JP2018142947A - Signal processing circuit and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To maintain the grayscale of high-saturation RGB signals when performing a linear matrix color correction process and a knee or limit process.SOLUTION: A linear matrix circuit 10-1 multiplies RGB signal values by a preset matrix LM1 and thereby performs a first instance of linear matrix color correction process and generates RGB signals (Rmid, Gmid, Bmid signals). A dynamic range compression circuit 102 performs a dynamic range compression process such as a knee process or a grip process on the RGB signals (Rmid, Gmid, Bmid signals) and generates RGB signals (Rmid', Gmid', Bmid' signals). A linear matrix circuit 11-1 multiplies the RGB signals (Rmid', Gmid', Bmid' signals) by a preset matrix LM2 and thereby performs a second instance of linear matrix color correction process and generates RGB signals (Rfin, Gfin, Bfin signals).SELECTED DRAWING: Figure 1

Description

The present invention relates to a circuit and program for processing RGB signals.

  2. Description of the Related Art Conventionally, in an imaging apparatus, color correction processing using a linear matrix is performed in order to obtain color reproduction that matches a target color system. In this color correction process using the linear matrix, the value of the RGB signal is approximated to the value of the target signal by multiplying the value of the RGB signal obtained from the image sensor by a matrix of 3 rows and 3 columns. This is a process of correcting to.

  By the color correction processing of the linear matrix, it is possible to correct errors caused by the image sensor and the optical system, and to reproduce negative imaging characteristics required in a general color system. For this reason, this process is very widely used. FIG. 9 is a diagram showing an example of negative imaging characteristics, and shows ideal imaging characteristics of a wide color gamut color system defined for ultra-high definition video defined in ITU-R rec.BT2020. The horizontal axis represents wavelength (nm) and the vertical axis represents sensitivity (arbitrary unit). By the linear matrix color correction processing, as shown in FIG. 9, negative imaging characteristics can be reproduced for each of the RGB signals.

  On the other hand, in an imaging device, nonlinear signal processing (dynamic range compression processing) such as knee and clipping is often performed in order to keep the RGB signal values obtained from the imaging device within a transmittable or displayable range. .

  There is known an imaging apparatus that performs signal processing on RGB signals obtained from an imaging element, including such linear matrix color correction processing and dynamic range compression processing (see, for example, Patent Documents 1 to 3).

  FIG. 7 is a block diagram illustrating a configuration example of a conventional signal processing circuit. The signal processing circuit 100 includes a linear matrix circuit 101 and a dynamic range compression circuit 102. In a conventional process for RGB signals obtained from an image sensor, a dynamic range compression process is generally performed after a linear matrix color correction process.

  FIG. 8 is a diagram for explaining a processing example of the conventional signal processing circuit 100. 8A to 8C, the horizontal axis represents the incident light intensity, and the vertical axis represents the signal value of the RGB signal. It is assumed that an RGB signal of subject light in which blue is dominant is output from the image sensor, and the RGB signal is input to the signal processing circuit 100.

  8A shows the characteristics of the RGB signal input to the linear matrix circuit 101, and FIG. 8B shows the characteristics of the RGB signal output from the linear matrix circuit 101 and input to the dynamic range compression circuit 102. Indicates. FIG. 8C shows the characteristics of the RGB signal output from the dynamic range compression circuit 102. The characteristics shown in FIGS. 8A to 8C are shown for convenience so that the degree of saturation and the degree of gradation can be intuitively understood in order to explain the processing example of the signal processing circuit 100. It is an outline.

  After RGB signals obtained from an image sensor not shown in FIG. 7 are subjected to processing such as gain adjustment and noise removal, the RGB signals are input to the linear matrix circuit 101. In general, the RGB signal obtained from the image sensor cannot achieve negative characteristics or is affected by crosstalk, so the saturation is lower than the ideal signal value and the signal becomes dull. End up. Therefore, color correction processing is performed by the linear matrix circuit 101. That is, processing for adjusting the balance of the colors of the RGB signals and processing for increasing the saturation are performed.

  The linear matrix circuit 101 performs linear matrix color correction processing (corrects color reproduction) by multiplying RGB signal values by a preset matrix of 3 rows and 3 columns LMorg. The matrix coefficient, which is each element of the 3 × 3 matrix LMorg, can take any value in order to achieve the desired color reproduction.

  For example, it is assumed that the RGB signal shown in FIG. 8A is input to the linear matrix circuit 101. By performing color correction processing of the linear matrix on the RGB signal shown in FIG. 8A, the signal value of the B signal is increased, and the signal values of the R and G signals are reduced, as shown in FIG. 8B. An RGB signal is generated. This increases the saturation of the RGB signal.

  The dynamic range compression circuit 102 performs dynamic range compression processing such as knee processing and clip processing on the values of the RGB signals that have undergone linear matrix color correction processing by the linear matrix circuit 101. The knee process is a process of reducing the value of the RGB signal exceeding the threshold (knee point) according to a certain ratio (knee slope). The clipping process is a process of limiting the RGB signal value exceeding the threshold (clip point) to the threshold. Here, processing for changing the gradation expression from linear is generally defined as dynamic compression processing.

  For example, by performing dynamic compression processing (clip processing) on the RGB signal shown in FIG. 8B, the signal value of the B signal is limited to the upper limit value in the region subjected to dynamic range compression processing, and FIG. RGB signals shown in FIG. The RGB signal shown in FIG. 8C is blue in the entire range of incident light intensity.

Japanese Patent Publication No.59-41357 JP-A-11-191893 Japanese Patent No. 4004849

  As described above, the linear matrix circuit 101 of the signal processing circuit 100 performs the linear matrix color correction processing on the RGB signal of the high-saturation subject, so that the high-saturation signal near the boundary of the color gamut in the target color system is obtained. May be obtained. In addition, a signal out of the color gamut may be obtained.

  When dynamic range compression processing is performed on the high saturation signal or the out-of-gamut signal by the dynamic range compression circuit 102, the dynamic range of the dominant signal value of the RGB signals is compressed. For this reason, as shown in FIG. 8C, in the region where the signal value of the B signal becomes constant by the dynamic range compression processing, the gradation of the RGB signal is impaired, so-called gradation collapse occurs.

  Here, it is assumed that light having a color equal to the primary primary color (primary color) of the color system is incident on the image sensor from the optical system. The gradation of the sensor output signal is 12 bits (0 to 4095).

  When correct color reproduction is obtained by the color correction processing of the linear matrix, the RGB signal after the color correction of the linear matrix is accompanied by (R, G, B) = (0, 0, 0: Black) to (0, 0, 4095: Blue) to (4095, 4095, 4095: White).

  Then, clipping processing is performed as dynamic range compression processing on the RGB signals after color correction of the linear matrix. Then, the RGB signal after the dynamic range compression process changes from (0, 0, 0: black) to (0, 0, 1023: blue) to (1023, 1023, 1023: white). Here, for example, when the signal value before dynamic range compression is (0, 0, 1023) to (0, 0, 4095), all the values after compression change to (0, 0, 1023).

  For this reason, the gradation of the B signal is impaired, and the video is uncomfortable. Even if the value of the R (red) signal or the G (green) signal is not 0, the same phenomenon occurs when R, G << B. The same applies to the case where knee processing is performed instead of clip processing as dynamic range compression processing. Here, the case where light of the same color as the primary color of blue of the target color system is incident is illustrated, but this phenomenon is not limited to blue but also when light of the same color as the primary color of red or green is incident. Occur as well.

