CN100388757C - Method for enhancing image contrast - Google Patents

Abstract

An image processing method is to count the number of original grayscale values according to the original grayscale value difference between each pixel and its adjacent pixels in a picture, and determine the contrast enhancement degree of the picture according to the counted number of grayscale values and the contrast change range of the previous and next pictures.

Description

Methods of Enhancing Image Contrast

technical field

The invention relates to an image processing method, and in particular to a method for enhancing image contrast.

Background technique

In the past, the method of enhancing image contrast is only to determine the degree of enhancing the image contrast by counting the number of gray values of each pixel in the image. However, this method can only improve the image contrast in the frame under certain specific circumstances. However, in some situations where contrast enhancement is most needed, such as when watching DVD movies, this method of enhancing image contrast can produce extremely poor images. As shown in FIG. 1 , it is a schematic diagram of a screen when playing a DVD movie. When DVD movies are played, the display screen 100 is mostly displayed at a ratio of 16:9, so the upper and lower parts UP and DP of the display screen 100 (such as an LCD screen) will display black screens. In this way, the previous method of image contrast enhancement would count the gray values of the black images of the UP and DP parts when performing contrast enhancement on the image 100, resulting in an extremely unnatural appearance of the image 100 after contrast enhancement. . In addition, when the difference between the statistical curves of the gray values of the front and rear images is too large, the conventional method of enhancing image contrast will cause the phenomenon of image flickering.

Therefore, how to enhance the contrast while maintaining the natural characteristics of the image and avoiding the flickering phenomenon of the image screen is an issue that needs to be solved at present.

Contents of the invention

In view of this, the object of the present invention is to provide an image processing method, which enhances the contrast of an image and maintains the stability of the image frame under the condition of maintaining the natural characteristics of the image.

According to the object of the present invention, an image processing method is proposed. The image is presented by at least a first frame f(N). The first frame f(N) is composed of a plurality of pixels. Each pixel corresponds to an original gray value, and the image processing method of the present invention is described as follows. According to the difference of the original gray value between two adjacent pixels, the number of the original gray value appears is counted, and expressed by a first transfer function F(X). A gamma curve is calculated according to the first transfer function F(X), and a plurality of adjusted gray values are selectively adjusted according to the gamma curve. The above-mentioned step of counting the number of occurrences of these original gray values also includes: when the difference between a first original gray value of a first pixel in these pixels and a second original gray value of an adjacent second pixel is When it is greater than n, n is 0 or a positive integer, and the statistical number of all original grayscale values or the statistical number of partial original grayscale values between the first original grayscale value and the second original grayscale value are increased by 1.

In order to make the above-mentioned purpose, features, and advantages of the present invention more comprehensible, a preferred embodiment is specifically cited below, and is described in detail with reference to the accompanying drawings.

Description of drawings

Figure 1 is a schematic diagram of the screen when playing a DVD movie.

FIG. 2 is a flowchart of an image processing method in a preferred embodiment of the present invention.

FIG. 3A is a schematic diagram of an unspecified current frame.

FIG. 3B is a schematic diagram of statistics of the number of original gray values.

FIG. 3C is a schematic diagram of the statistics of the number of original gray values.

FIG. 4A is a flowchart of calculating a gamma curve.

FIG. 4B is a schematic diagram of calculating a gamma curve.

FIG. 5 is a schematic diagram of adjusting the upper and lower limits of the third transfer function.

Detailed ways

The invention proposes an image processing method, which enhances the contrast of an image and maintains the stability of the image frame under the condition of maintaining the natural characteristics of the image. The image processing method of the present invention counts the number of occurrences of these original gray values according to the original gray value difference between each pixel and its adjacent pixels, and then based on the counted number of gray values and The contrast change range is used to determine the degree of contrast enhancement of the current picture.

Please refer to FIG. 2 , which is a flowchart of an image processing method according to a preferred embodiment of the present invention. An image is presented by multiple frames f(N), where N is a positive integer. Each frame f(N) is composed of multiple pixels, for example, 1028*768 pixels. Each pixel is respectively corresponding to an original gray level (graylevel) GL. The image processing method of the present invention includes the following steps. In step 200, according to the original gray value difference between each pixel and its adjacent pixels, the number of occurrences of these original gray values is counted and represented by the first transfer function F _i (X), where i is a value between 1 and N positive integer. In step 202, a gamma curve (Gamma Curve) is calculated according to the first transfer function F _i (X). Then in step 204, the original grayscale values are selectively adjusted according to the gamma curve to a plurality of adjusted grayscale values.

