to claim 23 is
The image processing device according to any one of claims 20 to 22, wherein the range compression unit sets a correction addition value based on the compression coefficient and the luminance average obtained by the compression coefficient calculation unit. It further has an average brightness correction unit including an addition value setting unit and an adder that adds the corrected addition value to all pixels after the suppression processing by the low frequency component suppression unit. In this configuration, the addition value setting means sets the correction addition value based on the compression coefficient and the luminance average obtained by the compression coefficient calculation means. Then, the adder adds the corrected addition value to all pixels after the suppression processing by the low frequency component suppression means.

This is because when the suppression processing is performed by the low frequency component suppressing means, the average brightness of the processed image is lower than the average brightness of the input image, and the entire image becomes dark. Therefore, in the case where the average brightness is substantially the same before and after the processing, the above-described corrected addition value is added to all pixels by the adder. This allows the input image and the output image to have substantially the same brightness (luminance average). In some cases, the brightness of the input image itself (original image) may be low overall. Even in that case, the present invention is effective. That is, if the correction addition value is added to all the pixels, the image after the compression processing can be made an image of a level that is easily visible.

The image processing apparatus according to the twenty-fourth aspect is
The image processing apparatus according to any one of claims 20 to 23, wherein the range compression means sets a correction multiplication value based on a compression coefficient, and a low frequency component suppression means suppresses the multiplication. And a multiplier that multiplies all processed pixels by a correction multiplication value. In this configuration, the multiplication value setting means sets the correction multiplication value based on the compression coefficient.
Then, the multiplier multiplies the corrected multiplication value by all the pixels after the suppression processing by the low frequency component suppression means. This is because, for example, the input image itself (original image) may be an image that does not need to be compressed or a low-contrast image. Also, the range of the output image may be smaller than the range of the display device. In this case, all the pixels are multiplied by the corrected multiplication value. As a result, the range of the display device to be output is effectively used. Also, the contrast of the output image can be increased. That is, for example, it is possible to provide an image that is more optimally adjusted according to the dynamic range of the display device.

The image processing apparatus according to the twenty-fifth aspect is:
The image processing apparatus according to any one of claims 20 to 24, wherein the range compression means includes a logarithmic conversion device at an input. This is because the brightness of the captured image is given as the product of the reflectance component of the object and the illumination intensity component.
Therefore, if the input image is logarithmically converted by the logarithmic conversion device, each luminance of the image becomes the sum of the reflectance component (high frequency component) and the illumination intensity component (low frequency component) of the object. That is, in the luminance distribution, the object information (high frequency component) is linearly offset by the illumination intensity component (low frequency component). Range compression is performed on the image in that state. That is, the compression processing is performed by subtracting the low frequency component by the range compression means. Therefore, the low frequency component can be suppressed more accurately. That is, the object reflectance (object information) can be extracted more accurately.

The image processing apparatus according to claim 26 is
The image processing device according to any one of claims 20 to 25, wherein the low-frequency component suppressing unit is a high-pass filter device. The high-pass filter device is, for example, a filter including a DSP (Digital Signal Processor) or the like. Since the filter characteristic is programmable, the suppression amount of low frequency components can be easily reduced. Further, by controlling the filter characteristic, the compression rate can be easily controlled.

The image processing apparatus according to the twenty-seventh aspect is
In the image processing device according to the twenty-sixth aspect, the high-pass filter device comprises high-pass filter means and image memory means. In this configuration, the high-pass filter means is, for example, a filtering process in which an input signal is temporarily stored in an image memory and the captured image is processed by software. That is, processing equivalent to that of the high-pass filter device according to claim 26 is performed by software. For example, a high-frequency emphasis filter represented by a matrix is operated as an image memory that is an image memory unit.
This suppresses the low frequency components of the image. Since the high-pass filter device is composed of the image memory and the software (program), it can be flexibly set according to various conditions such as the display device of the output destination. That is,
The image processing device has excellent flexibility. Incidentally, the high-pass filter means by software is not limited to the calculation of the filter expressed by the matrix. For example, the entire image may be Fourier transformed, low frequency components may be removed, and then inverse Fourier transform may be performed. The high-pass filter means includes all filtering processing that suppresses low frequency components using an image memory.

The image processing apparatus according to claim 28 is
The image processing apparatus according to any one of claims 20 to 25, wherein the low-frequency component suppressing unit is set by a low-pass filter device and each luminance and suppression characteristic setting unit after the low-pass filtering process. It is composed of a multiplier that multiplies the suppression coefficient and a subtracter that subtracts the multiplication result of the multiplier from the input image. In the above configuration, the low pass filter device extracts the low frequency component from the input image and generates the low frequency image. Then, the multiplier multiplies the low frequency image by the suppression coefficient set by the suppression characteristic setting means. This controls the amount of suppression. Then, the subtracted image subtracts the multiplied image from the input image. This allows
The low frequency image is subtracted (suppressed) from the input image according to the multiplication coefficient. That is, it becomes a high-pass filter device.
Therefore, an effect equivalent to that of the image processing apparatus of claim 26 or claim 27 is exhibited.

Further, the image processing apparatus according to claim 29,
The image processing device according to any one of claims 20 to 25, wherein the low-frequency component suppressing unit is a low-pass filter device, and each luminance after the low-pass filter process is nonlinearly corrected based on a compression coefficient. It comprises a non-linear correction means and a subtractor for subtracting the non-linearly corrected image from the input image. This configuration changes the suppression characteristic of the low frequency component suppression means in a non-linear manner according to the brightness. For example, in the image processing device according to the twenty-eighth aspect, the low-frequency image is multiplied by a constant suppression coefficient and subtracted from the input image. That is, the low frequency image is linearly subtracted. In this case, the effect is also linear.