  In recent years, due to advances in paint or dyeing technology and advances in lighting technology such as LEDs, there have been many opportunities to photograph high-saturation subjects, and such a phenomenon has produced an uncomfortable image.

  Therefore, the present invention has been made to solve the above-described problems, and its purpose is to maintain the gradation of a highly saturated RGB signal when performing linear matrix color correction processing and knee or limit processing. It is an object of the present invention to provide a signal processing circuit and a program that can be used.

  In order to solve the above problem, the signal processing circuit according to claim 1 is a signal processing circuit that performs predetermined processing on the RGB signal, and sets a first matrix of 3 rows and 3 columns set in advance for the RGB signal. By multiplying, a first linear matrix circuit that performs color correction processing of a linear matrix and a dynamic range that performs dynamic range compression processing on the RGB signals that have been subjected to the color correction processing by the first linear matrix circuit. A linear matrix color correction process is performed by multiplying an RGB signal that has been subjected to the dynamic range compression process by the compression circuit and the dynamic range compression circuit by a preset second matrix of 3 rows and 3 columns. And a second linear matrix circuit that performs the first linear matrix circuit using the first matrix. From the RGB signal when the color correction processing is performed using the 3 × 3 matrix which is the product of the first matrix and the second matrix, the RGB signal subjected to the color correction processing Is also characterized by a signal with reduced saturation.

  According to a second aspect of the present invention, there is provided the signal processing circuit according to the first aspect, wherein the first linear matrix circuit multiplies the first matrix with the RGB signal by multiplying the first matrix. By performing correction processing, the saturation of the RGB signal is reduced, and the second linear matrix circuit applies the second matrix to the RGB signal subjected to the dynamic range compression processing by the dynamic range compression circuit. And performing color correction processing of the linear matrix to increase the saturation of the RGB signal and adjust the color balance of the RGB signal.

  According to a third aspect of the present invention, there is provided the signal processing circuit according to the first aspect, wherein the first linear matrix circuit multiplies the first matrix with the RGB signal by multiplying the first matrix. By performing correction processing, the color balance of the RGB signal is adjusted, and the second linear matrix circuit performs the second matrix on the RGB signal on which the dynamic range compression processing has been performed by the dynamic range compression circuit. And the color correction processing of the linear matrix is performed to increase the saturation of the RGB signal.

According to a fourth aspect of the present invention, there is provided the signal processing circuit according to the first aspect, wherein a matrix of 3 rows and 3 columns, which is a product of the first matrix and the second matrix, is set to LMorg and α in advance. As the set positive real parameter, the first matrix LM1 and the second matrix LM2 are set as follows:
It is characterized by that.

The signal processing circuit according to claim 5 is a signal processing circuit according to claim 1, wherein a matrix of 3 rows and 3 columns, which is a product of the first matrix and the second matrix, is defined as LMorg, α _R , Using α _G and α _B as preset positive real parameters corresponding to the RGB signals, the first matrix LM1 and the second matrix LM2,
It is characterized by that.

A signal processing circuit according to a sixth aspect is the signal processing circuit according to the first aspect, wherein a matrix of 3 rows and 3 columns, which is a product of the first matrix and the second matrix, is preliminarily set as LMorg and β. As the set positive real parameter, the second matrix LM2 and the first matrix LM1 are set as follows:
It is characterized by that.

The signal processing circuit according to claim 7 is the signal processing circuit according to claim 1, wherein a matrix of 3 rows and 3 columns, which is a product of the first matrix and the second matrix, is defined as LMorg, β _R , Using β _G and β _B as preset positive real parameters corresponding to the RGB signals, the second matrix LM2 and the first matrix LM1
It is characterized by that.

  The signal processing circuit according to claim 8 is the signal processing circuit according to any one of claims 1 to 7, wherein the second matrix is an inverse matrix of the first matrix. Features.

  A signal processing circuit according to a ninth aspect is the signal processing circuit according to any one of the first to eighth aspects, wherein the first linear matrix circuit applies the RGB signal obtained by the image sensor. The color correction processing of the linear matrix is performed by multiplying the first matrix.

  A signal processing circuit according to a tenth aspect is the signal processing circuit according to any one of the first to ninth aspects, further comprising the second matrix and the first matrix for the RGB signal. Are multiplied by a preset third matrix of 3 rows and 3 columns, and the color correction is performed by the third linear matrix circuit that performs color correction processing of the linear matrix and the third linear matrix circuit. For each processed RGB signal, a signal that is smaller than a preset reference value is clipped to the reference value, and a clip circuit that generates an RGB signal after clipping, and the RGB generated by the clip circuit A fourth linear matrix that performs color correction processing of the linear matrix by multiplying the signal by a preset fourth matrix that is an inverse matrix of the third matrix. And the first linear matrix circuit multiplies the first matrix with the RGB signal that has been subjected to the color correction processing by the fourth linear matrix circuit. Color correction processing is performed.

  The signal processing circuit according to claim 11 is the signal processing circuit according to claim 10, comprising a fifth linear matrix circuit instead of the fourth linear matrix circuit and the first linear matrix circuit, The fifth linear matrix circuit is a preset fifth matrix of 3 rows and 3 columns that is the product of the first matrix and the fourth matrix for the RGB signal generated by the clip circuit. Are multiplied by a linear matrix color correction process, and the dynamic range compression circuit performs a dynamic range compression process on the RGB signal subjected to the color correction process by the fifth linear matrix circuit. It is characterized by that.

  Furthermore, a program according to a twelfth aspect causes a computer to function as the signal processing circuit according to any one of the first to eleventh aspects.

  As described above, according to the present invention, it is possible to maintain the gradation of a highly saturated RGB signal when performing linear matrix color correction processing and knee or limit processing.

FIG. 3 is a block diagram illustrating a configuration example of a signal processing circuit according to the first embodiment. FIG. 6 is a block diagram illustrating a configuration example of a signal processing circuit according to a second embodiment. FIG. 10 is a block diagram illustrating a configuration example of a signal processing circuit according to a third embodiment. FIG. 10 is a block diagram illustrating a configuration example of an imaging apparatus according to a fourth embodiment. It is a figure explaining the process example of the signal processing circuit using matrix LM1, LM2 of the 1st method. It is a figure explaining the experimental result of Example 1 when a high saturation RGB signal is input. It is a block diagram which shows the structural example of the conventional signal processing circuit. It is a figure explaining the process example of the conventional signal processing circuit. It is a figure which shows the example of a negative imaging characteristic. FIG. 10 is a block diagram illustrating a configuration example of a signal processing circuit according to a fifth embodiment. FIG. 10 is a block diagram illustrating a configuration example of a signal processing circuit according to a sixth embodiment. It is a figure explaining transition of a signal value when there is no signal processing outside a color gamut. It is a figure explaining transition of a signal value in case there is signal processing outside a color gamut. It is a figure explaining the experimental result of Example 6 when a RGB signal of high saturation is input.

Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the drawings.
[Example 1]
First, the first embodiment (Example 1) of the present invention will be described. FIG. 1 is a block diagram illustrating a configuration example of a signal processing circuit according to the first embodiment. The signal processing circuit 1 includes linear matrix circuits 10-1 and 11-1 and a dynamic range compression circuit 102.

  For example, when the signal processing circuit 1 is used in an image pickup apparatus, a signal obtained from an image pickup device (not shown in FIG. 1) is subjected to signal processing after processing such as gain adjustment and noise removal that are appropriately required. Input to the circuit 1. As in the case of the prior art shown in FIG. 7, the RGB signal obtained from the image sensor is affected by different adjacent elements, so that the saturation is lowered and dull compared to the actual subject light. It becomes a color signal. Therefore, the signal processing circuit 1 corrects the color to an optimum signal value in the target color system, and reduces the gradation loss of high saturation light caused by the dynamic range compression processing.

