CN100472602C - Gamma correction method and circuit, image processing device and display device - Google Patents

Abstract

The invention relates to a gamma-ray correction method, which comprises a reverse correction processing and a form conversion processing. The reverse correction processing can be used for inputting X bit hue inputting-signal Tin values with a nonlinear hue-brightness characteristic, and for outputting Y bit hue signal Tout values which has conversed linear hue-brightness characteristic; wherein, X is less than Y. The form conversion processing can be used for inputting Tout, and for outputting (N added to M) bit hue outputting signals which is mostly close to Tout and consists of Tout and Tin, when the index indicated by a constant A-and-N bit signal Tn and the mantissa indicated by M bit signal Tm indicate the Tout with ATn multiplied by Tm.

Description

Gamma correction method and circuit, image processing device and display device

technical field

The present invention relates to a gamma correction method, program and device, an image processing method using the gamma correction method, program and device, and a display device.

Background technique

Various display devices such as cathode ray tubes (CRTs), liquid crystal displays (LCDs), and plasma display panels (PDPs) have inherent gamma characteristics and chromaticity characteristics unique to each display type.

The gamma characteristic of a monitor is the relationship between the monitor's input signal level and output brightness. On the other hand, the chromaticity characteristic indicates the chromaticity of three primary colors or more (for example, RGB+white) of each display. Since the three primary colors RGB are generally used in displays, the chromaticity characteristic generally indicates RGB chromaticity.

These characteristics depend on the type of display device, so gamma characteristics are largely different in each system of CRT, LCD, and PDP, and chromaticity characteristics are also different.

On the other hand, gamma characteristics (relationship between signal level and brightness displayed thereby) and chromaticity characteristics of input image signals such as television signals and sRGB are determined by predetermined specifications. Therefore, gamma correction processing and color space color space conversion processing are performed to achieve matching of gamma characteristics and chromaticity characteristics between the video signal and the display device.

Considering a display on a CRT, TV signals are usually subject to gamma correction and chroma conversion that match the CRT's gamma characteristics. Therefore, in most television receivers using CRTs, no special gamma correction processing is required on the CRT side. However, for television signal display on an LCD or PDP, gamma correction (the process of converting the gamma characteristics of the video signal to the gamma characteristics of the display device) at least in a general sense must be implemented in order to maintain high image quality. The gamma characteristics differ more than the chroma characteristics. Because the difference in gamma characteristics involves the generation of false contours and the color taste conversion of midtones, the image quality will deteriorate significantly if matching is not obtained by performing gamma correction in the general sense.

Here, gamma correction in a general sense is assumed, in order to utilize "gamma correction (in the narrow sense)" for converting linear hue-brightness characteristics to nonlinear characteristics and for converting nonlinear hue-brightness characteristics to "Inverse gamma correction" converted to linear characteristics, a series of processing required to match the video signal and the gamma characteristics of the display device. Hereafter, in this specification, gamma correction in a narrow sense is expressed as "positive gamma correction", and gamma correction in a general sense is directly expressed as "gamma correction".

Moreover, in order to accurately reproduce images and pictures produced in accordance with the sRGB specification (in other words, to achieve corresponding color reproduction), color space conversion is also required after inverse gamma correction. More specifically, a 3×3 matrix operation needs to be performed on the input RGB signal, as disclosed in Tajima's "Color Image Reproduction" (pp.33-39, Equation 3.11) (Reference 3) of.

However, in the case of performing inverse gamma correction processing using digital processing, the output bit number must be larger than the input bit number in order to maintain its accuracy. If the number of output bits is insufficient, image quality will deteriorate due to quantization errors (eg false contours).

On the other hand, when the number of output bits increases, the circuit scale of the inverse gamma correction circuit and the arithmetic circuit provided in the subsequent stage will increase with the number of bits (number of bits), resulting in a problem that the manufacturing cost of the display device will increase.

Also, for corresponding color reproduction, color space conversion needs to be performed after inverse gamma correction is done. In this case, when the number of output bits generated during inverse gamma correction increases, the circuit scale of the color space conversion circuit also increases.

As a background technology for solving the problem of increasing the number of digits, JP 10-126648 (Reference 1) discloses "Input Signal Level Application Type GammaCorrection Circuit in Liquid Crystal Display Apparatus (Input Signal Level Application Type Gamma Correction Circuit in Liquid Crystal Display Apparatus) horse correction circuit)". Reference 1 relates to a gamma correction circuit that uses fewer bits in conversion processing without deteriorating image quality.

The input signal level application type gamma correction circuit in the liquid crystal display device described in Reference 1 has a structure including an analog-to-digital (AD) converter and a digital-to-analog (DA) converter, and additionally includes a variable gain control amplifier located before the AD converter and after the DA converter.

The amplification degrees of these variable gain control amplifiers have a complementary relationship. Reference 1 improves the tonal characteristics of dark areas well by changing the degree of amplification according to the level of the analog input signal. Therefore, high-precision gamma correction is realized without increasing the number of bits.

As other background art, an example in which a floating-point number is used to indicate a hue signal of a display device is described in, for example, US Patent No. 5,528,741 (Reference 2).

However, Reference 1 is based on the fact that the input signal is an analog signal. Therefore, when the input signal is a digital signal, a DA converter is further required, which also complicates the circuit. Television signals (high-definition television (HDTV) and the like) and image signals (sRGB) are almost input as digital signals, and digital image signal processing is often performed. Therefore, it is preferable to employ a digital inverse gamma correction processing circuit capable of realizing inverse gamma correction with sufficient accuracy without using an AD converter and a DA converter.

Reference 2 discloses a method for efficiently converting floating-point representations of pixel values to one-byte integer values. In computer graphics applications, it is more convenient to use floating-point representations in pixel value calculations. However, since almost all display devices handle pixel data (pixel values) for pixel display as integers, display is impossible without converting the floating-point representation of pixel data to integer representation.

Therefore, in reference 2, the data in floating point representation is converted to integer representation. Also, while converting to integer representation, gamma correction is also performed. However, this structure achieves a simplified gamma correction process, but still involves an unsolved problem of an increase in the number of bits after gamma correction. Therefore, in the case where some arithmetic operations are performed after gamma correction, improvement in precision and downsizing of the circuit in the latter stage cannot be achieved.

Contents of the invention

The present invention has been made in consideration of the above-mentioned problems, and it is therefore an object of the present invention to provide a gamma correction circuit for achieving matching between the gamma characteristics of an input signal and the gamma characteristics of a display device, and A gamma correction method that performs gamma correction processing or inverse gamma correction processing while controlling the scale of the arithmetic circuit itself and a circuit set in a subsequent stage, and it is also possible to provide a gamma correction program and apparatus using the A gamma correction method, a program and an image processing method of the device, and a display device.

In order to achieve the above object, as a first aspect, the present invention provides a gamma correction method, the gamma correction method includes: inverse gamma correction processing, for inputting an X-bit tone input signal Tin with nonlinear tone-brightness characteristics value, and output the Y-bit (X<Y) tone signal Tout having the hue-brightness characteristic converted into a linear characteristic; When the mantissa indicated by the M-bit signal Tm expresses Tout as A^Tn×Tm, the (N+M) bit tone output signal formed by a group of Tn and Tm closest to Tout is output, wherein the value of the constant A is 2 The power of natural numbers, M satisfies the relationship M≤Y-N, when A is equal to 2, the minimum value of N is 2, when A is greater than 2, the minimum value of N is 1, and the maximum value of N is determined according to the constants A and M is the maximum value at which a set of Tn and Tm values becomes a different set of values relative to all Tin values.