The statistical method of step 200 will be further described. Please refer to FIG. 3A, which is a schematic diagram of an unspecified current frame. A first frame f(1) is a certain frame in the video. Each space in the first frame f(1) represents a pixel P, and the number marked in the space represents the original gray value GL corresponding to each pixel P. Taking the 3*3 pixels P(1)-P(9) in the upper left corner of the first frame f(1) to illustrate the principle of counting the number of occurrences of the original gray value of the first frame f(1). This principle is: "When the original gray value difference between each pixel and adjacent pixels is greater than n, and n is a predetermined value of 0 or a positive integer, the original gray value corresponding to each pixel and the corresponding Add 1 to the statistical number of all original grayscale values or the statistical number of partial original grayscale values between the original grayscale values corresponding to adjacent pixels.” Take n as 0 and take the ninth pixel P9 as an example for illustration , the ninth pixel P9 is adjacent to the sixth pixel P6 and the eighth pixel P8. The original grayscale value difference between the ninth pixel P9 and the sixth pixel P6 is 21 grayscales (95−74=21>0), so all original grayscale values between 74 and 95 original grayscale values (74, 75, 76, . FIG. 3B is a schematic diagram of statistics of the number of original gray values. The original grayscale value difference between the ninth pixel P9 and the eighth pixel P8 is 19 grayscales (93-74=19>0), so all the original grayscale values between the original grayscale values 74 to 93 (74 , 75, ... 92, 93) plus 1, as indicated by the dotted arrow in Figure 3B. Using this principle to make statistics on each original gray value GL of the pixels P1-P9 will obtain a statistical map of the number of partial gray values of the first frame f(1), as shown in Figure 3C, which is the original gray value Schematic diagram of degree value statistics.

Compared with the previous method (that is, only counting the number of occurrences of the original gray value of each pixel in the frame), the statistical principle of the present invention takes into account the difference in original gray value between adjacent pixels. In this way, the statistically calculated first transfer function F _i (X) takes into account the difference in gray value at the edge of the image in the frame, so that the gamma curve obtained based on the first transfer function F _i (X) is corrected. Images are more natural in terms of contrast. In addition, the above n can also be 1 or other positive integers, and the statistical method can also be the partial original gray value between the original gray value corresponding to each pixel and the original gray value corresponding to the adjacent pixel The statistical number of values is increased by 1. Taking the ninth pixel P9 and the eighth pixel P8 as an example, it is possible to add 1 to the statistical number of the partial original gray value whose difference between the original gray value 74 and 95 is 2, such as 74, 76, 78.. .92, 94, 95, according to this rule, can choose the original gray value difference of 3, 4 or other partial original gray value as statistical object, the present invention does not limit the value of difference.

Next, the method of calculating the gamma curve will be described. After performing statistics on the original gray values of all pixels in the first frame f(1) in the above statistical manner to obtain the first transfer function F ₁ (X), a gamma curve G is obtained through appropriate calculation in step 202 (X). Please refer to FIG. 4A and FIG. 4B at the same time. FIG. 4A is a flowchart of calculating a gamma curve. FIG. 4B is a schematic diagram of calculating a gamma curve. In step 400, the first transfer function F _i (X) is linearly transformed. In order to limit the range of the output value of the first transfer function F _i (X), the first transfer function F _i (X) is linearly transformed, for example, the square root of the first transfer function F ₁ (X) is taken. Next, in step 402, the linearly converted first transfer function F _i (X) is accumulated (accumulate) to obtain a second transfer function F' _i (X). In step 404, normalize (nomalize) the second transfer function F' _i (X) to obtain a third transfer function F" _i (X). For example, the second transfer function F' ₁ of the first frame f(1) The statistical number of (X) is normalized to the maximum grayscale value (for example, 255 grayscales), so as to obtain the third transfer function F″ ₁ (X). Then in step 406, the third transfer function F" _i (X) is multiplied by K and a parameter P (X) is added to obtain a gamma curve G (X). Among them, as shown in Figure 4B, the parameter P (X ) is, for example, a curve with a slope of 1-K, and K is any value between 0 and 1, which can be pre-selected. So far, the gamma curve G ₁ (X) of the first frame f(1) can be calculated After the gamma curve G ₁ (X) is calculated, the original gray value GL of the first frame f(1) can be adjusted according to the gamma curve G ₁ (X).