Instead of this, the non-linear correction means non-linearly corrects each luminance after the low-pass filter processing. The non-linear correction means is, for example, a reference table and a brightness conversion device that are stored such that the suppression coefficient increases as the brightness increases. As a result, each luminance after the low-pass filter processing is non-linearly converted. Then, the non-linearly transformed image is subtracted from the input image by the subtractor (non-linear correction). In this way, the dynamic range is compressed more as the brightness increases. That is, for example, by storing a suppression coefficient that changes non-linearly according to the brightness in a reference table and converting the brightness according to the suppression coefficient, the degree of compression can be changed according to the brightness level. That is, the image processing apparatus can adjust the degree of compression of the dynamic range to a desired amount according to the brightness level and perform compression processing.

The image processing apparatus according to claim 30 is
The image processing apparatus according to any one of claims 10 to 29, wherein the input image is a grayscale image. In a grayscale image, the density of each pixel is almost equal to the brightness of each pixel. Therefore, it is possible to realize the image processing apparatus according to any one of claims 10 to 29 that processes the brightness.

The image processing apparatus according to claim 31 is
The image processing device according to any one of claims 10 to 29, wherein the input image is a color image, and the dynamic range is compressed with respect to the luminance component obtained from the color image. The image processing apparatus according to any one of claims 10 to 29, wherein, for example, when the input image is a color image, the input image is decomposed into R, G, and B,
Compress the dynamic range for each color. And
If each color after processing is output to the color image display device, the dynamic range can be compressed even for a color image. Also, an input image is decomposed into a luminance signal and a color difference signal like a YUV image or a YIQ image, and dynamic range compression processing is performed only on the luminance signal. Then, if the color difference signal is not changed, the dynamic range can be compressed even for a color image. That is, the image processing apparatus is also effective for color images.

The image processing apparatus according to the thirty-second aspect is:
The image processing device according to claim 31, wherein the range compression device converts the RGB data of the color image into a luminance component and an RG.
It has a decoding device for decomposing the B data into corrected RGB data obtained by dividing the B data by the luminance component. Further, the encoding device includes a dynamic range compression for the luminance component, and obtains the dynamic range compressed RGB data by the product of the compressed luminance component and the corrected RGB data. The decoding device decomposes the RGB data into a luminance component and corrected RGB data obtained by dividing the RGB data by the luminance component. Then, the image processing apparatus according to any one of claims 10 to 29 compresses the dynamic range for the luminance component. Then, the encoding device uses the compressed luminance component and the corrected RG.
Multiply B data to create RGB data with a compressed dynamic range. Finally, it is sent to the next color display device, for example. By doing so, the dynamic range of the color image can be compressed without changing the color components. That is, the image processing device according to claim 31 can be realized.

[0068]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Specific embodiments of the present invention will be described below with reference to the drawings. The present invention is not limited to the examples below. (First Embodiment) FIG. 1 shows an image processing system for compressing a dynamic range using the image processing apparatus of the present invention.
The figure is a system configuration diagram. The image processing system of the present embodiment comprises a monochrome image pickup device 1, a digital processing device 2, a brightness level conversion device 3 which is a logarithmic conversion device, an image processing device 10, an analog processing device 5, and a monochrome image display device 6. .

In the above structure, the monochrome image pickup device 1 is, for example, a CCD camera. The digital processing device 2 is an analog output (luminance) of the monochrome image pickup device 1.
Is converted into a digital value (luminance Y) by an A / D converter and output as a grayscale image. The brightness level conversion device 3 is a device for logarithmically converting the input brightness Y.
The image processing device 10 is a device that compresses the dynamic range based on the variance σ ² of the brightness distribution described below. The analog processing device 5 is a device for converting the signal, for example, a D / A converter into an analog signal. The monochrome image display device 6 is a display device that displays an image in accordance with the analog signal. Here, the brightness level conversion device 3 is described separately from the image processing device 10, but it may be included in the image processing device 10.

The processing process of the image processing system is as follows. First, the monochrome image pickup device 1
The image is sent to the digital processing device 2 as an analog signal. The digital processing device 2 A / D converts the signal into digital data and outputs it to the brightness level converter 3 as the brightness Y of each pixel in each frame. Brightness level converter 3
Outputs the input brightness Y to the image processing apparatus 10 by logarithmically converting it to brightness Y ′. The image processing device 10 converts the luminance Y ′ into the luminance Y ″ according to the processing based on the variance σ ² described later, and outputs the luminance Y ″ to the analog processing device 5 in the subsequent stage.
The analog processing device 5 D / A converts the luminance Y ″ of each pixel for each l frame to form an analog signal, and outputs the analog signal to the monochrome image display device 6. The image captured by the monochrome image capturing device 1 is processed in this manner.

The configuration and operation of the image processing apparatus 10 will be described below. The configuration of the image processing apparatus 10 and the processing contents are the features of this embodiment. The image processing apparatus 10 includes a high-pass filter device 11 that is a low-frequency suppressing unit, a filter setting unit 14 that is a suppressing characteristic setting unit, a brightness dispersion calculating unit 17 that is a brightness dispersion calculating unit, and a compression coefficient calculating unit 18 that is a compression coefficient calculating unit. An additional value setting unit 15 which is an additional value setting means,
Then, it is composed of an adder 16. In the above configuration, all the constituent elements except the luminance dispersion calculating section 17 which is the luminance dispersion calculating section constitute range compression means for compressing the dynamic range, and the addition value setting section 15 and the adder 16
An average brightness correction means for correcting the average brightness is configured. or,
The high-pass filter device 11 includes a frame memory 12 which is an image memory means and a high-pass filter 13 which is a high-pass filter means.

The luminance Y ′ input to the image processing apparatus 10 is the luminance dispersion calculating unit 17 and the high-pass filter device 1.
1 to the frame memory 12. Luminance variance calculator 1
7 is the average brightness E of the input brightness Y'in one frame.
And its variance _{^{σ 2 = Σ N i = 1}} (E-Y i) 2 / N and sends (i:: integer, N 1 number of pixels in a frame) is calculated in compression coefficient calculating unit 18. The compression coefficient calculation unit 18 calculates the variance σ
^The compression coefficient A is calculated from ^2, and is output to the next stage together with the brightness average E. The compression coefficient A is a coefficient for adjusting the gradation width of a pixel, and when the value is 1 or less, the dynamic range is compressed, and the smaller the compression coefficient A is, the more the dynamic range is compressed. When the compression coefficient A is 1 or more, the dynamic range is expanded.