  The signal processing circuit 1 is applicable to a video signal processing apparatus that performs conversion processing or the like on a video signal in addition to an imaging apparatus. As an example of the conversion process, there is a process of converting an HDR (High Dynamic Range) signal into an SDR (Standard Dynamic Range) signal. The same applies to the signal processing circuit 2 of the second embodiment and the signal processing circuit 3 of the third embodiment which will be described later.

  The linear matrix circuit 10-1 receives the RGB signal and performs a first linear matrix color correction process by multiplying the RGB signal value by a preset 3 × 3 matrix LM 1. As a result, an RGB signal with a lower saturation than the RGB signal output by the conventional linear matrix circuit 101 shown in FIG. 7 is generated.

  The linear matrix circuit 10-1 outputs the RGB signals (Rmid, Gmid, Bmid signals) after the first linear matrix color correction processing to the dynamic range compression circuit 102. Details of the preset matrix LM1 will be described later.

  The dynamic range compression circuit 102 inputs RGB signals (Rmid, Gmid, Bmid signals) that have undergone the first linear matrix color correction processing from the linear matrix circuit 10-1. Then, similarly to the dynamic range compression circuit 102 shown in FIG. 7, the dynamic range compression circuit 102 performs dynamic range compression processing such as knee processing and clip processing on the values of the RGB signals (Rmid, Gmid, and Bmid signals). Do. As a result, the dynamic range compression processing is performed on the RGB signals (Rmid, Gmid, Bmid signals) whose saturation is lower than that of the prior art, so that the RGB signals (Rmid ′, Gmid ′, Bmid signal ′) is generated.

  The dynamic range compression circuit 102 outputs the RGB signals (Rmid ′, Gmid ′, Bmid ′ signals) after the dynamic range compression processing to the linear matrix circuit 11-1.

  Note that the dynamic range compression circuit 102 may perform only knee processing or only clip processing as dynamic range compression processing. The dynamic range compression circuit 102 may perform dynamic range compression processing other than knee processing and clip processing. The dynamic range compression circuit 102 may perform a plurality of processes such as a knee process and a clip process.

  The linear matrix circuit 11-1 inputs RGB signals (Rmid ', Gmid', Bmid 'signals) subjected to dynamic range compression processing from the dynamic range compression circuit 102. The linear matrix circuit 11-1 then multiplies the value of the RGB signal (Rmid ′, Gmid ′, Bmid ′ signal) by a preset 3 × 3 matrix LM2 to obtain the second linear matrix. The color correction process is performed. The desired color reproduction is realized by the second linear matrix color correction processing. As a result, the same color correction processing as that of the prior art is performed, and RGB signals (Rfin, Gfin, Bfin signals) in which gradation collapse is suppressed as compared with the prior art are generated.

  The linear matrix circuit 11-1 outputs RGB signals (Rfin, Gfin, Bfin signals) after color correction processing of the second linear matrix. Details of the preset matrix LM2 will be described later.

(Matrix LM1, LM2)
Next, the preset matrices LM1 and LM2 will be described in detail. The matrices LM1 and LM2 are obtained by a designer's calculation using a predetermined method and set in advance. There are two types of setting methods. The first is a method in which the matrix LM1 is calculated first and then the matrix LM2 is calculated, and the second is a method in which the matrix LM2 is calculated first and then the matrix LM1 is calculated.

  In either method, the linear matrix coefficients that are the elements of the matrices LM1 and LM2 are obtained by combining the color correction processing of the first and second linear matrices by the linear matrix circuits 10-1 and 11-1 to obtain the target color. It is determined that correction processing is performed. That is, the linear matrix coefficients of the matrices LM1 and LM2 are subjected to color correction processing (target color correction processing for the entire signal processing circuit 1) expressed by LM2 · LM1 = LMorg which is the product of the matrices LM1 and LM2. To be determined. In other words, the color correction process performed using the matrix LM1 by the linear matrix circuit 10-1 and the color correction process performed using the matrix LM2 by the linear matrix circuit 11-1, and LM2 · The linear matrix coefficient is determined on the assumption that the color correction processing performed using the matrix of LM1 = LMorg is equivalent. The matrix LMorg is a matrix used by the linear matrix circuit 101 shown in FIG.

(First method)
As described above, the first method for setting the matrices LM1 and LM2 is to calculate the matrix LM1 first, and then calculate the matrix LM2. First, the linear matrix coefficient of the matrix LM1 is determined by defining as follows using α which is a positive real number as a parameter.

  As the parameter α, an RGB signal (Rmid, Gmid, Bmid signal) whose saturation is lower than that of the input signal of the conventional dynamic range compression circuit 102 shown in FIG. 7 is input to the dynamic range compression circuit 102 of the first embodiment. As described above, this is a parameter for determining the degree of desaturation based on the RGB signal (output signal of the image sensor) input to the linear matrix circuit 10-1. The greater the value of the parameter α, the greater the degree of saturation of the RGB signal (Rmid, Gmid, Bmid signal), and the smaller the value of the parameter α, the smaller the degree of saturation.

  Compared with the matrix LMorg of the color correction processing by the conventional linear matrix circuit 101 shown in FIG. 7, this matrix LM1 is not intended to mainly correct the color balance, and is a signal whose saturation is reduced with respect to the input signal. Functions to generate.

However, the color correction processing of the linear matrix acts on the RGB signal that is the input signal as shown in the following formula, and the RGB signal (Rout, Gout, Bout) that is the output signal, assuming that the matrix for performing the processing is LM. ) Shall be obtained.

Next, the linear matrix coefficients of the matrix LM2 are determined by defining as follows in order to make the color correction of the entire signal processing circuit 1 an optimum value.
LM1 ^-1 represents an inverse matrix of LM1. The matrix LMorg is used for the color correction processing of the conventional linear matrix circuit 101 shown in FIG. 7, and is a matrix for performing the target color correction processing as a whole, and is the product of the matrices LM1 and LM2.

  The matrix LM2 performs color balance processing in the same manner as the matrix LMorg for color correction processing by the conventional linear matrix circuit 101 shown in FIG. 7, but additionally functions to compensate for the decrease in saturation due to the matrix LM1. To do.

  Next, processing when the matrices LM1 and LM2 set by the first method are used will be described. FIG. 5 is a diagram for explaining a processing example of the signal processing circuit 1 using the matrices LM1 and LM2 of the first technique. 5A to 5D, the horizontal axis indicates the incident light intensity, and the vertical axis indicates the signal value of the RGB signal. FIG. 5A shows the characteristics of the RGB signals input to the linear matrix circuit 10-1, and FIG. 5B is output from the linear matrix circuit 10-1 and input to the dynamic range compression circuit 102. The characteristics of the RGB signal are shown. 5C shows the characteristics of the RGB signal output from the dynamic range compression circuit 102 and input to the linear matrix circuit 11-1, and FIG. 5D is output from the linear matrix circuit 11-1. The characteristic of the RGB signal to be performed is shown. 5A to 5D are schematic diagrams that are shown for the sake of convenience so that the degree of saturation and the degree of gradation can be intuitively understood in order to explain the processing example of the signal processing circuit 1. is there.