Since the first transfer function F ₁ (X) takes into account the difference in the gray value of the image edge in the frame, when the first frame f(1) is a 16:9 DVD movie, according to the gamma curve G ₁ (X) The adjusted first frame f(1) will present a better contrast effect.

In addition, in order to maintain the stability of the picture, that is to avoid the phenomenon of unnatural image or image flicker caused by the sharp change of the picture brightness, the change of the gamma curve can be made by limiting the output value of the above-mentioned various transfer functions F(X). rate is relatively stable. For example, a part of the upper and lower limits of the third transfer function F" _i (X) is adjusted to obtain a fourth transfer function F"' _i (X). As shown in FIG. 5 , it is a schematic diagram of adjusting the upper and lower limits of the third transfer function. For example, the third transfer function F"(X) is multiplied by the two transfer functions J1(X) and J2(X) to obtain a fourth transfer function F"' _i (X). When the third transfer function F”(X) is greater than the transfer function J1(X), the output transfer function J1(X), and when the transfer function F”(X) is smaller than the transfer function J2(X), the output transfer function J2(X ). Afterwards, another gamma curve G'(X) is generated according to the fourth transfer function F"' _i (X). For example, another gamma curve G'(X) is obtained after being processed in step 406 above.

Or judge whether to adjust the original gray value GL of the current picture according to the gamma curve G(X) according to the second transfer function F' _i (X). That is, when the accumulated maximum value of the second transfer function F' ₁ (X) is greater than a first default value W1, the original grayscale value of the current frame is adjusted according to the gamma curve G(X). For example, when the second transfer function F' ₁ (X) of the first picture f(1) is smaller than the first default value W1, that is, the accumulated number is not greater than the default value W1, then the first picture f(1) is defined is a smooth scene, otherwise it is a general scene. In a smooth scene, the original gray value GL of the first frame f(1) is adjusted without using the gamma curve G ₁ (X) to keep the image frame stable. In general scenarios, the original gray value GL of the first frame f(1) is adjusted using the gamma curve G ₁ (X). In addition, the original gray value GL of the first frame f(1) can also be adjusted through the gamma curve G ₁ ′(X) to keep the image frame stable.

Alternatively, since the image is presented by a plurality of frames f(N), each frame f(N) is respectively corresponding to a statistics map of the number of original gray values, that is, the first transfer function F _i (X). By comparing the two first transfer functions F _i (X) of the two frames before and after, it is judged whether to adjust the gamma curve G(X) of the current frame or the gamma curve G(X) of the previous frame to adjust the gamma curve of the current frame Raw gray value GL. The method of the comparison may be to compare whether the gray level difference of the same pixel or the same pixel area of the two frames before and after is outside or within a predetermined tolerance value. Take the first frame f(1) and the frame before the first frame f(1), a second frame f(0) as an example for illustration. The first frame f(1) corresponds to a first transfer function F ₁ (X), and the second frame f(0) corresponds to another first transfer function F ₀ (X). First, integrate the first transfer function F ₁ (X) of the first frame f(1) to obtain a first integral value E1. Next, after subtracting the first transfer function F ₁ (X) of the first frame f(1) from the first transfer function F ₀ (X) of the second frame f(0), the absolute value is taken and integrated to obtain a The second integral value E2. When the ratio of the second integral value E2 to the first integral value E1 is greater than a second default value W2, it is considered that there is a significant difference between the current first frame f(1) and the second frame f(0). When there is a significant difference, the original gray value GL of the first frame f(1) is adjusted according to the gamma curve G ₁ (X) generated by the first transfer function F ₁ (X), otherwise the first frame f(1 ) is adjusted according to another gamma curve G ₀ (X) generated by the first transfer function F ₀ (X) of the previous frame.