A method of calculating the compression coefficient A is shown in FIG. The figure shows the function (graph) that represents the relationship between the compression coefficient A and the variance σ ^2.
Is. The compression coefficient A is obtained by the function shown in FIGS. For example, as shown in FIG. 2A, when the variance σ ² is larger than σ _{1, the} compression coefficient A is set to a constant value smaller than 1, and when the variance σ ² is smaller than σ ₁ ,
The compression coefficient A should be proportional (with a negative coefficient) to the variance σ ² . That is, the function is set such that the compression coefficient A decreases in proportion to the increase of the variance σ ² .
When the variance σ ² is smaller than σ _{2, the} compression coefficient A becomes a value larger than 1, and the dynamic range is expanded at this time.

In addition, instead of the function shown in FIG. 2 (a), the function shown in FIG. 2 (b) may be applied. That is, it is a function that sets the compression coefficient A to 1 when the variance σ ² is smaller than σ ₂ . As a result, when the variance σ ² is smaller than σ ₂ , the dynamic range is not expanded. The compression coefficient A has a variance σ ² as shown in FIGS.
It is calculated by a monotonically decreasing function with respect to. This monotonically decreasing function may include a part that is invariant with respect to changes in the variance σ ² . In both functions, the variance σ ² is σ ₁
The compression coefficient A when it is larger is set to the predetermined value G smaller than 1 in order to limit the suppression amount of the low frequency component so that the edge portion is not overemphasized. If the compression coefficient A is less than the predetermined value G, the image is over-compressed and the edge portion is overemphasized, resulting in an image in which overshoot and undershoot are conspicuous.

The compression coefficient A obtained as described above is
To the filter setting unit 14. Also, the brightness is flat at the same time.
It is also sent to the added value setting unit 15 together with the average E (FIG. 1).
In the filter setting unit 14, the low frequency
Frequency characteristics of the high-pass filter 13 for suppressing the component
Sex. The frequency characteristic of the high pass filter 13 is
For example, the frequency characteristic is set as shown in FIG. Immediately
Then, when the compression coefficient A is 1 or less, as shown in FIG.
Set to a high-pass filter that suppresses such low frequency components
However, when the compression coefficient A is larger than 1, it is shown in FIG.
The frequency component does not change and the brightness is multiplied by A.
Set to the lupass filter. At this time, the horizontal axis is the space circumference.
It is the wave number f and the vertical axis is the component of the spatial frequency f transmitted.
It is a multiplication factor. And the frame memory for each frame
Y ′ stored in 12 is output by the high-pass filter 13.
Low-frequency components are suppressed by filtering,
Brightness Y ' _FIs recorded again in the frame memory 12 as
It That is, the filtering process of the high-pass filter 13
Thus, an image in which low frequency components are suppressed is formed. or,
Its brightness Y '_FAre sequentially output to the adder 16.

On the other hand, the addition value setting section 15 sets the addition value P as a correction addition value based on the input luminance average E and the compression coefficient A. Because the high pass filter 13
This is because suppressing the low frequency component reduces the average luminance. Therefore, the addition value setting unit 15 determines the addition coefficient P for setting the average value of the luminance Y ′ _F after the low frequency component is suppressed by the high pass filter 13 to a desired value. For example, when the filter having the characteristics shown in FIG. 3B is applied, the average brightness E of the brightness Y ′ _F of the image after the filtering process is multiplied by A to obtain EA. If you want to make the average values equal before and after filtering, add value P = E
Calculate (1-A). Also, if you want the center value C of the input range of the display device a mean value of the luminance Y _'F after filtering process for calculating the additional value P = C-E A. And
In the adder 16, the added value P calculated by the added value setting unit 15 is added to each luminance Y ′ _F , and converted into the luminance Y ″ having the same average value and output. Image processing device
10 is configured as described above and operates as described above.

As described above, in the image processing apparatus 10, the processing in which the low frequency component is largely suppressed is performed as the luminance variance σ ² of the input image is larger. Therefore, the compression of the dynamic range is performed at a higher compression rate for an image whose luminance is distributed in a wider brightness range on average. On the contrary, the image whose brightness is distributed in a narrower brightness range on average,
The dynamic range is compressed with a small compression rate.
The average spread of the luminance distribution is below a certain value (σ ² <σ
₂ ) In some cases, the dynamic range can be extended by multiplying the luminance by a constant without suppressing low frequency components. That is, if the external display device to be output has a margin of the dynamic range, it is possible to match it.

In the conventional method of determining the compression coefficient A from the maximum value and the minimum value of the luminance distribution (histogram), the influence of local light and dark of a small number of pixels cannot be avoided. However, in the present embodiment, the brightness variance calculator 17 calculates the variance σ ² which is the statistic as described above, and the compression coefficient calculator 18 determines the compression coefficient A based on the variance σ ^2. It is hardly affected by a small number of pixels such as a dark area and a dark area.
That is, there is no unnatural emphasis of edges due to excessive compression.
That is, the dynamic range can always be compressed with the appropriate compression coefficient A. Further, since the compression coefficient A does not change significantly due to the change in the brightness of a small number of pixels, the degree of edge enhancement does not change unnaturally in time even when processing a moving image.

(Modification of First Embodiment) In the first embodiment,
The low frequency component suppressing means is composed of the high-pass filter device 11, especially the high-pass filter 13. Then, the filter characteristic is changed according to the luminance dispersion to control the suppression amount of the low frequency component. The low frequency component suppressing means may be composed of a low pass filter device and a subtractor. That is, the image processing apparatus 10 of the first embodiment can be modified like the image processing apparatus 30 shown in FIG.
That is, the low frequency component suppressing means is used as the low pass filter device 3
It is also possible to realize by subtracting the low frequency component extracted by the low-pass filter device 31 from each luminance Y ′ of the input image.