  The linear matrix circuit 10-1 performs the first linear matrix color correction processing on the RGB signal values using the matrix LM1 set by the first method. Then, the linear matrix circuit 10-1 has RGB signals (Rmid, Gmid, Bmid signals) with reduced saturation as compared with the signal subjected to the color correction processing by the conventional linear matrix circuit 101 shown in FIG. Is generated.

  The first linear matrix color correction processing is performed on the RGB signal shown in FIG. 5A using the matrix LM1. As a result, the signal value of the B signal is reduced, and the signal values of the R and G signals are increased, so that the RGB signals (Rmid, Gmid, which are less saturated than the RGB signals, as shown in FIG. 5B). , Bmid signal) is generated.

  The dynamic range compression circuit 102 performs dynamic range compression processing on RGB signals (Rmid, Gmid, and Bmid signals) whose saturation is lower than that in the prior art.

  Dynamic range compression processing is performed on the RGB signals (Rmid, Gmid, and Bmid signals) shown in FIG. As a result, in the region of the incident light intensity that is not the target of the dynamic range compression processing, there is no gradation collapse and the saturation is low as in the RGB signal (Rmid, Gmid, Bmid signal). RGB signals (Rmid ′, Gmid ′, Bmid ′ signals) shown in FIG. Further, in the region of the incident light intensity to be subjected to the dynamic range compression processing, the signal value of the B signal is limited to the upper limit value, and the saturation is further lowered than the RGB signal (Rmid, Gmid, Bmid signal). The RGB signals (Rmid ′, Gmid ′, Bmid ′ signals) shown in c) are generated.

  The linear matrix circuit 11-1 uses the matrix LM2 set by the first method for RGB signals (Rmid ′, Gmid ′, Bmid ′ signals) with low saturation to perform color correction of the second linear matrix. Processing (in this case, color balance processing) is performed. Then, the linear matrix circuit 11-1 outputs RGB signals (Rfin, Gfin, Bfin signals) with no gradation collapse and high saturation.

  For the RGB signals (Rmid ′, Gmid ′, Bmid ′ signals) shown in FIG. 5C, a second linear matrix color correction process is performed using the matrix LM2. This increases the signal values of the R and G signals in the region of the incident light intensity that is the target of the dynamic range compression process. That is, as shown in FIG. 5D, RGB signals (Rfin, Gfin, and Bfin signals) with no gradation collapse and high saturation are generated. This RGB signal (Rfin, Gfin, Bfin signal) has a natural gradation that changes from a low incident light intensity region to a high region, from blue to white.

  As described above, by using the matrices LM1 and LM2 set by the first method, the same color correction processing as that of the conventional method is performed, and gradation collapse is suppressed as compared with the conventional method. Can do.

The matrix LM1 may use the following mathematical formula instead of the mathematical formula (1). The linear matrix coefficients of the matrix LM1 are determined using α _R , α _G , and α _B that are positive real numbers for each RGB signal as parameters.
The parameters α _R , α _G , and α _B are parameters that determine the degree of desaturation for each color of the RGB signal. When α _R = α _G = α _B = α, the matrix LM1 is expressed by the above equation (1).

(Second method)
As described above, in the second method for setting the matrices LM1 and LM2, the matrix LM2 is calculated first, and then the matrix LM1 is calculated. First, the linear matrix coefficient of the matrix LM2 is determined by defining the following as a parameter, which is a positive real number β.

  As the parameter β, RGB signals (Rmid, Gmid, and Bmid signals) whose saturation is lower than that of the input signal of the conventional dynamic range compression circuit 102 shown in FIG. 7 are input to the dynamic range compression circuit 102 of the first embodiment. As described above, it is a parameter that determines the degree of desaturation based on the RGB signals (Rfin, Gfin, Bfin signals) output from the linear matrix circuit 11-1. The greater the value of the parameter β, the greater the degree of saturation of the RGB signal (Rmid, Gmid, Bmid signal), and the smaller the value of the parameter β, the smaller the degree of saturation.

  The matrix LM2 functions to generate a signal with increased saturation without performing color balance processing on the input signal.

Next, the linear matrix coefficients of the matrix LM1 are determined by defining as follows in order to set the overall color correction to an optimum value.
LM2 ⁻¹ represents an inverse matrix of LM2.

  The matrix LM1 performs color balance processing similarly to the matrix LMorg of the color correction processing by the conventional linear matrix circuit 101 shown in FIG. 7, but generates a signal with lower saturation than the matrix LMorg. To work.

  When the matrices LM1 and LM2 set by the second method are used, the same color correction as that of the conventional method is performed and the gradation is crushed compared to the conventional method as in the case of the first method. Can be suppressed.

The matrix LM2 may use the following formula instead of the formula (5). The linear matrix coefficients of the matrix LM2 are determined using β _R , β _G , and β _B that are positive real numbers for each RGB signal as parameters.
The parameters β _R , β _G , β _B are parameters that determine the degree of desaturation for each color of the RGB signal. When β _R = β _G = β _B = β, the matrix LM2 is expressed by the above equation (5).

  The difference between the first method and the second method is a difference in RGB signals (Rmid, Gmid, Bmid signals) that are input signals of the dynamic range compression circuit 102.

  In the first method, an RGB signal (Rmid, Gmid, Bmid signal) that is an input signal of the dynamic range compression circuit 102 is converted into an RGB signal (for example, an RGB signal obtained from an image sensor) of the signal processing circuit 1. ) As a reference, it can be a signal whose saturation is lower than that of the input signal of the conventional dynamic range compression circuit 102 shown in FIG. On the other hand, in the second method, RGB signals (Rmid, Gmid, Bmid signals) that are input signals of the dynamic range compression circuit 102 are converted into RGB signals (Rfin, Gfin, Bfin) that are output signals of the signal processing circuit 1. Signal) as a reference, it can be a signal whose saturation is lower than that of the input signal of the conventional dynamic range compression circuit 102 shown in FIG.

  Therefore, for example, a more preferable result can be obtained by appropriately selecting and using the first method or the second method according to the characteristics of the imaging element.

  In addition, when the incident light is R, G, and B, respectively, and there is no significant difference in the saturation of the RGB signal obtained from the image sensor, the equation (1) using the parameter α common to the RGB signal is used. Matrix LM1 or the matrix LM2 of the equation (5) with the parameter β common to the RGB signals is used.

On the other hand, when there is a difference in saturation, the matrix LM1 of the equation (4) using the parameters α _R , α _G , α _{B for} each RGB signal, or the parameters β _R , β _G , β _B for each RGB signal. The matrix LM2 of Equation (7) is used.

That is, the matrix LM1 of the equation (4) or the matrix LM2 of the equation (7) is used when there is a difference in saturation, and gradation collapse can be effectively suppressed. For example, when the saturation of the RGB signal when shooting a blue subject is larger than the saturation of the RGB signal when shooting a red or green subject, the parameter α _B or β _{B of the} B signal is another parameter. The matrix LM1 or the matrix LM2 set to a value larger than α _R , α _G or β _R , β _G is used.

(Experimental result of Example 1)
Next, the experimental result by the computer simulation of Example 1 will be described. FIG. 6 is a diagram for explaining experimental results of Example 1 when a high-saturation RGB signal is input. (1) shows changes in the signal value by the signal processing circuit 100 of the prior art shown in FIG. 7, and (2) shows changes in the video signal in (1). Further, (3) shows the change of the signal value by the signal processing circuit 1 of the first embodiment shown in FIG. 1, and (4) shows the change of the video signal in (3). However, the signal values shown in (1) and (3) are values after gamma correction, and the change in the signal value is represented as a curve.