According to the present invention, it is possible to provide a gamma correction circuit for achieving matching between gamma characteristics of an input signal and a gamma characteristic of a display device, and for performing gamma correction processing or inverse gamma correction processing with sufficient accuracy , a gamma correction method for simultaneously controlling the scale of the arithmetic circuit itself and a circuit set in a subsequent stage, and also providing a gamma correction program and device, an image processing method using the gamma correction method, program and device, and display devices.

Description of drawings

1A and 1B are graphs showing the relationship between the signal value before and after inverse gamma correction and the assigned tone with numerical lines.

Fig. 2 shows the processing of the gamma correction method to which the present invention is applied.

Fig. 3 shows a method for preventing tone recession when there is a pair of Tout' closest to Tout.

Figure 4 shows the relationship between M, N values and output signal accuracy;

FIG. 5 shows an input-output table for explaining a method of obtaining appropriate M, N values from the input-output table of inverse gamma correction processing in the gamma correction method.

FIG. 6 shows the structure of the liquid crystal display device of the first embodiment of the present invention.

FIG. 7 shows the structure of the digital inverse gamma correction circuit of the liquid crystal display device of the first embodiment.

FIG. 8 shows the configuration of the color space conversion circuit of the liquid crystal display device of the first embodiment.

FIG. 9 shows the structure of the positive gamma correction circuit of the liquid crystal display device of the first embodiment.

FIG. 10 shows the structure of the LUT provided in the positive gamma correction circuit of the liquid crystal display device of the first embodiment.

FIG. 11 shows the structure of a display device using a display system other than liquid crystal.

FIG. 12 shows the structure of a liquid crystal display device of a second embodiment of the present invention.

FIG. 13 shows the configuration of the digital inverse gamma correction circuit of the liquid crystal display device of the second embodiment.

FIG. 14 shows the structure of the color space color space conversion circuit of the liquid crystal display device of the second embodiment.

Fig. 15 shows the structure of a liquid crystal display device according to a third embodiment of the present invention.

FIG. 16 shows the structure of the digital inverse gamma correction circuit of the liquid crystal display device of the third embodiment.

FIG. 17 shows the configuration of the brightness-contrast correction circuit of the liquid crystal display device of the third embodiment.

Fig. 18 shows the structure of a liquid crystal display device of a fourth embodiment of the present invention.

FIG. 19 shows the relationship between the input-output voltage of the DAC in the source driver-signal generator and the output voltage of the reference tone signal generator.

FIG. 20 shows the relationship between the input-output voltage of the DAC in the source driver-signal generator and the output voltage of the reference tone signal generator when Tn is 3 (“11”).

FIG. 21 shows the relationship between the input-output voltage of the DAC in the source driver-signal generator and the output voltage of the reference tone signal generator when Tn is 2 (“11”).

Detailed ways

Standard image signals such as sRGB, scRGB, and television image signals are generally gamma-corrected. The hue value of an image signal that has been positively gamma corrected has a non-linear behavior with respect to luminance:

(hue value)=(brightness)^(1/γ)

As explained above, for corresponding color reproduction, a 3*3 matrix operation is required. However, when matrix operations are performed directly on a signal that has been positively gamma corrected, the hue of an image that includes a large number of complementary colors such as cyan, magenta, and yellow, as well as halftones, will change.

Therefore, before performing matrix operations, inverse gamma correction must be performed in order to provide linear hue values with respect to luminance.

However, since the number of input bits (the number of bits) is set larger than the number of output bits in order to perform digital inverse gamma correction, the scale of the matrix operation circuit of the latter stage becomes large.

As explained above, the increase in the number of bits required for inverse gamma correction results in a cause that allows an increase in circuit scale to achieve corresponding color reproduction.

Therefore, in the present invention, it is necessary to obtain tone accuracy after inverse gamma correction processing while controlling an increase in circuit scale by expressing the tone value of an image signal in a method different from the background art. More specifically, a method based on a floating-point representation is used to represent hue values.

The definition of hue expression using this method will now be described. In a conventional method, an 8-bit hue is expressed as a hue of 0 to 255 using an 8-bit binary number. On the other hand, this method expresses hue in the form of 2^Tn×Tm similar to floating-point representation. Here, Tn is a number called an exponent in the floating-point representation, and Tm is a number called a mantissa in the floating-point representation.

The difference from ordinary floating-point representation is that there is no need to add "1" to the bit one digit higher than the most significant bit of the mantissa. In the case of ordinary floating-point representation, it is determined to add "1" to the bit higher than the most significant bit of the mantissa to express the same using Tn and Tm combinations (2^1×38=2^2×19) value becomes impossible.

However, a direct representation that does not add "1" to the bit higher than the most significant bit of the mantissa by one digit is used in this method. However, the most significant bit is set to "1" whenever possible.

For example, in the case of 2^2×19, since the most significant bit of the mantissa (19: decimal number=010011: binary number) is “0”, it is expressed as 2^1×38, so that the most significant bit becomes "1". Also, 2^0×19 is represented as-is because the most significant bit of the mantissa is 0, but there is no exponent smaller than 0.

For example, where hue is represented as 2^Tn×Tm, it is possible to get:

0＝2^0×0≤2^Tn×Tm≤2^3×63＝504

And when the domain of Tn is set to 0 to 3 (2 bits) while the domain of Tm is set to 0 to 63 (6 bits), 8 bits are used to represent the value of 0 to 504.

This approach to hue representation is especially effective for exponentiation operations such as inverse gamma correction. Also, by introducing hue after inverse gamma correction based on floating-point representation, the circuit scale of 3×3 matrix operation and positive gamma correction can be reduced.

Advantages of using a floating-point representation-based method for hue representation:

The advantages of this approach to tone representation will now be explained. In inverse gamma correction, a floating-point representation-based method can distribute hues more efficiently with higher precision than ordinary hue representation methods. FIGS. 1A and 1B show the relationship between signal values before and after inverse gamma correction and assigned hues using numerical lines. For simplicity of illustration, the input is taken as three (3) bits and the output is taken as six (6) bits (in methods based on floating point representations, the exponent is two (2) bits and the mantissa is four (4) bits). The gamma value is 2.2 which is generally used.

When inverse gamma correction is performed, the input value is converted to a smaller value than before correction. The value after inverse gamma correction is assigned to one of the hues of the existing representation and the floating-point representation, as indicated by the number lines in the intermediate stages. In the lowest level of number lines, indicated by vertical lines, numbers can be represented by methods based on existing six (6) bit representations and six (6) bit floating point representations.

In the case of the existing representation, when going to the darker tone side, the number of digits that can be represented exists between the digits assigned to the input value after inverse gamma correction. This fact suggests that quantization errors are likely to occur when the tone is dark and the operation accuracy of matrix operations to be performed in the subsequent stage becomes low.

On the other hand, in the present method, almost the same number of digits to be represented exists between digits assigned to the input value after inverse gamma correction regardless of the darkness of the hue. This fact implies that the quantization error becomes smaller and the operation precision of the matrix operation to be performed in the subsequent stage becomes higher.

More specifically, when the Q value is equal to or greater than one (1) in an (output)=(input)^Q operation such as inverse gamma correction, the operation accuracy of the floating-point representation-based method becomes high.

Furthermore, since human visual characteristics show linear luminance characteristics equal to the 1/3 power of luminance according to the formula for obtaining L* of the L*a*b color display system (CIE1976), when performing inverse gamma correction The accuracy of dark tones needs to be improved. In other words, because quantization error directly manifests as chromatic aberration, this chromatic aberration can be reduced by reducing the quantization error of dark tones in the floating-point representation compared to existing representations.