In this case, the frequency characteristic of the low-pass filter device 31 is a fixed characteristic as shown in FIG. 5 and is not changed depending on the input image. Instead, a coefficient calculator 35, which is a multiplication value setting means for setting a correction multiplication value, is provided,
The coefficient calculator 35 determines the suppression coefficient B, which is the low frequency component suppression amount, and the expansion coefficient C, which is the correction multiplication value for expanding the dynamic range, based on the compression coefficient A obtained by the compression coefficient calculation unit 18. To do. This changes the filter characteristics. That is, the multiplier 32 multiplies the suppression coefficient B to adjust the suppression amount of the low frequency component, and the multiplier 34 multiplies the expansion coefficient C to expand the dynamic range. Suppression coefficient B
The expansion coefficient C and the expansion coefficient C are obtained from the compression coefficient A by the functions shown in FIGS. For example, when the compression coefficient A is 1 or less, the suppression coefficient B is calculated by B = 1-αA, and when the compression coefficient A is 1 or more, it is a constant value. Further, the expansion coefficient C is calculated by C = βA when the compression coefficient A is 1 or more, and is a constant value when it is 1 or less. Here, α and β are predetermined coefficients. In the present embodiment, by using the configuration using the low-pass filter, the frequency characteristic of the filter can be a fixed characteristic that does not change depending on the characteristic of the original image.
It is possible to simplify the configuration of the filter. The image processing device 30 may be configured in this way. Regarding the compression of the dynamic range, it is possible to realize the same function as that of the image processing apparatus 10. In this case, the frame memory 12 of the first embodiment is not used, and sequentially, in real time,
This is an example of converting the luminance Y ′ of each pixel.

A convolution filter is used as the low-pass filter device 31, but the invention is not limited to this, and an edge-preserving smoothing filter such as a median filter may be used. Further, these filters may be applied to an image obtained by thinning out pixels of the original image. Further, the low frequency component may be obtained by creating a blurred image by optical means. In short, any means can be used as long as it suppresses the high frequency component of the image.

Further, the image processing apparatus 30 of FIG. 4 can be modified like the image processing apparatus 40 shown in FIG.
That is, the function of the multiplier 32 may be non-linearly corrected by the brightness converter 41 based on the look-up table which is a non-linear correction means. The conversion characteristics of the lookup table are shown in FIG.
The characteristics are as shown in. FIG. 8A shows the characteristic of compressing only the high luminance portion, and FIG. 8B shows the example of the characteristic of compressing only the low luminance portion. As shown in the figure, by changing the conversion characteristic according to the compression coefficient A, the suppression amount of the low frequency component can be changed nonlinearly according to the luminance and the dispersion of the luminance distribution. In this way, the variance of the brightness distribution of the input image, that is, the low-frequency component may be non-linearly converted according to the compression coefficient A and the brightness, and the subtracter 33 may subtract the low-frequency component from the input image. By doing so, it becomes possible to compress only a desired specific range such as a low-luminance portion or a high-luminance portion at a desired compression rate. In this example, a coefficient calculator 42 is provided instead of the coefficient calculator 35 of FIG. 4, and only the expansion coefficient C is output.

(Second Embodiment) In the first embodiment, the suppression amount of the low frequency component is changed according to the variance σ ^{2 of the} input image 1
An example was given. In this embodiment, the suppression amount of the low frequency component is changed based on the amount representing the bimodality of the luminance distribution of the input image.
Here is an example. This method is effective for processing a scene in which a bright portion and a dark portion are clearly separated due to a difference in light and shade due to a shadow and the like, and the difference in lightness is large. An example of the image processing apparatus 20 of this embodiment is shown in FIG. The image processing apparatus according to the present embodiment includes the brightness variance calculation unit 17 and the compression coefficient calculation unit 18 of the image processing apparatus 10 according to the first embodiment, and includes the bimodal coefficient calculation unit and the compression coefficient calculation unit illustrated in FIG. It is configured by replacing the peak coefficient / compression coefficient calculation unit 50. The other components are the same as those of the first embodiment. The bimodal coefficient calculating means is equivalent to the brightness dispersion calculating means in terms of the function of outputting the dynamic range compression coefficient A and the brightness average as described later. Therefore, the brightness variance calculating unit 17 and the compression coefficient calculating unit 18 can be directly replaced by the bimodal coefficient / compression coefficient calculating unit 50.

The bimodal coefficient / compression coefficient calculation unit 50 of this embodiment.
Is a brightness level determiner 51, a mean square calculator 52, a class calculator 53, a pixel number adder 54, a brightness adder 55, a total variance calculator 56, an interclass variance calculator 57, a maximum value output device 58, It is composed of a discrimination reference value calculator 59. The brightness level determiner 51 is a device that determines the brightness level of each pixel in one input frame and outputs it to any one of (n + 1) output lines of X _{0 to} X _n . The mean square calculator 52 is a calculator that calculates the mean square of all pixels. Also, the class calculator 53 determines that the luminance Y ′
_This is an arithmetic unit for calculating the sum R _i of the number of pixels M _i input with _i and its luminance Y ′. The pixel number adder 54 is an adder that adds the sum of M _{0 to} M _{i−1 to} the output (number of pixels) M _i of the class calculator 53 and outputs the cumulative number of pixels N _i . The brightness adder 55 outputs R ₀ to R _i to the output R _i of the class calculator 53.
It is an adder that adds the sum of R _i−1 and outputs the cumulative luminance S _i .

The inter-class variance calculator 57 binarizes the accumulated pixel number inputs N _i , N _n , and the accumulated luminance S _i , S _n with the threshold value k _i using the equation shown in the equation (5) described later. This is a computing unit that obtains the _interclass variance σ _Bi in the case. The maximum value output device 58 receives the n _interclass variances σ _Bi (i =
It is an arithmetic unit that detects the maximum value among 1, 2, ..., n). The total variance calculator 56 calculates the total pixel variance σ from the output of the mean square calculator 52, the cumulative number of pixels N _n , and the cumulative luminance S _n.
It is an arithmetic unit that calculates _T and calculates the average luminance E of all pixels from the cumulative number of pixels N _n and the cumulative luminance S _n . The discriminant reference value calculator 59 calculates the discriminant reference value η _max from the ratio between the maximum value of the inter-class variance σ _Bi and the total pixel variance σ _T based on the above equations (1) and (2). Is. Here, the determination reference value eta _{ma x} is a value corresponding to a bimodal coefficient representing the degree of bimodality of the luminance distribution.