  6 (1) and 6 (3), the horizontal axis indicates the incident light intensity, and the vertical axis indicates the signal value of the RGB signal. This RGB signal is an output signal of the signal processing circuit 1 shown in FIG. 1, that is, an RGB signal (Rfin, Gfin, Bfin signal). 6 (2) and 6 (4), the horizontal axis indicates the same incident light intensity as in FIGS. 6 (1) and 6 (3). 6 (2) and 6 (4) show gradations in the entire range of incident light intensity.

  6 (1) to 6 (4) show the matrix LMorg of the linear matrix circuit 101 used in the signal processing circuit 100 of the prior art and the linear matrix circuits 10-1 and 11- used in the signal processing circuit 1 of the first embodiment. It is an experimental result when the product of 1 matrix LM1, LM2 is made the same. That is, LMorg = LM2 · LM1 as shown in the mathematical expressions (3) and (6).

  Assume that an RGB signal of subject light in which blue is dominant is output from the image sensor, and the RGB signal is input to the signal processing circuit 100 and the signal processing circuit 1 of the first embodiment.

  From FIG. 6 (1), in the conventional technique, the signal value of the B signal is suppressed by the dynamic range compression process in the range where the incident light intensity is about 80 or more and 600 or less, but the signal value of the R and G signals is It can be seen that the signal value does not increase and is stagnated as a whole. For this reason, as shown in FIG. 6B, the gradation of the video signal is crushed.

  On the other hand, from FIG. 6 (3), in Example 1, the signal value of the B signal is suppressed by the dynamic range compression processing in the range where the incident light intensity is about 80 or more and 600 or less. It can be seen that the value increases, and the closer to the white, the higher the incident light intensity. For this reason, as shown in FIG. 6 (4), the video signal maintains a natural gradation without being crushed.

  As described above, according to the signal processing circuit 1 of the first embodiment, the linear matrix circuit 10-1 multiplies the value of the RGB signal by the preset matrix LM1 to obtain the first linear matrix. Color correction processing is performed to generate RGB signals (Rmid, Gmid, Bmid signals). The dynamic range compression circuit 102 performs dynamic range compression processing such as knee processing and clip processing on the values of the RGB signals (Rmid, Gmid, and Bmid signals) to generate RGB signals (Rmid ′, Gmid ′, and Bmid ′ signals). Generate. The linear matrix circuit 11-1 performs color correction processing of the second linear matrix by multiplying the value of the RGB signal (Rmid ′, Gmid ′, Bmid ′ signal) by a preset matrix LM2, RGB signals (Rfin, Gfin, Bfin signals) are generated.

  As a result, the RGB signals (Rfin, Gfin, and Bfin signals) are subjected to the same color correction processing as that of the prior art, and the gradation collapse is suppressed as compared with the prior art. The RGB signals (Rfin, Gfin, and Bfin signals) are subjected to the same color correction processing as that of the prior art, and the linear matrix circuits 10-1 and 11- so that LMorg = LM2 · LM1 is satisfied. This is because the process 1 is performed. In addition, the RGB signals (Rfin, Gfin, Bfin signals) are signals with suppressed gradation loss compared to the conventional RGB signals (Rmid, Gmid, Bmid signals) with lower saturation than in the past. This is because the dynamic range compression processing is performed, and the RGB signals (Rmid ′, Gmid ′, Bmid ′ signals) after the dynamic range compression processing become signals having gradations as compared with the conventional case.

  Therefore, the input RGB signal can be color-corrected to the optimum signal value in the target color system, and the gradation loss of high saturation light caused by the dynamic range compression process can be reduced, and the dynamic range can be reduced. Even after compression processing, natural gradation can be maintained. That is, when performing linear matrix color correction processing and dynamic range compression processing, the gradation of RGB signals with high saturation can be maintained.

[Example 2]
Next, a second embodiment (Example 2) of the present invention will be described. FIG. 2 is a block diagram illustrating a configuration example of the signal processing circuit according to the second embodiment. The signal processing circuit 2 includes linear matrix circuits 10-2 and 11-2 and a dynamic range compression circuit 102. As a whole, the signal processing circuit 2 does not perform linear matrix color correction processing but performs only dynamic range compression processing.

When the signal processing circuit 1 of the first embodiment shown in FIG. 1 and the signal processing circuit 2 of the second embodiment shown in FIG. 2 are compared, the dynamic range compression circuit 102 is provided under the same configuration as a whole. Common. On the other hand, the signal processing circuit 2 according to the second embodiment includes linear matrix circuits 10-2 and 11-2 that are different from the linear matrix circuits 10-1 and 11-1 of the signal processing circuit 1 according to the first embodiment. Is different. Specifically, in the first embodiment, the matrix LM2 is used in the linear matrix circuit 11-1, whereas in the second embodiment, the matrix LM1 ^-1 (linear matrix circuit 10- 2) is used.

  The linear matrix circuit 10-2 performs the first color correction process of the linear matrix by multiplying the value of the RGB signal by a preset matrix LM1 of 3 rows and 3 columns.

  The dynamic range compression circuit 102 performs the same processing as the dynamic range compression circuit 102 shown in FIG.

The linear matrix circuit 11-2 multiplies the value of the RGB signal (Rmid ′, Gmid ′, Bmid ′ signal) by a preset 3 × 3 matrix LM1 ⁻¹ to obtain the second linear matrix. The color correction process is performed.

  The linear matrix circuit 11-2 outputs an RGB signal (Rfin, Gfin, Bfin signal) that has been subjected to the same color correction processing as that of the prior art and that has suppressed gradation collapse compared to the prior art. The RGB signal (Rfin, Gfin, Bfin signal) that is the output signal of the signal processing circuit 2 is different from the RGB signal that is the input signal only in that it is a signal that has undergone dynamic range compression processing. is there. That is, depending on the processing of the signal processing circuit 2, the characteristics of the linear matrix circuit 10-2 and the characteristics of the linear matrix circuit 11-2 cancel each other within a range where the signal value does not change due to the dynamic range compression processing. The signal color does not change.

The matrices LM1 and LM2 (= LM1 ⁻¹ ) are obtained by a designer's calculation using a predetermined method and set in advance. As this setting method, the same first method as that in the first embodiment is used, and a condition (LM2 · LM1 = I) in which LM2 is an inverse matrix of LM1 is further added. I is a unit matrix of 3 rows and 3 columns.

As described above, according to the signal processing circuit 2 of the second embodiment, the linear matrix circuit 10-2 is preset with respect to the values of the RGB signals, like the linear matrix circuit 10-1 of the first embodiment. By multiplying the matrix LM1, the first linear matrix color correction processing is performed to generate RGB signals (Rmid, Gmid, Bmid signals). The dynamic range compression circuit 102 performs dynamic range compression processing such as knee processing and clip processing on the values of the RGB signals (Rmid, Gmid, and Bmid signals) to generate RGB signals (Rmid ′, Gmid ′, and Bmid ′ signals). Generate. The linear matrix circuit 11-2 multiplies the value of the RGB signal (Rmid ′, Gmid ′, Bmid ′ signal) by a preset matrix LM1 ⁻¹ to perform the second linear matrix color correction process. Then, RGB signals (Rfin, Gfin, Bfin signals) are generated.

  As a result, the RGB signals (Rfin, Gfin, and Bfin signals) have the same color as the input RGB signal in the signal range that is not affected by the dynamic range compression, and gradation collapse is suppressed compared to the conventional case. Signal. Therefore, as in the first embodiment, when performing linear matrix color correction processing and dynamic range compression processing, it is possible to maintain the gradation of RGB signals with high saturation.