As explained above, it can also be understood from the viewpoint of visual characteristics that this method is more preferable in terms of complete representation because the color difference becomes smaller.

It is also possible to make the circuit scale smaller by reducing the exponent and mantissa to two (2) bits and three (3) bits, respectively, according to the required precision of operation, by subtracting 1 bit is obtained.

Reference 2 introduced in Background Art discloses a gamma correction method for converting pixel values represented by floating points into one-byte pixel values. As explained above, there does exist a gamma correction process in which an input is given with a floating-point representation value to output an integer. However, the floating-point representation input data in Reference 2 is generated using operations in the field of computer graphics, and it is assumed that integer-valued data obtained using ordinary digital cameras, scanners, etc. to process. The present invention is characterized by a conversion method for converting an integer representation value into a value in a floating point representation based method. Therefore, Reference 2 cannot achieve "improvement in precision of an arithmetic circuit in a subsequent stage and reduction in circuit scale" which are objects of the present invention.

The above illustrates the advantages of methods based on floating-point representations. However, it is not necessarily possible to obtain the effects of the present invention by introducing the above-mentioned structure. That is, it is necessary to appropriately set the ranges of A, Tn, and Tm values in the A^Tn×Tm expression method according to the input-output table of actual inverse gamma correction. The effect of the present invention cannot be obtained with the desired A, Tn, and Tm values in the floating-point representation described in Reference 2.

The method for setting these values is explained below.

FIG. 2 shows the flow of the inverse gamma correction method applicable to the present invention. This inverse gamma correction method uses an X-bit signal as an input signal and an N+M-bit signal as an output signal.

The inverse gamma correction method includes: inverse gamma correction processing for inputting an X-bit tone input signal Tin value and outputting a Y-bit (X<Y) tone signal Tout; format conversion processing for inputting a Y-bit tone signal Tout , and output Tn and Tm which become closest to Tout when the hue signal Tout is expressed as A^Tn×Tm with a constant A, an exponent Tn of N bits, and a mantissa Tm of M bits; Processing of the output signal consisting of Tn and M bits Tm.

In this inverse gamma correction method, an X-bit tone input signal Tin is input (step S1). The tone input signal is subjected to inverse gamma correction (step S2). Here, in the inverse gamma correction process, Tin having a nonlinear hue-brightness characteristic is converted to linear and then output as Tout.

Here, since the hue input signal Tin is transmitted in compliance with, for example, the sRGB standard, it is sufficient to obtain Tout for converting the hue-brightness characteristic of Tin into a linear characteristic based on such a standard. It is assumed that inverse gamma correction processing is performed with a structure in which a conversion table such as a common look-up table (LUT) of the input value Tin and the output value Tout is stored in the memory. In summary, an inverse gamma correction process is performed in order to convert the nonlinear characteristics of Tin into linear characteristics.

In the Y-bit signal Tout obtained by the inverse gamma correction process, Y is usually larger than X (step S3). The reason for this is that, as described above, in the inverse gamma correction process, if the output bit number is not larger than the input bit number, the accuracy will decrease.

Thereafter, the Tout signal is converted in a format conversion process (steps S4, S5). In the format conversion process, the Tout value is converted into an N-bit signal Tn and an M-bit signal Tm.

First, the value of Tout' closest to Tout among the values that can be represented by the formula A^TnxTm is obtained in the format conversion process 1 (step S4). When there are two values of Tout' closest to Tout, in other words, when both values of Tout±α can be expressed as A^Tn×Tm, any value of Tout+α and Tout-α can be used Make Tout'. However, the rules for selecting the values used are preferably determined. Its reason is, when relational expression Tout(x)+β=Tout'(X), Tout'(X)+β=Tout(X+1), Tout(X+1)+β=Tout(X+1)+β= During Tout'(X+1), Tout'(X+1)+β=Tout(X+2), if Tout(X)+β=Tout'(X) is used as corresponding to Tout(X) value and Tout(X+1)-β=Tout'(X+1) is used as the value corresponding to Tout(X+1), tone degradation (recess) occurs. However, when Tout(X)+β=Tout'(X) is used as the value corresponding to Tout(X) and Tout(X+1)+β=Tout'(X+1) is used as the value corresponding to Tout( X+1), then even in this case there is never hue degradation.

Here, the constant A must be a natural number power of 2. When A is set to a natural number power of 2, multiplication of A^Tn and Tm can be performed using only data shift without using a multiplier circuit. If A is not a natural number power of 2, the scale of the operation circuit in the subsequent stage of the inverse gamma correction circuit cannot be reduced, and the advantage of using floating-point representation is lost. For example, when trying to perform 3^2×33+3^1×25 addition where the mantissa is defined as six (6) bits and A is defined as 3 bits, one needs to do 3^2×33=3^1 ×(3×33) to obtain a match between the Tn value of the first term and the Tn value of the second term. Also, after addition of the mantissa Tm (99+25=124), division (124/3) needs to be performed in order to set the mantissa of the mantissa to six (6) bits again. As described above, since addition requires multiplication and division, the effect that can be obtained when using floating-point representation cannot be obtained.

In addition, it is desirable that the value of M is equal to or smaller than Y-N. The reason for this is that gamma correction processing is often used to maintain accuracy, and the circuit scale of the LUT can be maintained or reduced by employing the inverse gamma correction circuit of the present invention. In order to reduce the circuit scale of the latter stage circuit, at least the number of digits of the mantissa must be smaller than the number of digits represented by the original integer, that is, the relational expression M<Y is required.

In addition, when A is 2, the value of N is required to be 2 or greater, and when A is greater than 2, the value of N is required to be 1 or greater. The reason for this is that if the value of N is 1 when A is 2, the value of A^Tn is only 1 or 2, and the one (1) bit data shift can be realized by setting Tm to M+1 bits. In addition, when A is equal to or greater than 4, even if Tm is set to M+1 bits, two (2) bit data shift of A^Tn cannot be realized. Therefore, it is reasonable to use N=1.

Here, the value of N is not sufficient to take the maximum value. The maximum value of N can be obtained from the value of Tout'. When N becomes large, the value of M must be subtracted from the relation Y≥M+N. As the N value becomes larger, the dynamic range of tones that can be represented becomes wider, but because the M value becomes smaller, the accuracy decreases. When the accuracy is reduced, the same value of Tout' cannot be obtained from other Tin values. When displayed, this is considered a tone recession problem and becomes far from the subject of the present invention. Therefore, from the above description, if a set of Tn and Tm values becomes a set of different values for all Tin values using the preset constant A and the number of digits M of the mantissa, it is impossible to use a value larger than the maximum N value. value.

From the above description, A, M, and N values are restricted, and Tn and Tm values can be obtained from these values in the format conversion process 2 (step S5). Here, the Tn and Tm values are set such that the Tm value becomes as large as possible. The reason for this is that, as the Tm value becomes larger, the accuracy of the color tone improves as a whole.

Figure 4 shows, when the value of M is increased (when M=m2>m1) and the value of N is increased (when N=n2>n1), the hue relative to M=m1 and N=n1 - Schematic diagram of changes in hue-accuracy characteristics of precision characteristics, where the horizontal axis represents hue and the vertical axis represents precision (the larger the value, the higher the precision). First, the tone-accuracy characteristics when M=m1 and N=n1 as a reference will be described. In this case, since the floating-point number cannot be shifted to less than 0 (when N=0) in the range where the tone value is small, when the tone value decreases, it is covered by the Tm value, thereby reducing precision. When the hue becomes higher, the precision of Tm can be used to the maximum. Therefore, the precision becomes constant (M digits). Next, as the value of N increases (when N=n2>n1), the floating point number is shifted in a wider tonal range. Therefore, the precision becomes almost constant except in the range where the hue takes only small values. However, since the M value does not change (M=m1), the accuracy when the tone value takes a larger value is the same as that when N=n1. Meanwhile, when the value of M increases (when M=m2>m1), the precision increases by (m1-m2) bits for all hue values. Therefore, it is preferable to make the Tm value preferentially larger than the Tn value.