Below, the bimodal coefficient / compression coefficient calculation unit 50
Discriminant reference value η _max is obtained in
_The operation of calculating the compression coefficient A based on _max and further calculating the brightness average E of all images will be described. In the above equations (1) and (2), the occurrence probability ω in each class
_The following relationship holds between ₁ , ω ₂ and class mean value μ ₁ , μ ₂ and overall mean value μ _T.

## EQU00003 ## ω ₁ + ω ₂ = 1 ω ₁ μ ₁ + ω ₂ μ ₂ = μ _T (3) Then, if the above relational expression (3) is used, the expression (2) becomes It is expressed as follows.

[Equation 4]         σ_B ²= Ω₁(Μ₁-Μ₂)²/ (1-ω₁) ・ ・ ・ ・ ・ ・ (4) Therefore, the threshold value k that maximizes η in equation (1) is selected.
To solve this, σ in equation (4) _B ²Choose a threshold that maximizes
Good. This selection is made by the following operation.

It is input to the bimodal coefficient / compression coefficient calculation unit 50.
The luminance Y ′ is calculated by the luminance level determiner 51 and the root mean square calculation.
Sent to the container 52. The brightness level determiner 51 inputs
The brightness level of each pixel₀~ X_nOf (n
+1) Output to any one of the output lines. Concrete
Specifically, k₁<k₂<・ ・ <K_nN thresholds k _i
(I = 1, 2, ..., N) is provided to adjust the brightness of the input luminance.
Bell Y is Y <k₁When X₀To k_i<Y <k
_{i + 1}X when (i = 1, 2, ..., N-1)_iTo Y>
 k_nWhen X_nOutput to. And X_iIs input
The generated class calculator 53 is input during l frames.
Number of pixels M_iAnd the sum R of the input luminance Y ' _iOutput
It

In the pixel number adder 54, the class calculator 53
Output (number of pixels) M_iTo M₀~ M_{i- 1}Add the sum of
Number of product pixels N_iIs output. In the brightness adder 55, the class
Output R of calculator 53_iTo R₀~ R_i-1Add the sum of
Product brightness S_iIs output. That is, the cumulative number of pixels N_iIs the brightness
Y'is the threshold value k_iIs the number of pixels that is less than or equal to_i
The luminance Y'is the threshold value k_iIs the sum of the brightness of the pixels
It Also, the cumulative number of pixels N_n, And the cumulative brightness S_nIs all pixels
Is the sum of the number of pixels and the luminance. Inter-class distributed computing unit 57
Then, the cumulative number of pixels N_i, Cumulative brightness S_i, Cumulative pixel count
N _n, Cumulative brightness S_nFrom the calculation formula shown in Formula (5)
And the threshold k_iInter-class variance σ when binarized by_BiSeeking
And output.

Σ _Bi ² = N _i · (S _i / N _i −S _n / N _n ) ² / (N _n −N _i ) ... (5)

In the maximum value output device 58, the n input
Interclass variance σ_BiMaximum value in (i = 1, 2, 2, ..., N)
Is detected and output. On the other hand, from the mean square calculator 52
The root mean square of all pixels is calculated and output. Total variance
In the computing unit 56, the root mean square and the cumulative pixel number N_n,Accumulation
Brightness S_nFrom the total pixel variance σ_TTo calculate the discrimination standard value
Output to the calculator 59. In addition, the total variance calculator 56
Prime number N_n, Cumulative brightness S _nTo the average luminance E of all pixels (= cumulative
Product brightness S_n/ Cumulative number of pixels N_n) Is calculated and output. Size
The different reference value calculator 59 is input at the maximum value output device 58.
Between classes σ_BiMaximum value and total variance calculator 56
Total pixel variance σ calculated in_TFrom the ratio of
Discriminant reference value η which is a bimodal coefficient according to_maxAsk for. So
Then, the determined reference value η_maxFig. 11 from (bimodal coefficient)
The compression coefficient A is calculated and output according to the characteristics shown in
It After that, the bimodal coefficient and the discrimination reference value η_maxAre used interchangeably
To use.

Discrimination reference value η_maxCalculate compression factor A from
The method is shown in FIG.²Calculate the compression factor A from
It is similar to the method described above. That is, the discrimination reference value η_maxIs η₁Yo
If it is larger than 0, set the compression coefficient A to a constant value J smaller than 1.
The judgment reference value η_maxIs η ₁Compressor when smaller
The number A is the discrimination reference value η_maxProportional to (with a negative coefficient)
To do so. That is, the discrimination reference value η_maxBecomes larger
The function is set so that the compression coefficient A decreases in proportion to it.
Set. That is, the discrimination reference value η_maxIs large (low frequency
Min., For example, the brightness distribution of the illumination intensity is large)
For the image, the compression coefficient A is small and the low frequency component is large.
Suppress. On the contrary, the discrimination reference value η_maxIs small (low frequency
Min., For example, the difference in the brightness distribution of the illumination intensity is small)
The image is processed with a reduced suppression rate. By this
Therefore, the edges are not overemphasized.
Yes. In addition, the input image with a large difference
Black in dark areas when displayed on a narrow display device
Crushing and whitening in bright areas are also alleviated. The above
The processing shows two bimodal peaks on the histogram.
B₁, B₂Are processed to approach one peak A
(FIG. 20). That is, the background darkness, which is a low-frequency component,
For example, it is processed so that the light and darkness due to lighting is reduced.