The signal processing circuit 2 may use the matrices LM2 ⁻¹ and LM2 set by the second method instead of using the matrices LM1 and LM1 ⁻¹ set by the first method. . Specifically, the linear matrix circuit 10-2, to the value of the RGB signal, by multiplying the matrix LM2 ^-1 preset three rows and three columns, performs color correction processing of the first linear matrix . The linear matrix circuit 11-2 multiplies the value of the RGB signal (Rmid ′, Gmid ′, Bmid ′ signal) by a preset 3 × 3 matrix LM2 to thereby change the color of the second linear matrix. Perform correction processing.

The matrices LM1 (= LM2 ⁻¹ ) and LM2 in this case are obtained by a designer's calculation using a predetermined method and set in advance. As this setting method, the second method similar to that of the first embodiment is used, and further, a condition (LM2 · LM1 = I) in which LM1 is an inverse matrix of LM2 is added.

Example 3
Next, a third embodiment (Example 3) of the present invention will be described. FIG. 3 is a block diagram illustrating a configuration example of the signal processing circuit according to the third embodiment. The signal processing circuit 3 includes linear matrix circuits 10-1 and 11-1 and a level diagram adjustment circuit 103.

  When the signal processing circuit 1 of the first embodiment shown in FIG. 1 and the signal processing circuit 3 of the third embodiment shown in FIG. 3 are compared, the overall configuration is the same, and the linear matrix circuits 10-1 and 11-1 are the same. Common in that it has. On the other hand, the signal processing circuit 3 of the third embodiment is different in that a level diagram adjustment circuit 103 different from the dynamic range compression circuit 102 of the signal processing circuit 1 of the first embodiment is provided.

  Similar to the linear matrix circuit 10-1 shown in FIG. 1, the linear matrix circuit 10-1 performs a first color correction process of the linear matrix using a preset 3 × 3 matrix LM1. The linear matrix circuit 10-1 then outputs RGB signals (Rmid, Gmid, Bmid signals) to the level diagram adjustment circuit 103.

  The level diagram adjusting circuit 103 receives RGB signals (Rmid, Gmid, Bmid signals) from the linear matrix circuit 10-1. The level diagram adjustment circuit 103 performs level diagram adjustment processing by performing clip processing for aligning saturation points between RGB signals (Rmid, Gmid, and Bmid signals). The level diagram adjusting circuit 103 outputs RGB signals (Rmid ′, Gmid ′, Bmid ′ signals) to the linear matrix circuit 11-1.

  As a result, the RGB signals (Rmid, Gmid, Bmid signals) that have undergone color balance processing are clipped to the same signal value when a difference occurs between these signals. Rmid ′, Gmid ′, Bmid ′ signals) are output.

  The linear matrix circuit 11-1 receives RGB signals (Rmid ′, Gmid ′, Bmid ′ signals) that have undergone level diamond adjustment processing (clip processing) from the level diamond adjustment circuit 103. Then, similarly to the linear matrix circuit 11-1 shown in FIG. 1, the linear matrix circuit 11-1 performs a second linear matrix color correction process using a preset 3 × 3 matrix LM2. Do.

  The linear matrix circuit 11-1 outputs an RGB signal (Rfin, Gfin, Bfin signal) that has been subjected to the same color correction processing as that of the prior art and in which gradation collapse is suppressed as compared with the prior art.

  As described above, according to the signal processing circuit 3 of the third embodiment, the linear matrix circuits 10-1 and 11-1 perform the same processing as that of the first embodiment, and the level diagram adjustment circuit 103 performs the RGB signal (Rmid , Gmid, Bmid signals), the level diagram adjustment processing is performed by performing the clipping processing for aligning the saturation points.

  Also in this case, the RGB signals (Rfin, Gfin, and Bfin signals) output from the linear matrix circuit 11-1 are subjected to the same color correction processing as that in the past, and the gradation collapse is suppressed as compared with the conventional case. Signal. Therefore, when performing linear matrix color correction processing and level diagram adjustment processing, the gradation of RGB signals with high saturation can be maintained. The third embodiment is particularly effective in HDR imaging when the signal processing circuit 3 is applied to an imaging apparatus.

  The signal processing circuit 3 includes linear matrix circuits 10-1 and 11-1 as in the first embodiment shown in FIG. 1, but instead of these, the signal processing circuit 3 of the second embodiment shown in FIG. Linear matrix circuits 10-2 and 11-2 may be provided.

Example 4
Next, a fourth embodiment (Example 4) of the present invention will be described. FIG. 4 is a block diagram illustrating a configuration example of the imaging apparatus according to the fourth embodiment. The imaging device 20 includes an imaging device 21, a black balance / white balance circuit 22, signal processing circuits 1, 2, 3 and a video processing circuit 23.

  In the fourth embodiment, the signal processing circuit 1 of the first embodiment shown in FIG. 1, the signal processing circuit 2 of the second embodiment shown in FIG. 2, and the signal processing circuit 3 of the third embodiment shown in FIG. This is an example applied to FIG. The signal processing circuits 1, 2, and 3 are the signal processing circuit 1 of the first embodiment shown in FIG. 1, the signal processing circuit 2 of the second embodiment shown in FIG. 2, and the signal processing of the third embodiment shown in FIG. Any one of the circuits 3 is provided.

  The image sensor 21 receives subject light through an optical system (not shown), converts the light amount into an electrical signal, and outputs the signal as an RGB signal. The RGB signal is amplified, and the analog RGB signal is converted into a digital RGB signal.

  The black balance / white balance circuit 22 receives an RGB signal from the image sensor 21, performs various processes such as a black level adjustment process and a white level adjustment process on the RGB signal value, and outputs the RGB signal after the adjustment process. Output to the processing circuits 1, 2, 3.

  As described above, the signal processing circuits 1, 2, and 3 are any one of the signal processing circuits 1, 2, and 3, and input the adjusted RGB signal from the black balance / white balance circuit 22. Then, the signal processing circuits 1, 2, and 3 perform the first linear matrix color correction process, the dynamic range compression process or the level diagram adjustment process, and the second linear matrix color correction process. The signal processing circuits 1, 2, 3 output RGB signals (Rfin, Gfin, Bfin signals) to the video processing circuit 23.

  The video processing circuit 23 receives RGB signals (Rfin, Gfin, Bfin signals) from the signal processing circuits 1, 2, 3, and performs noise reduction processing and detail enhancement on the values of the RGB signals (Rfin, Gfin, Bfin signals). Various video processing such as processing is performed. Then, the video processing circuit 23 outputs the RGB signal after the video processing as a video signal.

  As described above, according to the imaging device 20 of the fourth embodiment, the signal processing circuits 1, 2, and 3 can be applied to the imaging device 20. Therefore, as in the first to third embodiments, when performing linear matrix color correction processing, dynamic range compression processing, or level diagram adjustment processing, it is possible to maintain the gradation of a highly saturated RGB signal.

  4 includes the black balance / white balance circuit 22, but the black balance / white balance circuit 22 may not necessarily be provided. In this case, the signal processing circuits 1, 2, and 3 receive RGB signals from the image sensor 21.

Example 5
Next, a fifth embodiment (Example 5) of the present invention will be described. As described above, the signal processing circuits 1, 2, and 3 of the first, second, and third embodiments perform the same color correction processing as that of the prior art, and RGB that suppresses the collapse of gradation compared to the prior art. Signals (Rfin, Gfin, Bfin signals) can be obtained.