However, if operation accuracy is required when the tone is low, it is preferable to increase the operation accuracy with respect to the lower tone while setting the Tn value to a certain value according to the required accuracy.

Finally, the obtained Tn and Tm are output as output signals (step S6).

These steps can be performed with a single process by summarizing them with only one LUT.

With the above structure, since Y≧M+N, the LUT size required for inverse gamma correction processing can be reduced compared to the existing structure. Furthermore, since Y>M, it is possible to reduce the operation scale of the multiplier circuit when performing multiplication in the subsequent stage of the inverse gamma correction processing (corresponding to the corresponding color reproduction processing). Furthermore, these processes yield a precision nearly equal to or higher than existing structures for providing Y-bit output. That is, with the inverse gamma correction of the present invention, image quality never deteriorates.

Conversely, in the case of positive gamma correction, from the above description, it can be expressed as Tout=A^Tn ×Tm (A is a constant), and then convert Tout into an X-bit (X<Y) tone signal Tin with nonlinear tone-brightness characteristics to reduce the LUT size required for positive gamma correction processing.

The ranges of the values of A, M and N for providing the effects of the present invention are explained above. Next, appropriate M and N values for inverse gamma correction processing will be investigated.

In this example, the value of A is set to 2 for the sake of illustration. However, even if A takes other values (natural number powers of 2), appropriate values of M and N can be obtained by changing the values by the same method as described below.

Appropriate M and N values can be obtained from the Tout value obtained in the inverse gamma correction process. An example of the input-output characteristics obtained in the inverse gamma correction process in the case of X=6 and Y=10 in FIG. 5 will be described below. FIG. 5 shows a table of comparison results between the outputs Tn and Tm of the format conversion process when the number of digits M of the mantissa is set to 5 and 6 under the condition of X=6 and Y=10. The input-output characteristics obtained by the inverse gamma correction processing are expressed by a monotonically increasing function in order to eliminate the occurrence of tone inversion (a phenomenon in which the relationship between lightness and darkness of an inherent signal is reversed).

First, study the Tout value near the darkest tone (minimum value) among the Tin values. Because the Tout value is converted into a value smaller than the input value in the inverse gamma correction process, the precision of Tin having a smaller value is lowered, as shown in FIG. 1 . Therefore, when Tin varies from 0 to about 3, the accuracy required for inverse gamma correction processing (ie, the number of output bits of inverse gamma correction processing: Y) is judged from Tout. In Fig. 5, because the values of Tout are 0, 1, 2 and 4, and the minimum difference is 1, when using other input values, never take the same value (that is, never produce tone degradation (recess)) . Therefore, the (Y) value may take a value of 10 as may be estimated.

Next, the M and N values are set such that the difference between adjacent tones of the Tm value obtained by performing conversion on Tout obtained from all the Tin values becomes equal to one (1) or more. More specifically, study the difference from the minimum Tout with the most significant bit being one (1). When Y=10, the Tout value that the most significant bit of Tout becomes one (1) for the first time (that is, becomes equal to or greater than 2^9=512) is equal to 520 (1000001000), as shown in Figure 5, and Tout increases Large, such as 546 (1000100010), 571 (1000111011) and 598 (1001010110). In this case, when the M value is equal to 5, the Tm values obtained from these Tout values become 10000, 10001, 10001, and 10010, respectively. In this case, Tm of Tout=546 and Tm of Tout=571 become 10001, the same value continues, and tone recession occurs. Meanwhile, when the M value is equal to six (6), the Tm values obtained from these Tout values become 100000, 100010, 100011, and 100101, respectively, and the Tm values corresponding to the respective Tout values become different. Therefore, a suitable M value is six (6) when the accuracy is equal from dark to light tones, and Tn values must be set from 0 to 4. In order to represent Y-bit (10-bit) data as ÂTn×Tm under the conditions of M=6, A=2, and Y=10, it is required to use Tn that makes 2Tn×63≈2̂10−1. Because Tm is equal to 10010 when Tout is equal to 546 (1000100010), it can be expressed as A^Tn×Tm and the value closest to 546 is equal to 544 (=2^4×34), and the value of Tn is required until 4.

In addition, it is also required to satisfy the relationship Tn=2^(2^N-1)≥2^(Y-M), because the remaining Y-M bits of data obtained by removing M bits of higher significant bits from the initial Y bits of data The shift is represented by N. When solving for N, the result of N≥log(1+Y-M)/log2 can be obtained. Therefore, when the M value is determined, the N value that is the smallest integer value equal to or greater than log(1+Y-M)/log2 can be obtained. In the above example, since log(1+10-6)/log2=2.322, the value of N becomes 3 as the minimum value.

Also, increase the M or N value for greater precision. For example, since the accuracy of bright tone is mainly determined by M, it is sufficient to increase the M value. In addition, since the accuracy of the dark tone is also related to the M and N values, it is enough to increase the M or N value.

Using the method described above, the values of inverse gamma correction processing can be used to set appropriate M and N values.

Preferred embodiments of the present invention based on the above principles will be described below.

first embodiment

A first preferred embodiment of the present invention will now be described. FIG. 6 shows the structure of the liquid crystal display device of this first embodiment. This liquid crystal display device is equipped with an image processor 1 and a liquid crystal display 12 . This liquid crystal display device inputs a digital image signal (for example, RGB signal, six bits for each color) into the image processor 1 as an input signal, and outputs the digital image signal to the liquid crystal display after performing various operations in the image processor 1. 12. Determine the transmittance of the pixels of the liquid crystal display 12 according to the digital image signal, and display an image according to the digital image signal.

Here, the input-output signal of the image processor is represented by hue representation of 6 bits per color, but the number of bits is not limited to six (6) bits.

The image processor 1 is a functional unit for performing corresponding color reproduction by converting an image signal according to the chromaticity characteristics of the liquid crystal display 12 . The image processor 1 includes an inverse gamma correction circuit 2 , a color space conversion circuit 3 and a positive gamma correction circuit 4 . An inverse gamma correction circuit 2 and a positive gamma correction circuit 4 are provided individually for each color element. The digital image signal input to the image processor 1 is subjected to sequential operation processing in the inverse gamma correction circuit 2, the color space conversion circuit 3 and the positive gamma correction circuit 4, and the output of the positive gamma correction circuit 4 is sent as an output signal to A DA converter (not shown) in the source driver and control signal generator 34 .

The LCD 12 includes a scan driver 33 , a source driver and control signal generator 34 and a pixel matrix 38 .

The pixel matrix 38 includes a plurality of scanning lines 31 , a plurality of signal lines 32 , a plurality of pixels 35 , a plurality of auxiliary capacitors 36 and a thin film transistor (TFT) 37 . Each scanning line 31 and each signal line 32 cross each other, and a pixel 35 is provided at each intersection of the scanning line 31 and the signal line 32 through a TFT 37. Each pixel 35 is connected in parallel with an auxiliary capacitor 36 .

The scan driver 33 controls signals to be respectively input to the plurality of scan lines 31 . The source driver and control signal generator 34 controls signals to be respectively input to the plurality of signal lines 32 .