To sum up, the above methods are all based on various forms of transfer functions F _i (X) to adjust the rate of change of the gamma curve or to select the gamma curve of the previous frame, so as to avoid sharp changes in the brightness of the frame and cause image distortion. Unnatural or flickering images.

In addition, the above-mentioned third method is to compare the first transfer function F _i (X) corresponding to the two frames before and after, so as to judge whether to use the gamma curve G(X) of the current frame or the gamma curve G(X) of the previous frame ) to adjust the original gray value GL of the current picture. It is also possible to judge whether to adjust according to the gamma curve G(X) of the current picture or the gamma curve G(X) of the previous picture by comparing the two first pixel transfer functions H _i (X) of the two pictures before and after The original gray value GL of the current picture. The first pixel transfer function H _i (X) is used to represent the counted number of original grayscale values of a frame, that is, the number of occurrences of original grayscale values of each pixel in the previous counted frame. Further, the first frame f(1) and the second frame f(0) are also taken as examples for illustration. First count the number of occurrences of each original gray value in the first frame f(1), and express it with a first pixel transfer function H ₁ (X); and count the number of each original gray value in the second frame f(0) The number of occurrences of the value is also represented by another second pixel transfer function H _N-1 (X). Next, after integrating the first pixel transfer function H _N (X) to obtain another first integral value E1', after subtracting the first pixel transfer function H ₁ (X) from the second pixel transfer function H ₀ (X) The absolute value is taken and integrated to obtain a second integral value E2'. When the ratio of the second integral value E2' to the first integral value E1' is greater than a third default value W2, it is considered that there is a significant difference between the current first frame f(1) and the previous frame f(0). At this time, the original gray value GL of the first frame f(1) is adjusted to the adjusted gray value GL' according to the gamma curve G(X) generated by the first transfer function F _N (X), otherwise the first frame f The original gray value GL of (1) is adjusted to the adjusted gray value GL′ according to another gamma curve G(X) generated by the first transfer function F ₀ (X) of the previous frame. Therefore, the present invention can also use the conventional statistical method, that is, the first pixel transfer function H _i (X), to determine whether to use the gamma curve of the current frame or the previous frame to adjust the original gray value GL of the current frame, In order to avoid unnatural or flickering images caused by drastic changes in screen brightness.

In addition, no matter what kind of gamma curve the original gray value GL uses, such as gamma curve G _i (X), G' _i (X), G _i-1 (X) or G' _i-1 ( X) After adjustment, there may be a color shift in the color of some pixels. For example, the original RGB gray value GL of a pixel that is approximately red is (255, 12, 12). After the pixel is adjusted by the above-mentioned gamma curve, its RGB gray value GL is adjusted to (255, 30, 30), so it appears pink. Therefore, an adjustment called "color purity weighting mechanism" is required, such as reducing the ratio of contrast enhancement, so that the picture f(N) looks more natural. In other words, this mechanism is used to ensure that when any one of the three colors of RGB in a pixel is close to saturation, there will be no color cast caused by the above-mentioned image processing method. That is to avoid the situation that the original gray level GL (255, 12, 12) of the pixel is adjusted to (255, 30, 30). The color purity weighting mechanism is GL _new = [GL*max(RGB)+GL'*(B-max(RGB))]/B. Where B is a positive integer, GL is the original gray value, and GL' is, for example, the gamma curve G _N (X), G' _N (X), G _N-1 (X) or G' _N-1 (X ) adjusted grayscale value GL' after adjustment, and max(RGB) in the formula is the maximum grayscale value among the original grayscale values GL. Taking the original gray value GL as (255, 12, 12), the adjusted gray value GL' as (255, 30, 30) and B as 256 as an example, then L _new = {(255, 12, 12 )X255+(255, 30, 30)X(256-255)}/256, that is, GL _new is approximately (255, 12, 12). In this way, when the original gray value GL (255, 12, 12) is adjusted by the above-mentioned various gamma curves, when the original red becomes close to pink (255, 30, 30), and then through this color purity weight The machine is adjusted to approximate the original chroma performance, that is, the new adjusted gray value GL _new (255, 12, 12), which makes the picture look more natural.