  However, there are cases where an out-of-gamut signal is generated by the color correction processing of the signal processing circuits 1, 2, 3, and gradation collapse cannot be sufficiently suppressed. Here, the out-of-gamut signal is the minimum value of the predetermined range in which any value of the signal is not included in the predetermined range of the color gamut in each of the RGB signals after the color correction processing is performed. A signal having a value (eg, negative value) smaller than a reference value.

  FIG. 12 is a diagram for explaining the transition of the signal value in the signal processing circuit 1 of the first embodiment shown in FIG. is there. It is assumed that an RGB signal of subject light in which blue is dominant is input to the signal processing circuit 1.

  12A to 12D, the horizontal axis indicates the incident light intensity, and the vertical axis indicates the signal value of the RGB signal. FIG. 12A shows the characteristics of the RGB signals input to the linear matrix circuit 10-1, and FIG. 12B is output from the linear matrix circuit 10-1 and input to the dynamic range compression circuit 102. The characteristics of RGB signals (Rmid, Gmid, Bmid signals) are shown. FIG. 12C shows the characteristics of the RGB signals (Rmid ′, Gmid ′, Bmid ′ signals) output from the dynamic range compression circuit 102 and input to the linear matrix circuit 11-1, and FIG. ) Shows the characteristics of the RGB signals (Rfin, Gfin, Bfin signals) output from the linear matrix circuit 11-1. The signal value transitions shown in FIGS. 12A, 12B, and 12C are the same as those shown in FIGS. 5A, 5B, and 5C.

  Referring to FIG. 12D, in the RGB signals (Rfin, Gfin, Bfin signals) generated by the color correction processing of the signal processing circuit 1, the signal values of the R and G signals are submerged, and the signal values are reduced. There is a negative value area. Specifically, there is a predetermined position (point P2) at which the R and G signals change from a negative value to a positive value in the region of the incident light intensity in which the B signal is suppressed by the dynamic range compression circuit 102. In this area, the signal values of the R and G signals are negative values.

  Therefore, the region of the incident light intensity between the points P1 and P2 (between the start point where the signal value of the B signal is suppressed and the point where the signal value of the R and G signals changes from a negative value to a positive value) In the region), the signal value of the B signal is not changed, and the signal value of the R and G signals is changed, but the signal value is a negative value. The R and G signals in this area are out-of-gamut signals. For this reason, in the region of the incident light intensity between the points P1 and P2, gradation collapse occurs. As described later, in the fifth embodiment, this out-of-gamut signal is clipped to a reference value.

  As described above, an out-of-gamut signal is generated by the color correction processing of the signal processing circuits 1, 2, and 3, and in the region of the incident light intensity between the points P 1 and P 2, the gradation collapse is sufficiently suppressed. Can not. In other words, the effect of maintaining gradation with respect to a highly saturated RGB signal can be obtained, but the effect may not be sufficient in a predetermined area where an out-of-gamut signal exists.

  Therefore, in the fifth embodiment, the out-of-gamut signal is clipped in advance to the reference value of the gamut range before the color correction process so that the out-of-gamut signal is not generated by the color correction process.

  FIG. 10 is a block diagram illustrating a configuration example of the signal processing circuit according to the fifth embodiment. The signal processing circuit 4 includes an out-of-gamut signal processing unit 30 in front of the signal processing circuits 1, 2, and 3 described above. The signal processing circuits 1, 2, and 3 are the signal processing circuit 1 of the first embodiment shown in FIG. 1, the signal processing circuit 2 of the second embodiment shown in FIG. 2, and the signal processing of the third embodiment shown in FIG. Any one of the circuits 3 is provided.

  1 to 3 and the signal processing circuit 4 of the fifth embodiment shown in FIG. 10 are compared with the signal processing circuit 4 of the fifth embodiment. The signal processing circuits 1, 2, and 3 are different from the signal processing circuits 1, 2, and 3 of the first, second, and third embodiments in that an out-of-gamut signal processing unit 30 is provided.

  The out-of-gamut signal processing unit 30 includes linear matrix circuits 31 and 33 and a clip circuit 32. The linear matrix circuit 31 receives an RGB signal and performs a color correction process on the linear matrix by multiplying the value of the RGB signal by a preset 3 × 3 matrix LMorg. The matrix LMorg is a matrix composed of linear matrix coefficients applied to the entire signal processing circuits 1, 2, 3 in the subsequent stage, as expressed by the equations (3) and (6). Used. As a result, the output signal obtained by the color correction processing by the signal processing circuits 1, 2 and 3 in the subsequent stage is reproduced.

  The linear matrix circuit 31 outputs RGB signals (R1, G1, B1 signals) after color correction processing of the linear matrix to the clip circuit 32.

  The clip circuit 32 receives the RGB signals (R1, G1, B1 signals) that have undergone linear matrix color correction processing from the linear matrix circuit 31, and outputs the RGB signals (R1, G1, B1 signals) to the respective signals. Thus, a process of clipping an out-of-gamut signal that is a signal smaller than a preset reference value to the reference value is performed.

  As the reference value set in advance, for example, 0 is used for a general image signal. In addition, when the minimum signal value of the color gamut range that can be taken as the RGB signal is defined by the standard, the minimum value is used as the reference value. For example, since the minimum value of the 10-bit SDI signal is 4, 4 is used as the reference value. Further, a negative value may be used as the reference value.

  The clip circuit 32 outputs the RGB signals (R2, G2, B2 signals) after the clip process to the linear matrix circuit 33.

The linear matrix circuit 33 inputs the RGB signal (R2, G2, B2 signal) subjected to the clipping process from the clip circuit 32, and multiplies the matrix LMorg ⁻¹ of 3 rows and 3 columns set in advance, so that the linear matrix circuit 33 Perform matrix color correction processing. The matrix LMorg ⁻¹ used in the linear matrix circuit 33 is an inverse matrix of the matrix LMorg used in the linear matrix circuit 31. As a result, the RGB signal that is the original input signal is reproduced for the in-gamut signal that is not affected by the clip. For signals outside the color gamut, RGB signals clipped within the color gamut are reproduced.

  The linear matrix circuit 33 outputs the RGB signals (R ′, G ′, B ′ signals) after the linear matrix color correction processing to the signal processing circuits 1, 2, 3.

  As described above, the linear matrix circuits 31 and 33 and the clip circuit 32 included in the out-of-gamut signal processing unit 30 allow the out-of-gamut signal (the signal processing circuits 1, 2, and 3 in the subsequent stages are input signals of the signal processing circuit 4). An out-of-gamut signal generated when color correction processing is performed on an RGB signal is clipped into the color gamut. That is, the RGB signals (R ′, G) input to the signal processing circuits 1, 2, 3 are prevented so that out-of-gamut signals are not generated when the color correction processing is performed in the signal processing circuits 1, 2, 3. ', B' signal) is generated.

  The signal processing circuits 1, 2, and 3 receive RGB signals (R ′, G ′, B ′ signals) after linear matrix color correction processing from the linear matrix circuit 33 of the out-of-gamut signal processing unit 30. The same processing described in FIG. 3 is performed. Then, the signal processing circuits 1, 2, and 3 output RGB signals (Rfin, Gfin, and Bfin signals) in which desired color reproduction is achieved and gradation collapse is suppressed.

The matrices LMorg and LMorg- ¹ are obtained by a designer's calculation using a predetermined method and set in advance. In this case, the product of the matrix LMorg and the matrix LMorg ⁻¹ is a unit matrix (LMorg ⁻¹ · LMorg = I, I is a 3 × 3 unit matrix).