Here, the processing from inputting a digital image signal to the liquid crystal display device until displaying an image on the liquid crystal display 12 will be described below. The image processor 1 outputs digital signals by performing operation processing on input digital signals. Output image signals and display control signals (not shown) are sent to the scan driver 33 and the source driver and control signal generator 34 . The source driver and control signal generator 34 performs DA conversion on the digital image signal based on the applied voltage-brightness characteristic of the pixel 35 of the liquid crystal display 12 and the conversion characteristic obtained from the gamma characteristic of the input image signal. The signal converted into an analog signal is applied to the pixel 35 connected to the scan line 31 to which the ON voltage is selectively applied by the scan driver 33 through the TFT 37, and is then output as an image signal by being converted into luminance.

FIG. 7 shows the detailed structure of the inverse gamma correction circuit 2. As shown in FIG. The inverse gamma correction circuit 2 is composed of an exponent look-up table (LUT) 21 and a mantissa LUT 22. The exponent LUT 21 and the mantissa LUT 22 receive the same digital signal as an input signal. In these exponent LUT 21 and mantissa LUT 22, a value according to an input signal is obtained by referring to the LUT, and then the value is output as an output digital signal. The output of the exponent LUT 21 is a three (3) bit value (exponent signal) and the output of the mantissa LUT 22 is a six (6) bit value (mantissa signal). Signals respectively output from the exponent LUT 21 and the mantissa LUT 22 are sent to the color space conversion circuit 3 .

Here, the exponent LUT and the mantissa LUT are formed as separate structures, but it is also possible to provide only one LUT with a nine (9) bit output, where six (6) bit outputs are provided for the mantissa and the remaining three (3) bits are provided give index.

FIG. 8 shows the detailed structure of the color space conversion circuit 3. As shown in FIG. Color space conversion circuit 3 is made up of following: nine multiplier circuits, are used for the multiplication of RGB mantissa signal and three constants (axy (x, y=1,2,3)); Data shifter 23 is used for according to The RGB exponent signals perform data shifts on the corresponding multiplication results; six (6) adders for adding up the data shifted results; and a signal converter 24 for splitting the addition results into exponents and mantissas again. Obviously, from the structure shown, the color space conversion circuit 3 performs 3×3 matrix operations.

FIG. 9 shows the structure of the positive gamma correction circuit 4 . The positive gamma correction circuit 4 is constituted by a two-dimensional LUT 25, in which the exponential signal and the mantissa signal are respectively defined as the number of rows and columns in the secondary two-dimensional arrangement. The positive gamma correction circuit 4 refers to the two-dimensional LUT 25 to obtain a value according to the input exponent signal and the input mantissa signal, and then outputs the value as an output signal (six (6) bits).

Next, the operation of the liquid crystal display device of this embodiment and the achieved effects will be described.

In the image processing circuit of FIG. 6, by adopting the color tone representation based on the floating-point number system as the output of the inverse gamma correction circuit 2, the circuit of the color space conversion circuit 3 in the subsequent stage can be reduced without reducing the accuracy scale. The operation of each circuit will now be described in detail along the flow of image signals.

The inverse gamma correction circuit 2 shown in FIG. 7 performs inverse gamma correction using a LUT, and converts the tone expression method to a method based on floating-point expression.

In the case of performing inverse gamma correction on a six (6) bit signal, if the corrected output is set to 10 bits or more (2^(9-1) = 512 tones or more), use normal tones The representation (fixed-point) cannot maintain precision, but with methods based on floating-point representations, it is possible to obtain a value equal to Or a value larger than 10 bits (2^4×63=1008 tones) displayed by the fixed-point method. That is, since the ten bits required for ordinary representation can be reduced to nine (9) bits (=3+6), the inverse gamma The structural size of the horse correction circuit 2 can be reduced.

In the subsequent stage of each multiplication circuit, unlike an ordinary matrix operation circuit, the matrix operation circuit (color space conversion circuit 3) shown in FIG. A circuit that shifts data by exponent value. When the exponent of the input signal is defined as Xa, and the mantissa is defined as Xb (X=R, G, B), the following formula (1) is used to represent from the signal input to the color space conversion circuit 3 until the signal converter 24 processing in the processing.

Rout=2^Ra×(Rb×a11)+2^Ga×(Gb×a12)+2^Ba×(Bb×a13),

Gout=2^Ra×(Rb×a21)+2^Ga×(Gb×a22)+2^Ba×(Bb×a23),

Bout＝2^Ra×(Rb×a31)+2^Ga×(Gb×a32)+2^Ba×(Bb×a33) (1)

In the data shifter 23, the operation (X=R, G, B) in the formula (1) is performed 2^Xa times by Xa bit shifting. Then, the operation output is output by being converted into an exponent and a mantissa by the signal converter 24 . Since the data shifter 23 performs exponent operation, the multiplication process can be simplified. Therefore, when the exponent operation performed by the data shifter 23 is, for example, a power-raising operation of 2, such as 4^Ra times and 8^Rb times, the effect of the present invention can be obtained.

Here, since the number of bits of the mantissa is equal to the number of input bits, the circuit scale of the multiplier circuit can be significantly reduced. In a common structure, 10 bits x 6 bits operation is performed, but this operation is converted into six (6) bits x six (6) bits operation, resulting in a circuit scale reduction of about 40%. Also, since the circuit scales of the data shifter 23 and the signal converter 24 are not too large, the circuit structure can be reduced by about 30% compared to the conventional structure.

Next, positive gamma correction based on floating-point representation is performed using a two-dimensional LUT (two-dimensional LUT 25 ) in which exponent values and mantissa values are respectively set as xy axes. Figure 10 shows the LUT structure of a positive gamma correction circuit, where the input consists of a three (3) bit exponent (using only 0 to 4) + six (6) bits of a mantissa, and the output consists of six (6) bits. Since the exponent value ranges from 0 to 4 and the mantissa value ranges from 0 to 63, the number of bits required for the LUT reaches 1920 bits (=5×64×6 bits). However, when the value of the exponent is equal to or greater than one (1), and the most significant bit of the mantissa is 0, the value of the exponent is decreased by one (1) and the value of the mantissa is doubled. Thus, LUTs do not necessarily need to be saved for all combinations of exponent and mantissa (the fields enclosed by the double box in Figure 10 (fields with exponent values equal to or greater than one (1) and the most significant bit of the mantissa being 0) are not required).

As shown in FIG. 10, when the exponent is equal to "0", the mantissa takes any value from "0 to 63", and when the exponent is equal to "1 to 4", the mantissa takes any value from "32 to 63". Since the output is six (6) bits in any case, the size of the positive gamma correction circuit 4 can become as follows.

When the exponent is equal to "0", the mantissa can take values from "0 to 63", that is, 2^6 values. Since the output is six (6) bits for each value, 2Λ6x6 bits of memory are required. At the same time, when the exponent is equal to "1 to 4", the mantissa can take a value from "32 to 63", that is, 2^5 kinds of values. Since the output is six (6) bits for each value, 2Λ5x6x4 bits of memory are required. That is, (2̂6×6) bits + (2̂5×6×4) bits of memory are required for each color. In the case of an RGB color image, as many bits as above are required for each color element (however, it is assumed that the input-output bit number for each color element of RGB is the same). Therefore, the number of bits required for the LUT is expressed as follows.

Input 3+6 bits --> output six (6) bits:

(2^6×6+2^5×6×4)×3(RGB)=152×3 bits

In the background art case, a positive gamma correction circuit converts a 10-bit input signal into a 6-bit output signal, so the required number of bits is 2^10×6×3=6144×3 bits. Therefore, the circuit scale can be greatly reduced in the present invention.

Here, the circuit scale of the matrix operation circuit (adder circuit, multiplier circuit) and positive gamma correction circuit (LUT) in the subsequent stage can be reduced by applying a method based on floating-point representation.