After some pixels are adjusted by the above-mentioned "color purity weighting mechanism", the contrast of the new adjusted gray value GL _new can actually be adjusted higher. Therefore, the present invention also includes another color purity weighting mechanism. The color purity weight mechanism is described as follows:

GL' _new ＝(P _LC *GL'+P _L *GL)/B

The above P _L =n*max(RGB)+m*color_gap, m+n=1 and P _LC = _BPL , color_gap=max(RGB)-min(RGB). Wherein B is also a positive integer, and color_gap is the difference between the largest gray value in the original gray value GL and the smallest gray value in the original gray value GL. This color purity weighting mechanism is used when the color of some pixels is close to white, that is, the gray values of the RGB three colors are very close to each other, and the contrast can be enhanced by this formula. That is, through the second color purity weighting mechanism, when the gray values of the RGB three colors of some pixels are very close to each other and the gray values of the RGB three colors are close to 255, the gray values of the RGB three colors can also be adjusted to enhance the intensity contrast. In other words, after the "second color purity weighting mechanism" is adjusted by the "first color purity weighting mechanism", the ratio of certain contrasts can be adjusted to a larger gray value because of the "first color purity weighting mechanism" When the adjustment causes the ratio of the contrast to decrease, re-adjust the ratio of the contrast to make the contrast of the image more obvious. For example, when GL (200, 198, 202), it is close to white. This GL (200, 198, 202 ) is adjusted by the "first color purity weighting mechanism", which is assumed to be GL _new (211, 210, 213). However, the contrast ratio of GL _new (200, 198, 202) can be a little higher, namely That is to say, the contrast is enhanced. Therefore, the ratio is adjusted to be larger through the "second color purity weighting mechanism", so that the contrast of the image is strengthened.

The image processing method disclosed in the above-mentioned embodiments of the present invention enhances the contrast of an image and maintains the stability of the image frame under the condition of maintaining the natural characteristics of the image.

In summary, although the present invention has been disclosed as above with a preferred embodiment, it is not intended to limit the present invention, and those skilled in the art can make various modifications without departing from the spirit and scope of the present invention. Movement and retouching, so the protection scope of the present invention shall be determined by the claims of the present invention.

Claims (8)

1. image treatment method, this image is presented by one first picture f (N) at least, and this first picture f (N) is made up of a plurality of pixel, and each pixel corresponds to an original gray value respectively, and this image treatment method comprises:

Neighbor original gray value to each other according to each pixel and this pixel is poor, adds up the appearance number of these original gray value, and with one first transfer function F _i(X) expression, i is the positive integer between 1 to N, and N is a positive integer, and X is 0 to 255 integer;

According to this first transfer function F _i(X) calculate a gamma curve;

Optionally adjusting these original gray value according to this gamma curve is the whole back of many styles gray value;

Wherein, this image also comprises one second picture f (N-1), this second picture f (N-1) is in the preceding appearance of this first picture f (N), and these original gray value of described adjustment comprise for these steps of adjusting the back gray value: this first transfer function F of integration this first picture f (N) _N(X), think a first integral value; The first transfer function F with this first picture f (N) _N(X) deduct the first transfer function F of this second picture f (N-1) _N-1(X) after, take absolute value and according to this integration go out a second integral value; And

When the ratio of this second integral value and this first integral value during greater than one second default value, these original gray value of this first picture f (N) are first transfer function F according to this first picture _N(X) be adjusted into these and adjust the back gray value, otherwise these original gray value of this first picture f (N) are first transfer function F according to this second picture f (N-1) _N-1(X) be adjusted into these and adjust the back gray value.

2. image treatment method as claimed in claim 1, wherein add up the step that number appears in these original gray value and also comprise:

When one second original gray value difference of one first original gray value of one first pixel in these pixels and adjacent one second pixel during greater than n, n is 0 or positive integer, the statistics number of all these original gray value between this first original gray value and this second original gray value or partly a statistics number average of these original gray value add 1.