As described above, the bimodal coefficient of the luminance distribution is
It is clearly divided into two parts, a light part and a dark part, like a scene including shadows and sun rays, and it is possible to more faithfully show the difference in an image with a large difference in brightness. An image that is clearly divided and has a large difference in lightness can be compressed more accurately and largely. In addition, since the bimodal coefficient representing the bimodal nature of the luminance distribution is also a statistical quantity, it is hardly affected by a small number of pixels such as a local bright part or dark part of the input image. Therefore, the dynamic range can always be stably and appropriately compressed. Also, since the compression coefficient does not change significantly due to the change in the brightness of a few pixels,
In processing a moving image, the degree of edge enhancement does not change unnaturally with time.

(Third Embodiment) The first and second embodiments are examples in which the compression coefficient A is calculated and processed from the input image after the luminance conversion (without filter processing). The present embodiment is an example in which the compression coefficient A is calculated from the low-frequency image after the filter processing and is processed. The image processing device 60 of this embodiment is shown in FIG.
Shown in. The figure is a configuration block diagram. The image processing device 60 of the present embodiment is provided with a low-pass filter 61, which is a low-frequency image extracting means, in the preceding stage of the brightness dispersion calculating unit 17 in the image processing device 10, and the brightness dispersion calculating unit 17 calculates the characteristics of the low-frequency image. The characteristic feature is that it is a characteristic amount calculating means.
At this time, the characteristic amount is, of course, dispersion. This focuses on the fact that low-frequency components such as uneven illumination are low-frequency images. That is, it is rational to obtain the compression coefficient and the like from the low-frequency image because the high-frequency component is naturally relaxed. The other configurations are the same as those of the image processing apparatus 10, and the same parts are denoted by the same reference numerals.

In the above configuration, since the low-pass filter 61 is provided in the preceding stage of the brightness dispersion calculating unit 17, the brightness dispersion calculating unit 17 removes the local bright and dark parts, which are rapid brightness changes of the image, as high frequency components. Entered.
Alternatively, the changed amount is relaxed and input. Therefore, the compression coefficient calculation unit 18 calculates the compression coefficient A in which the influence of the local bright portion or dark portion is reduced. Also, the compression coefficient A
The influence of the local bright portion or dark portion in the calculation of is controlled by adjusting the cutoff frequency and / or the transmission coefficient of the low pass filter 61. For example, if the cutoff frequency of the low-pass filter 61 is reduced, it is possible to further reduce the local influence of the bright portion and the dark portion in the calculation of the compression coefficient A. Even with this configuration, the same effect as that of the image processing apparatus of the first embodiment can be obtained.

In addition, the luminance dispersion calculation unit 17 shown in this embodiment.
The compression coefficient calculation unit 18 can be replaced with the bimodal coefficient / compression coefficient calculation unit 50 shown in the second embodiment. In that case, the characteristic amount becomes a bimodal coefficient, and the same effect as that of the image processing apparatus of the second embodiment can be obtained. Furthermore, the brightness variance calculation unit 17
Can be replaced with a maximum / minimum calculation device (not shown) that is a characteristic amount calculation means. The maximum / minimum calculation device obtains the difference or ratio between the maximum value and the minimum value of the input image (low frequency image),
The compression coefficient calculation unit 18 may calculate the compression coefficient A according to the obtained difference or ratio. The maximum / minimum calculation device is, for example, a histogram search device (not shown), and if the histogram of the image is searched, the difference between the maximum value and the minimum value,
Alternatively, the ratio can be easily obtained. The difference or ratio between the maximum value and the minimum value is approximately proportional to the difference between the brightness and darkness of the low-frequency image, and if the characteristic amount (difference or ratio) is large, the low-frequency component may be greatly suppressed. Similar effects are obtained. Obtaining the difference or ratio between the maximum value and the minimum value from the histogram is easy with a small amount of calculation as compared with the variance and the bimodal coefficient.
Since the calculation takes a short time, the image processing apparatus can perform high-speed compression processing. Such a configuration may be adopted.

(Modification of Third Embodiment) The third embodiment is as follows.
This is an example in which a low frequency component is created by a low pass filter and the compression coefficient A is calculated based on the low frequency component. That is, the low-pass filter 61 is provided in the preceding stage of the luminance dispersion calculation unit 17 of the image processing apparatus 10 (FIG. 1) of the first embodiment. This configuration can be further simplified. That is, the image processing apparatus 70 shown in FIG. 13 may be configured by providing the low-pass filter device 31 in the preceding stage of the luminance dispersion calculation unit 17 of the image processing apparatus 30 of FIG. The frame memory 12 and the high pass filter 13 can be omitted. Further, as shown in FIG. 14, a second low-pass filter device 71 having a characteristic different from that of the low-pass filter device 31 in the image processing device 70 of FIG. 13 may be provided in the preceding stage of the luminance dispersion calculator 17. The compression coefficient A can be changed by changing the filter characteristic. That is, the suppression amount of the low frequency component may be controlled from outside the compression coefficient calculation unit 18.

(Fourth Embodiment) The first to third embodiments are examples of the image processing apparatus for compressing the dynamic range of a monochrome grayscale image. The present embodiment is an example of an image processing apparatus capable of compressing a dynamic range for a color image (grayscale image). FIG. 15 shows an image processing system of the present invention. The figure shows a system using a color image processing apparatus 80 capable of compressing a color image. The image processing system according to the present embodiment mainly includes a color image pickup device 1c, a color image processing device 80, and a color display device 6c. The color image processing device 80 includes a video decoder 81 which is a decoding device, an rgb memory 82 which is a color signal storage device, multipliers 83r, 83g and 83b which are encoding devices, a video encoder 84, a brightness level converter 3, and an image. It is composed of the processing device 10. The brightness level converter 3 and the image processing device 10
Is the same as that of the first embodiment.

The processing process of this color image processing system is as follows. First, the color image pickup device 1
The analog image signal obtained in c is sent to the video decoder 81. The video decoder 81 uses RGB color images.
The data is decomposed into a luminance Y which is a luminance component and r, g and b signals which are corrected RGB data obtained by dividing an RGB signal by the luminance Y, and output as digital values to the next stage. That is, the brightness (signal) Y is sent to the brightness level converter 3 and the image processing device 10 and is compressed by the same process as in the first embodiment.
The r, g and b signals are input to and stored in the rgb memory 82.