With the above-described effects, it is possible to provide an image processing device and a display device that reduce the circuit scale by applying a digital gamma correction circuit.

In the above structure, the inverse gamma correction circuit is composed of a combination of an inverse conversion circuit and a format conversion circuit. The embodiment is not limited thereto, and the effects of the present invention can also be obtained when the inverse gamma correction circuit and the format conversion circuit are separately formed.

In the above description, the number of digits of digital tone is set to six (6) bits, but the present invention is not limited thereto, and when such number of digits of digital tone is set to a desired number of bits, such as eight (8) bits, The same effect can be obtained.

Furthermore, in the present invention, a liquid crystal display is used as the display, but the present invention is not limited thereto. Similar effects can be obtained even when other display devices (PDPs, displays using electroluminescence elements, or liquid crystal projectors, etc.) are used as shown in FIG. 11 .

second embodiment

A second embodiment of the present invention will be described below.

Fig. 12 shows the structure of the liquid crystal display device of the second embodiment. This liquid crystal display device is almost the same as the image processor 1 of the liquid crystal display device of the first embodiment, but the structures of the inverse gamma correction circuit 2A, the color space conversion circuit 3A, and the positive gamma correction circuit 4 are the same as those of the liquid crystal display device of the first embodiment. The equipment is different. The positive gamma correction circuit 4 is the same as the first embodiment except that six (6) bits of data are output with respect to two (2) bits+six (6) bits of input data.

FIG. 13 shows the configuration of the inverse gamma correction circuit 2A. The inverse gamma correction circuit 2A includes an exponent LUT 21A and a mantissa LUT 22. The mantissa LUT 22 is similar to that of the liquid crystal display device of the first embodiment. The index LUT 21A differs from that of the first embodiment in that, for each color, two (2) bit values are output to six (6) bit input image signals. As in the first embodiment, the same signals are input to the exponent LUT 21A and the mantissa LUT 22.

In the first embodiment, a three (3) bit value output as the index LUT 21 is directly used as an index for hue representation. Meanwhile, in this second embodiment, the output of the index LUT 21A is not directly used as an index for tone representation, and it is encoded (assigned) to other values (such as 0, 2, 3, 4 instead of 0, 1, 2, 3). That is, the number of output items of the index LUT 21A is equal to the number of values that the output value of the index LUT 21A can take, and the output of the index LUT 21A is assigned to an integer number equal to or greater than 0 that is incremented from the initial item by unequal difference. The value of any item of the formed sequence.

A value greater than the maximum value "3" of the output of the LUT 21A can be set to Ra by encoding such that Ra = 0 when the two (2) bit outputs of the index LUT 21A are 0, and Ra = 0 when the output is 1. =2, when the output is 2, Ra=3, and when the output is 3, Ra=4. When Ra is 1, the output value of the mantissa part changes, so that Ra becomes 2 (2^1×18=2^2×9).

By performing this encoding, tone representation can be achieved over a wider dynamic range while the number of output bits of the Index LUT 21A is controlled. Since the number of output bits is controlled, the circuit scale of the index LUT 21A can be reduced.

Exponential data assigned other values is sent to the color space conversion circuit 3A. FIG. 14 shows the configuration of the color space conversion circuit 3A. This circuit is almost the same as the color space conversion circuit 3 of the liquid crystal display device of the first embodiment except that the data shifter 23A is substituted for the data shifter 23 . The data shifter 23A performs decryption corresponding to encoding in the index LUT 21A (resetting the LUT 21A output value converted into another value to the initial value). When the scale of the circuit used for this decoding is sufficiently small, the total circuit scale can also be reduced. In other words, when the scale of the circuit for converting the output value of the index LUT 21A assigned to other values is smaller than the circuit scale of the LUT 21A reduced by assigning the output value to other values, the total circuit scale can also be reduced .

With the above configuration, it is possible to reduce the circuit scale of the color space conversion circuit (matrix operation circuit (adder circuit and multiplier circuit)) and the positive gamma correction circuit (LUT) in the subsequent stage of the inverse gamma correction circuit.

In addition, by implementing the digital gamma correction circuit, it is also possible to reduce the circuit scale of the image processing device and the display device.

third embodiment

A third embodiment of the present invention will be described below.

In the first and second embodiments, the image processing circuit is provided in the subsequent stage of the digital gamma correction circuit for corresponding color reproduction. In this embodiment, a configuration will be described in which an image processing circuit for a purpose different from that of the corresponding color reproduction is provided in a subsequent stage of the inverse gamma correction circuit.

Fig. 15 shows the structure of the liquid crystal display device of the third embodiment. This liquid crystal display device is almost the same as that of the first embodiment except that an inverse gamma correction circuit 2B is substituted for the inverse gamma correction circuit 2, and a brightness-contrast correction circuit 5 is substituted for the color space conversion circuit 3. .

FIG. 16 shows the configuration of the inverse gamma correction circuit 2B. As in the inverse gamma correction circuit 2 of the first embodiment, the same image signal is input to both the exponent LUT 21B and the mantissa LUT 22B. The exponent LUT 21B outputs a three (3) exponent signal with respect to a six (6) bit input signal, and the mantissa LUT 22B outputs a six (6) bit mantissa with respect to a six (6) bit input signal.

The RGB image signal input to the image processing circuit 1 is divided into an exponent and a mantissa in the inverse gamma correction circuit 2B, and is output as an exponent signal and a mantissa signal.

In the inverse gamma correction circuit 2B, inverse gamma correction of the gamma value 1 can be performed. In the case of performing inverse gamma correction with a gamma value other than one (1), it is sufficient to use LUTs similar to those in the first and second embodiments, as the exponent LUT 21B and the mantissa LUT 22B. Meanwhile, when inverse gamma correction is performed with a gamma value of 1, it is sufficient to use a LUT to convert an input signal into a hue representation of a floating-point system using a method similar to that described above.

As shown in FIG. 16, in the inverse gamma correction circuit 2B, since the input bit number is six (6) bits, the exponent output bit number is three (3) bits, and the mantissa output bit number is six (6) bits, therefore The conversion is performed so as to obtain sufficient precision to such an extent that the maximum value of the exponential part is not allowed to increase too much as in the case of the above-described embodiments. That is, when the input signal is equal to the maximum value of 63 of the tone signal, the exponent is equal to 3, and the mantissa is equal to 63, 2^3×63=504. When the input signal is 1, the exponent is equal to 0, and the mantissa is equal to 8, because 504/63=8=2^0×8.

The signal output from the inverse gamma correction circuit 2B is then input to the brightness-contrast correction circuit 5 . As shown in FIG. 17, when the exponent of the signal input to the brightness-contrast correction circuit 5 is a and the mantissa is b, the following processing is performed.

(output)=2^a×(b×c1)+c0

In the data shifter 23B, the formula operation is performed 2a times or more by the bit shift of a. Then, the operation output is again converted into an exponent and a mantissa by the signal converter 24 for output.

Here, when the c1 value is set to 1 and the c0 value is set to 0, it is equivalent to doing nothing at this step. When the c1 value is set to 1 or more, a higher contrast image can be obtained, and when c1 is set to 1 or less, a lower contrast image can be obtained. In addition, c0 represents compensation, and when c0 takes a value greater than 0, image brightness can be increased. Furthermore, by changing the combination of c0 and c1, the desired contrast-brightness correction can be achieved.