3. image treatment method as claimed in claim 1, the step that wherein calculates this gamma curve also comprises:

This first transfer function of linear transformation F _i(X);

This first transfer function F after linearity conversion adds up _i(X), to obtain one second transfer function F ' _i(X);

This second transfer function of standardization F ' _i(X), to obtain one the 3rd transfer function F " _i(X); And

With the 3rd transfer function F " _i(X) be multiplied by K and add a parameter P (X) to obtain this gamma curve, this parameter P (X) is the transfer function of 1-K for slope, and K is between 0～1.

4. image treatment method as claimed in claim 1, the step that wherein calculates this gamma curve also comprises:

This first transfer function of linear transformation F _i(X);

This first transfer function F after linearity conversion adds up _i(X), to obtain one second transfer function F ' _i(X);

This second transfer function of standardization F ' _i(X), to obtain one the 3rd transfer function F " _i(X);

Adjust the 3rd transfer function F " _i(X) part upper lower limit value is to obtain one the 4th transfer function F " ' _i(X); And

With the 4th transfer function F " ' _i(X) be multiplied by K and add a parameter P (X) to obtain this gamma curve, this parameter P (X) is the transfer function of 1-K for slope, and K is between 0～1.

5. as claim 3 or 4 described image treatment methods, wherein adjust these original gray value and also comprise for these steps of adjusting the back gray value:

As this second transfer function F ' _i(X) the maximum that adds up is during greater than one first default value, and these original gray value adjust according to this gamma curve.

6. image treatment method as claimed in claim 2, wherein, these original gray value correspond to an original red gray value, an original green gray value and an original blue gray value respectively, and this image treatment method also comprises:

According to this original red gray value, this original green gray value or this original blue gray value ratio corresponding to the standard maximum gradation value, adjusting this adjustment back gray value is that one first colorimetric purity weight mechanism is adjusted the back gray value.

7. image treatment method as claimed in claim 6, wherein, this image treatment method also comprises: after the step of adjusting this adjustment back gray value, also according to this original red gray value, this original green gray value and this original blue gray value proportionate relationship to each other, adjusting this first colorimetric purity weight mechanism adjustment back gray value is that one second colorimetric purity weight mechanism is adjusted the back gray value, and this second colorimetric purity weight mechanism is:

GL’ _new＝(P _LC*GL’+P _L*GL)/B，

Wherein, GL is an original gray value, GL ' _NewBe that first colorimetric purity weight mechanism is adjusted the back gray value, B is a positive integer, P _LC=B-P _L, P _L=n*max (RGB)+m*color_gap, max (RGB) get maximum gray value among the original gray value GL, min (RGB) gets minimum gray value, color_gap=max (RGB)-min (RGB), m+n=1 among the original gray value GL.

8. image treatment method, this image is presented by one first picture f (N) at least, and this first picture f (N) is made up of a plurality of pixel, and each pixel corresponds to an original gray value respectively, and this image treatment method comprises:

Neighbor original gray value to each other according to each pixel and this pixel is poor, adds up the appearance number of these original gray value, and with one first transfer function F _i(X) expression, i is the positive integer between 1 to N, and N is a positive integer, and X is 0 to 255 integer;

According to this first transfer function F _i(X) calculate a gamma curve;

Optionally adjusting these original gray value according to this gamma curve is the whole back of many styles gray value;

Wherein, this image also comprises one second picture f (N-1), and this second picture f (N-1) is in the preceding appearance of this first picture f (N), and these original gray value of described adjustment comprise for these steps of adjusting the back gray value:

Number appears in these original gray value of adding up this first picture f (N), with one first pixel transfer function H _N(X) expression;

Number appears in these original gray value of adding up this second picture f (N-1), with one second pixel transfer function H _N-1(X) expression;

This first pixel transfer function of integration H _N(X), think a first integral value;

With this first pixel transfer function H _N(X) with this second pixel transfer function H _N-1(X) take absolute value after subtracting each other and, think a second integral value its integration; And

When the ratio of this second integral value and this first integral value during greater than one the 3rd default value, these original gray value of this first picture f (N) are this first transfer function F according to this first picture _N(X) be adjusted into these and adjust the back gray value, otherwise these original gray value of this first picture f (N) are this first transfer function F according to this second picture f (N-1) _N-1(X) be adjusted into these and adjust the back gray value.

Images