The image processing apparatus 10 outputs a luminance signal Y ″ which has been appropriately compressed according to the luminance distribution of the image. From the rgb memory, r, g corresponding to this luminance Y ″
, B signals are read out, and R ′ = RY ″ / in each of the multipliers 83r, 83g, 83b which are encoding devices.
Y, G ′ = GY ″ / Y, B ′ = BY ″ / Y are calculated. The R ′, G ′, and B ′ signals obtained in this way become new RGB signals in which the difference in brightness is appropriately suppressed according to the luminance distribution of the input image. Then, the video encoder 84 generates an analog color signal based on this RGB signal and outputs it to the color image display device 6c. In this way, the color image processing device 80 can compress the difference in brightness without affecting the color information.

(Modification) The present invention can be implemented in various other forms. For example, in the first embodiment (FIG. 1), the high-pass filter device 11 is composed of the frame memory 12 and the high-pass filter 13 and processed by software. However, as shown in FIG.
The high-pass filter device 11a composed of sor) or the like may be used. At this time, a coefficient calculator 35 is provided in place of the filter setting unit 14 to control the high-pass filter device 11a. That is, the suppression coefficient B controls the characteristics of the high-pass filter device 11a. You may comprise in this way. The same effect can be obtained.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an image processing system using an image processing apparatus according to a first embodiment of the present invention.

FIG. 2 is an explanatory diagram of a compression coefficient calculation method using dispersion according to the first embodiment of the present invention.

FIG. 3 is a characteristic diagram of a high-pass filter according to the first embodiment of the present invention.

FIG. 4 is a configuration diagram of an image processing apparatus according to a modification of the first embodiment of the present invention.

FIG. 5 is a characteristic diagram of a low-pass filter used in the image processing apparatus according to the modified example of the first embodiment of the present invention.

FIG. 6 is an explanatory view (a) of a method of calculating a suppression coefficient B and an explanatory view (b) of a method of calculating an expansion coefficient C in a modified example of the image processing apparatus according to the first embodiment of the present invention.

FIG. 7 is a configuration diagram of an image processing apparatus capable of nonlinear suppression according to a modification of the first embodiment of the present invention.

FIG. 8 is an explanatory diagram illustrating a non-linear suppression function according to a modification of the first embodiment of the present invention.

FIG. 9 is a configuration diagram of an image processing apparatus according to a second embodiment of the present invention.

FIG. 10 is a configuration diagram of a bimodal coefficient / compression coefficient calculation unit according to the second embodiment of the present invention.

FIG. 11 is an explanatory diagram of a compression coefficient calculation method using a bimodal coefficient according to the second embodiment of the present invention.

FIG. 12 is a configuration diagram of an image processing apparatus according to a third embodiment of the present invention.

FIG. 13 is a configuration diagram showing a modified example of the image processing apparatus according to the third embodiment of the present invention.

FIG. 14 is a configuration diagram showing a modification of the image processing apparatus according to the third embodiment of the present invention.

FIG. 15 is a configuration diagram of a color image processing device according to a fourth embodiment of the invention.

FIG. 16 is a configuration diagram of an image processing system using an image processing apparatus of a modified example according to the first embodiment of the present invention.

FIG. 17 is a one-dimensional luminance distribution diagram of an input image including low frequency components and high frequency components in the conventional example.

FIG. 18 is a histogram of an input image for explaining a change in dynamic range in the conventional example.

FIG. 19 is a histogram showing the difference in dispersion according to the illumination difference of the present invention.

FIG. 20 is a histogram showing peak shift due to bimodality and suppression of the present invention.

[Explanation of symbols]

3 ... Luminance level converter 10, 20, 30 ... Image processing device 12 ... Frame memory 13 ... High-pass filter 14 ... Filter setting unit 15 ... Additional value setting section 16 ... Adder 17 ... Luminance variance calculator 18 ... Compression coefficient calculation unit 31 ... Low-pass filter 32, 34 ... Multiplier 33 ... Subtractor 35, 42 ... Coefficient calculator 40, 60, 70 ... Image processing device 41 ... Luminance converter 50 ... Bimodal coefficient / compression coefficient calculation unit 81 ... Video decoder 82 ... rgb memory 83r, 83g, 83b ... Multiplier 84 ... Video encoder

Continued front page    F-term (reference) 5B057 BA02 CA01 CA02 CA08 CA12                       CA16 CB01 CB02 CB08 CB12                       CB16 CC01 CE11 CH08                 5C021 PA02 PA33 PA34 PA77 PA78                       PA85 PA86 RB03 XA31                 5C066 AA01 CA07 EA03 EA11 GA01                       GA05 JA01 KC02 KC03 KE03                       KE07                 5C077 LL04 MP01 MP08 PP01 PP15                       PP16 PP32 PQ12 PQ19 SS06                       TT09

Claims (32)