When the input tone value is small, the operation accuracy in the brightness-contrast correction circuit 5 becomes higher when the tone expression is performed with respect to the exponent and the mantissa, respectively, compared with ordinary tone expression. Therefore, even when the circuit of the subsequent stage of inverse gamma correction performs image processing for a purpose other than that of the corresponding color reproduction, it is possible to achieve this without enlarging the arithmetic circuits (adder circuit, multiplier circuit) and the positive gamma correction circuit ( LUT) in the case of the circuit scale, to achieve higher precision calculations.

In addition, application of this digital gamma correction circuit enables reduction of the circuit scale of the image processing device and the display device.

Fourth embodiment

A fourth embodiment of the present invention will be described below.

In each of the embodiments described above, the gamma correction processing is performed using digital processing in the gamma correction circuit of the image processing circuit. In this embodiment, a structure in which positive gamma correction processing is performed in the DA converter will be described.

Fig. 18 shows the structure of the liquid crystal display device of this embodiment. The liquid crystal display device includes an image processor 1 , a selector 6 , a reference tone voltage generator 7 and a liquid crystal display 12 .

The selector 6 sequentially selects one set from a plurality of sets of exponent signals and mantissa signals input from the color space conversion circuit 3, and then outputs the selected set.

The reference tone voltage generator 7 outputs a plurality of reference voltages for DAC conversion.

The liquid crystal display device of this embodiment utilizes the converter (DAC) and the reference tone voltage generator 7 in the source driver and control signal generator 34 to perform the positive gamma correction circuit 4 used in the liquid crystal display device of the first embodiment. Processing performed (positive gamma correction).

The exponential signal among the output signals of the selector 6 is output to the reference tone voltage generator 7 , and the mantissa signal is output to the DAC in the source driver and control signal generator 34 .

The configuration shown in FIG. 18 performs positive gamma correction by performing serial conversion on the RGB parallel signals selected by the selector 6, but it is also possible to use this configuration to separately include a reference for each color of RGB separately. The tone signal generator 7 and the configuration of the DAC perform processing in parallel.

The reference tone voltage generator 7 outputs desired reference tone voltages V1 to V9 corresponding to the input exponential signal, and sends these voltages to the DAC in the source driver and control signal generator 34 . In this DAC, reference tone values D1 to D9 of the mantissa signals are set as reference tone voltages V1 to V9 as output voltages when input digital signals are input, respectively. That is, when the input mantissa signal is D1, V1 is output as the output of the DAC, and when the input mantissa signal is D2, V2 is output. If it is assumed that D1<D2<...<D9, when a hue between D1 and D2 is input to the DAC, the voltage obtained by linear interpolation of the two reference tone voltages V1 and V2 sandwiching this hue is output. Figure 19 shows the input-output characteristics of this DAC. A dotted line indicates a desired input-output characteristic (an input-output characteristic suitable for the gamma characteristic of the liquid crystal display 12), while a solid line indicates an input-output characteristic of a DAC. As shown, the input-output characteristics of the DAC approximate those obtained from the desired input-output characteristics using multiple segments. This input-output characteristic can be uniquely determined when a display device is determined.

Here, a method for determining the reference tone voltage in the reference tone voltage generator 7 will be explained. The reference tone voltage generator 7 is capable of generating as many sets of reference tone voltages V1 to V9 as the number of values that the input exponential signal can take. For example, when the exponential signal is two (2) bits, four (=2 ² ) groups of voltages V1 to V9 can be generated. An example will be described below with reference to FIGS. 20 and 20 .

For example, it is assumed that the exponent signal Tn and the mantissa signal Tm which are the output values of the selector 6 in FIG. 18 are two (2) bits and six (6) bits, respectively (the exponent is two (2) bits and the mantissa is six (6) bits) , and the hue value is expressed as 2^Tn×Tm (ie, A=2). Furthermore, when Tn="11 (3 in decimal)" is set, it is assumed that a set of reference tone voltages generated by the reference tone voltage generator 7 are V1M, V2M, . . . , V9M. Assume that the mantissa signal values corresponding to this voltage value are D1M, D2M, ..., D9M, and also assume D1M="000000", D5M="011111", D8M="111110", D9M="111111" below. In this case, the tone value corresponding to the V5M output voltage becomes 2̂Tn×D5M=2̂3×31=248 (refer to FIG. 20 ).

Here, when Tn="10 (2 in decimal)", sets of reference tone voltages V1M2, V2M2, . . . , V9M2 are set as follows. For example, since the hue value corresponding to V1M2 is 2^Tn×D1M=2^2×0=0, V1M2=V1M can be obtained (because 2^3×0=0). In addition, since the hue value corresponding to V8M2 is 2^Tn×D8M=2^2×62=248, V8M2=V5M (because 2^3×31=248) can be obtained (refer to FIG. 21 ). As described above, based on V1M to V9M obtained when Tn="11", V1 to V9 when Tn is other values are set. In other words, a reference tone signal when Tn takes a maximum value is first obtained, and then based on this reference tone signal, a reference tone signal when Tn takes a value different from the maximum value is obtained.

In the above example, nine reference tone values (D1 to D9) of the mantissa signal were considered. However, the number of reference tone values of the mantissa signal is by no means limited thereto.

In FIG. 19, D1 is required to take the minimum value of the mantissa signal, while D9 is required to take the maximum value of the mantissa signal, so as to convert all the mantissa signals according to the characteristic of positive gamma correction. Also, as a linear interpolation method, it is considered to use resistor strings to obtain partial resistor voltages.

From the above description, it is obvious that the output of the reference tone signal generator 7 (reference tone voltage) is used to determine the voltage after conversion in the DAC.

Here, the processing from inputting a digital image signal to the image processor 1 until displaying an image on the liquid crystal display 12 will be described. In the DAC, DA conversion is performed on the signal output from the image processor 1 in accordance with the applied voltage-brightness characteristic of the pixel 35 of the liquid crystal display 12 and the conversion characteristic obtained from the gamma characteristic of the input image signal.

The reference tone signal generator 7 generates reference tone voltages V1 to V9 for the DAC to perform conversion, and then outputs them to the DAC. Here, the reference tone voltages V1 to V9 are set using the conversion characteristic and the exponential signal. Then, the signal converted into an analog signal in the DAC is applied to the pixel 35 through the source driver and control signal generator 34 and the TFT 37. This analog signal is converted to a luminance signal, which is then used to generate an image.

More specifically, the index signal output from the selector 6 is sent to the reference tone signal generator 7 as shown in FIG. 18 . The reference tone signal generator 7 receives the exponential signal, generates a reference voltage providing 2^(exponential signal) times the tone of the input signal, and then outputs this reference voltage to the DAC. Since the reference tone signal generator 7 generates a reference tone signal that provides 2^(exponential signal) times the tone of the input signal, although the mantissa signal is input to the DAC, the output analog signal becomes 2^(exponential signal)×(mantissa signal ), and then output.

With the above-described structure, positive gamma correction can be realized using the reference tone signal generator 7 and the DAC without digitally synthesizing signals divided into exponents and mantissas. Furthermore, it was demonstrated that conversion can be performed with higher precision by exponentially modulating the reference tone signal while maintaining the precision of the DAC at six (6) bits.

With the above-described structure, as is the case with the above-described respective embodiments, it is possible to perform high-precision operation by dividing the tone signal into an exponent and a mantissa.

Furthermore, by implementing this digital gamma correction circuit, the circuit scales of the image processing device and the display device can be reduced.

Each of the embodiments described above is an example of a preferred embodiment of the present invention, and the present invention is by no means limited thereto. For example, a liquid crystal display device is mainly described in each embodiment, but the present invention is by no means limited thereto, and the present invention is also applicable to a display device using a PDP and organic electroluminescence.

Furthermore, the image processor 1 is formed as a circuit in each of the above-described embodiments, but it is also possible to use a computer as the image processor 1 by software processing. In this case, the inverse gamma corrector and the positive gamma corrector are formed in an arithmetic device (central processing unit (CPU) etc.) for executing the inverse gamma correction program and the positive gamma correction program so as to perform processing in one embodiment. Naturally, both inverse gamma correction and positive gamma correction can be realized by software processing, and any of these corrections can be realized by software processing. In addition, it is also possible to perform these corrections by software processing including processing (color space conversion processing, brightness-contrast correction processing, etc.) to be performed between inverse gamma correction and positive gamma correction.

In addition, it is also possible to combine structures related to the respective embodiments within a possible range. As described above, the present invention allows various modifications.

The previous embodiments are described to enable any person skilled in the art to make and use the invention. Moreover, various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles and specific examples defined herein may be applied to other embodiments without the teachings of the present invention. Thus, the invention is not to be limited to the embodiments described herein, but is to be accorded the widest scope as defined by the limitations of the claims and equivalents.

Further, it is the inventor's intent to retain all equivalents of the claimed invention even if the claims are amended during prosecution.

Claims (19)

1. A gamma correction method, comprising: Input an X-bit tone input signal Tin value with a nonlinear tone-brightness characteristic, and output a Y-bit (X<Y) tone signal Tout with a tone-brightness characteristic converted into a linear characteristic, thereby realizing inverse gamma correction; as well as Input the Tout, and when the exponent indicated by the constant A, the N-bit signal Tn and the mantissa indicated by the M-bit signal Tm express the Tout as A^Tn×Tm, output a group of Tn closest to the Tout and the (N+M) bit tone output signal formed by Tm, thereby realizing format conversion; Where A is a natural power of 2, M satisfies the relation M≤Y-N, When A is equal to 2, the minimum value of N is 2, when A is greater than 2, the minimum value of N is 1, and The maximum value of N determined from A and M is such that a set of Tn and Tm values becomes a different set of values for all Tin values. 2. The gamma correction method according to claim 1, wherein Tn has a predetermined value encoded as an element of a sequence consisting of N integers equal to or greater than 0 incremented from an initial value by unequal difference. 3. The gamma correction method according to claim 1, wherein A=2, M=Y-Z', and N is set to a minimum integer value equal to or greater than 1+logZ'/log2, when used for input The inverse gamma correction table of the Tin value and outputting the Tout value is predetermined, under the condition that the Tout value after being reduced by Z bits less significant bits is set to Tz, obtained with respect to all Tout values Of the Tz values, only those Tz values whose most significant bit value is 1 are extracted, and the maximum value among Z values of different values of all extracted Tz values is set to Z′. 4. A gamma correction method, comprising: Input comprises the (N+M) bit signal of N bit signal Tn and M bit signal Tm, as input signal, and output with constant A, N bit exponent Tn and M bit mantissa Tm to express as Tout= ^ATn *Tm, have Y-bit tone signal Tout of linear tone-brightness characteristics, thereby realizing format conversion; and Input Tout value and output X bits with non-linearly converted hue-luminance characteristic (X<Y) tone signal Tin, thereby realizing gamma correction processing, Where A is a natural power of 2, M satisfies the relation M≤Y-N, When A is equal to 2, the minimum value of N is 2, and when A is greater than 2, the minimum value of N is 1, and The maximum value of N determined from A and M is such that a set of Tn and Tm values becomes a different set of values for all Tin values. 5. The gamma correction method according to claim 1, wherein the inverse gamma correction and the format conversion are performed using only one lookup table. 6. An image processing method, comprising: After the gamma correction according to claim 1, a color space conversion process is performed. 7. An image processing method, comprising: After the gamma correction method according to claim 1, brightness-contrast correction processing is performed. 8. A gamma correction circuit, comprising: It is used to input the X-bit tone input signal Tin value with nonlinear tone-brightness characteristics, and output the Y-bit (X<Y) tone signal Tout with tone-brightness characteristics converted into linear characteristics, thereby realizing inverse gamma corrected circuitry; and It is used to input the Tout, and when the Tout is expressed as A ^Tn ×Tm with the constant A, the exponent indicated by the N-bit signal Tn and the mantissa indicated by the M-bit signal Tm, the output is composed of a group of values closest to the Tout The (N+M) bit tone output signal formed by Tn and Tm, thereby realizing the circuit of format conversion; Where A is a natural power of 2, M satisfies the relation M≤Y-N, When A is equal to 2, the minimum value of N is 2, when A is greater than 2, the minimum value of N is 1, and The maximum value of N determined from A and M is such that a set of Tn and Tm values becomes a different set of values for all Tin values. 9. The gamma correction circuit according to claim 8, further comprising a predetermined element for encoding said Tn as an element of a sequence consisting of N integers equal to or greater than 0 incremented by unequal difference from the initial term value device. 10. The gamma correction circuit according to claim 8, wherein A=2, M=Y-Z', and N is set to a minimum integer value equal to or greater than 1+logZ'/log2, when used for input The inverse gamma correction table of the Tin value and outputting the Tout value is predetermined, under the condition that the Tout value after being reduced by Z bits less significant bits is set to Tz, obtained with respect to all Tout values Of the Tz values, only those Tz values whose most significant bit value is 1 are extracted, and the maximum value among Z values of different values of all extracted Tz values is set to Z′. 11. The gamma correction circuit according to claim 8, wherein the tone output is input to at least any one of an adder circuit, a multiplier circuit and a look-up table circuit. 12. The gamma correction circuit according to claim 11, wherein the hue output is input to a circuit for performing color space conversion processing according to chromaticity characteristics of the display device. 13. The gamma correction circuit according to claim 11, wherein the tone output is input to a circuit for performing luminance-contrast conversion processing according to luminance-contrast characteristics of a display device. 14. A gamma correction circuit comprising: a format conversion circuit for inputting (N+M)-bit signals including N-bit signal Tn and M-bit signal Tm as an input signal, and outputting a constant A, an N-bit index indicated by said Tn, and an index indicated by said Tm The M-bit mantissa is represented as a Y-bit tone signal Tout with a linear tone-brightness characteristic of Tout= ^ATn ×Tm; and A gamma correction circuit for inputting a Tout value and outputting an X-bit (X<Y) tone signal Tin having a tone-brightness characteristic that is nonlinearly converted, Wherein the value of the constant A is a natural number power of 2, The M satisfies the relationship M≤Y-N, When A is equal to 2, the minimum value of N is 2, and when A is greater than 2, the minimum value of N is 1, and The maximum value of N determined from said constant A and said M is such that a set of Tn and Tm values becomes a different set of values for all Tin values. 15. The gamma correction circuit of claim 8, wherein the inverse gamma correction circuit and the format conversion circuit are implemented using only one look-up table. 16. An image processing apparatus comprising the gamma correction circuit according to claim 8. 17. An image processing device for outputting an analog voltage to determine the color tone of a display device, comprising: a digital-to-analog converter for performing digital-to-analog conversion on an M-bit input signal among input signals including an N-bit signal and an M-bit signal; and a reference voltage generator for outputting a reference voltage for digital-to-analog conversion in said digital-to-analog converter, Wherein, the exponent Tn indicated by the N-bit signal and the mantissa Tm indicated by the M-bit signal are provided, so that a constant A is used to express the Y-bit tone signal Tout with linear tone-brightness characteristics as Tout= ^ATn ×Tm The value of A is a natural power of 2, and The reference voltage generator generates the reference voltage according to the N-bit signal. 18. A display device comprising: display device; and The image processing apparatus according to claim 16. 19. The display apparatus according to claim 18, wherein the display device is a liquid crystal display.

Images