[Claims] 1. An image processing method for compressing a dynamic range by suppressing low-frequency components of an input image, wherein a variance-related value related to a variance of a luminance distribution of an input image is obtained, and based on the obtained variance-related value. And compressing the dynamic range by suppressing low frequency components of the input image. 2. The image processing method according to claim 1, wherein the suppression amount of the low frequency component is increased as the dispersion related value is increased. 3. An image processing method for compressing a dynamic range by suppressing a low frequency component of an input image, wherein the luminance distribution of the input image is based on a bimodal coefficient representing a degree of bimodality, An image processing method characterized by compressing a dynamic range by suppressing low-frequency components. 4. The image processing method according to claim 3, wherein the suppression amount of the low frequency component is increased as the bimodal coefficient is increased. 5. An image processing method for compressing a dynamic range by suppressing low-frequency components of an input image, wherein a low-frequency image composed of low-frequency components of the input image is obtained, and the spread of luminance distribution of the low-frequency image is expanded. An image processing method characterized by compressing a dynamic range by suppressing a low-frequency component of an input image based on a characteristic amount shown. 6. The image processing method according to claim 5, wherein the characteristic amount is a dispersion related value related to the dispersion of the luminance distribution. 7. The image processing method according to claim 5, wherein the characteristic amount is a bimodal coefficient that represents a degree of bimodality of the luminance distribution. 8. The image processing according to claim 5, wherein the characteristic amount is a difference between a maximum value and a minimum value of brightness in the brightness distribution, or a ratio between the maximum value and the minimum value. Method. 9. The image processing method according to claim 6, wherein the amount of suppression of the low frequency component is increased as the characteristic amount is increased. 10. An image processing apparatus for compressing a dynamic range by suppressing a low frequency component of an input image, a brightness variance calculating means for calculating a variance related value related to a variance of a brightness distribution of an input image, and the brightness. Image processing, characterized in that, in the input image, based on the dispersion-related value obtained by the dispersion calculation means, range compression means for compressing a dynamic range by suppressing low-frequency components of the input image is provided. apparatus. 11. The image processing apparatus according to claim 10, wherein the range compression unit is a device that increases the suppression amount of the low frequency component as the dispersion-related value increases. 12. An image processing apparatus for compressing a dynamic range by suppressing a low frequency component of an input image, wherein a bimodal coefficient calculation for calculating a bimodal coefficient representing a degree of bimodality in a luminance distribution of the input image. An image comprising: means for compressing a dynamic range by suppressing a low frequency component of an input image based on the bimodal coefficient calculated by the bimodal coefficient calculating means. Processing equipment. 13. The image processing apparatus according to claim 12, wherein the range compression unit is a device that increases the suppression amount of the low frequency component as the bimodal coefficient increases. 14. An image processing apparatus for compressing a dynamic range by suppressing low frequency components of an input image, comprising: A characteristic amount calculation unit that calculates a characteristic amount indicating the spread of the luminance distribution of the low frequency image extracted by the low frequency image extraction unit; and the input based on the characteristic amount obtained by the characteristic amount calculation unit An image processing apparatus, comprising: a range compression unit that compresses a dynamic range by suppressing a low frequency component of the input image. 15. The characteristic amount is a dispersion-related value related to the dispersion of the luminance distribution.
The image processing device according to item 1. 16. The bimodal coefficient representing the bimodal degree of the luminance distribution, wherein the characteristic amount is a bimodal coefficient.
4. The image processing device according to item 4. 17. The image processing according to claim 14, wherein the characteristic amount is a difference between a maximum value and a minimum value of brightness in the brightness distribution, or a ratio between the maximum value and the minimum value. apparatus. 18. The image processing apparatus according to claim 15, wherein the amount of suppression of the low frequency component is increased as the characteristic amount is increased. 19. The range compression means comprises an average brightness correction means for correcting the average brightness of an image after suppressing low frequency components.
The image processing apparatus according to any one of 1. 20. The range compressing means determines a compression coefficient of at least a dynamic range based on the dispersion-related value calculated by the luminance dispersion calculating means, and suppresses low frequency components of an input image. The image processing according to claim 10 or 11, further comprising: low-frequency component suppressing means, and suppression characteristic setting means for setting characteristics of the low-frequency component suppressing means based on the compression coefficient. apparatus. 21. The range compression means determines a compression coefficient of at least a dynamic range based on the bimodal coefficients calculated by the bimodal coefficient calculation means, and suppresses low frequency components of an input image. 14. The image according to claim 12 or 13, further comprising: a low-frequency component suppressing unit configured to perform: a suppression characteristic setting unit configured to set a characteristic of the low-frequency component suppressing unit based on the compression coefficient. Processing equipment. 22. The range compression means includes a compression coefficient calculation means for determining at least a compression coefficient of a dynamic range based on the characteristic amount calculated by the characteristic amount calculation means, and a low-frequency component for suppressing a low frequency component of an input image. 19. A frequency component suppressing means, and a suppression characteristic setting means for setting a characteristic of the low frequency component suppressing means on the basis of the compression coefficient, according to any one of claims 14 to 18. The image processing device described. 23. The range compression means, an addition value setting means for setting a correction addition value based on the compression coefficient and the luminance average obtained by the compression coefficient calculation means, and suppression by the low frequency component suppression means. For all pixels after processing,
23. The image processing apparatus according to claim 20, further comprising an average brightness correction unit including an adder that adds the correction addition value. 24. The range compression means multiplies the correction multiplication value by a multiplication value setting means for setting a correction multiplication value based on the compression coefficient, and all the pixels after the suppression processing by the low frequency component suppression means. 24. The image processing device according to claim 20, further comprising a multiplier. 25. The range compression means is provided with a logarithmic conversion device at an input, and the output of the logarithmic conversion device is used as the input image. Image processing device. 26. The image processing apparatus according to claim 20, wherein the low-frequency component suppressing unit is a high-pass filter device. 27. The image processing apparatus according to claim 26, wherein the high-pass filter device includes high-pass filter means and image memory means. 28. The low frequency component suppressing means is a low-pass filter device, a multiplier for multiplying each luminance after the low-pass filter processing by a suppression coefficient set by the suppressing characteristic setting means, and the multiplication from an input image. 21. A subtractor for subtracting the multiplication result of the multiplier.
The image processing apparatus according to claim 25. 29. The low-frequency component suppressing means, a low-pass filter device, a non-linear correcting means for non-linearly correcting each luminance after low-pass filtering based on the compression coefficient, and the non-linearly corrected image from an input image. 26. The image processing apparatus according to claim 20, further comprising a subtractor that subtracts 30. The image processing apparatus according to claim 10, wherein the input image is a grayscale image. 31. The input image is a color image, and compression of the dynamic range is performed with respect to a luminance component obtained from the color image.
The image processing device according to any one of claims 0 to 29. 32. The range compressing device decomposes RGB data of the color image into a luminance component and corrected RGB data obtained by dividing the RGB data by a luminance component, and a decoding device, and a dynamic range for the luminance component. 32. The image processing device according to claim 31, further comprising: an encoding device that obtains RGB data having a compressed dynamic range by compressing the compressed luminance component and the corrected RGB data.