As described above, according to the signal processing circuit 4 of the fifth embodiment, the linear matrix circuit 31 of the out-of-gamut signal processing unit 30 linearly multiplies the RGB signal value by the preset matrix LMorg. Matrix color correction processing is performed to generate RGB signals (R1, G1, B1 signals). The clipping circuit 32 clips a signal smaller than a reference value set in advance for each value of the RGB signals (R1, G1, B1 signals) to a reference value, and outputs the RGB signals (R2, G2, B2 signals). Generate. The linear matrix circuit 33 performs color correction processing of the linear matrix by multiplying the value of the RGB signal (R2, G2, B2 signal) by a preset matrix LMorg ⁻¹ to obtain the RGB signal (R ′, G ′ and B ′ signals).

  As a result, an out-of-gamut signal is not generated by the color correction processing of the signal processing circuits 1, 2, and 3 in the subsequent stage for the RGB signals (R ', G', B 'signals).

  Then, the signal processing circuits 1, 2, and 3 perform the same processing as that of the first, second, and third embodiments on the values of the RGB signals (R ′, G ′, B ′ signals), and the RGB signals (Rfin, Gfin). , Bfin signal).

  That is, the out-of-gamut signal processing unit 30 provided in the previous stage of the signal processing circuits 1, 2, and 3 performs the clip processing after performing the same color correction processing as that of the signal processing circuits 1, 2, and 3, and outputs the out-of-gamut signal. Is changed to a signal in the color gamut and returned to the RGB signal (R ′, G ′, B ′ signal) corresponding to the original RGB signal before the color correction processing. Then, the signal processing circuits 1, 2, and 3 perform color correction processing on the RGB signals (R ′, G ′, and B ′ signals) to generate RGB signals (Rfin, Gfin, and Bfin signals).

  Thereby, the RGB signals (Rfin, Gfin, and Bfin signals) do not include out-of-gamut signals, and the gradation collapse can be sufficiently suppressed. Therefore, it is possible to sufficiently maintain the gradation of the RGB signal of high saturation when performing the color correction processing of the linear matrix and the knee or limit processing.

  The signal processing circuit 4 according to the fifth embodiment is applicable to the imaging device 20 according to the fourth embodiment illustrated in FIG. That is, the imaging device 20 according to the fourth embodiment may include the signal processing circuit 4 instead of the signal processing circuits 1, 2, and 3.

Example 6
Next, a sixth embodiment (Example 6) of the present invention will be described. In the sixth embodiment, the linear matrix circuit 33 of the out-of-gamut signal processing unit 30 and the linear matrix circuits 10-1 and 10-2 of the signal processing circuits 1, 2, and 3 are combined in the fifth embodiment to form one linear. It is an example which comprises a matrix circuit. Thereby, compared with Example 5, the number of circuits can be decreased and the whole can be reduced in size and cost.

  FIG. 11 is a block diagram illustrating a configuration example of the signal processing circuit according to the sixth embodiment. The signal processing circuit 5 includes linear matrix circuits 31 and 40, a clip circuit 32, a dynamic range compression circuit 102 or a level diagram adjustment circuit 103, and a linear matrix circuit 11-1 or a linear matrix circuit 11-2.

  When the signal processing circuit 4 of the fifth embodiment shown in FIG. 10 and the signal processing circuit 5 of the sixth embodiment shown in FIG. 11 are compared, the signal processing circuits 4 and 5 are equivalent. On the other hand, the signal processing circuit 5 includes a linear matrix circuit 40 in which the linear matrix circuit 33 of the signal processing circuit 4 and the linear matrix circuits 10-1 and 10-2 of the signal processing circuits 1, 2, and 3 are combined. This is different from the signal processing circuit 4.

  The linear matrix circuit 31 and clip circuit 32 of the out-of-gamut signal processing unit 30, the dynamic range compression circuit 102, the level diagram adjustment circuit 103, and the linear matrix circuits 11-1 and 11-2 of the signal processing circuits 1, 2, and 3 are described above. Since it is as follows, description is abbreviate | omitted.

The linear matrix circuit 40 receives the RGB signal (R2, G2, B2 signal) subjected to the clipping process from the clipping circuit 32, and multiplies the matrix LM1 · LMorg ⁻¹ of 3 rows and 3 columns set in advance. The color correction processing of the linear matrix is performed.

The matrix LM1 is a preset 3 × 3 matrix used in the linear matrix circuits 10-1 and 10-2. The matrix LMorg ⁻¹ is a preset 3 × 3 matrix used in the linear matrix circuit 33. The linear matrix circuit 40 is the product of a matrix LM1 and matrix LMorg ^-1 matrix LM1 · LMorg ^-1 is used.

  The linear matrix circuit 40 outputs RGB signals (Rmid, Gmid, Bmid signals) after color correction processing of the linear matrix to the dynamic range compression circuit 102 or the level diagram adjustment circuit 103.

As described above, according to the signal processing circuit 5 of the sixth embodiment, the linear matrix circuit 40 is configured by combining the linear matrix circuit 33 and the linear matrix circuits 10-1 and 10-2. By multiplying the values of the RGB signals (R2, G2, B2 signals) subjected to the clip processing by a preset matrix LM1 · LMorg ⁻¹ , color correction processing of the linear matrix is performed, and the RGB signal (Rmid , Gmid, Bmid signals).

  The signal processing circuit 5 includes one linear matrix circuit 40 equivalent to these circuits instead of the two circuits including the linear matrix circuit 33 and the linear matrix circuits 10-1 and 10-2 of the fifth embodiment.

  As a result, as in the fifth embodiment, when performing linear matrix color correction processing and knee or limit processing, it is possible to sufficiently maintain the gradation of the RGB signal of high saturation. Further, compared to the fifth embodiment, the number of circuits can be reduced, and the overall size and cost can be reduced.

  The signal processing circuit 5 according to the sixth embodiment is applicable to the imaging device 20 according to the fourth embodiment illustrated in FIG. That is, the imaging device 20 according to the fourth embodiment may include the signal processing circuit 5 instead of the signal processing circuits 1, 2, and 3.

  Next, the transition of the signal value in the signal processing circuit 5 according to the sixth embodiment will be described. FIG. 13 is a diagram for explaining signal value transition in the signal processing circuit 5 of the sixth embodiment illustrated in FIG. 11, that is, signal value transition when there is out-of-gamut signal processing. It is assumed that an RGB signal of subject light in which blue is dominant is input to the signal processing circuit 5.

  13A to 13F, the horizontal axis indicates the incident light intensity, and the vertical axis indicates the signal value of the RGB signal. 13A shows the characteristics of the RGB signals input to the linear matrix circuit 31, and FIG. 13B shows the RGB signals (R1, G1) output from the linear matrix circuit 31 and input to the clip circuit 32. , B1 signal). 13C shows the characteristics of the RGB signals (R2, G2, B2 signals) output from the clip circuit 32 and input to the linear matrix circuit 40, and FIG. 13D shows the linear matrix circuit 40. The characteristics of the RGB signals (Rmid, Gmid, Bmid signals) output by the above and input to the dynamic range compression circuit 102 are shown.

  FIG. 13E shows the characteristics of the RGB signals (Rmid ′, Gmid ′, Bmid ′ signals) output from the dynamic range compression circuit 102 and input to the linear matrix circuit 11-1. FIG. ) Shows the characteristics of the RGB signals (Rfin, Gfin, Bfin signals) output from the linear matrix circuit 11-1.

  The linear matrix circuit 31 performs linear matrix color correction processing on the RGB signals shown in FIG. 13A using the matrix LMorg, and the RGB signals (R1, G1, B1 signals) shown in FIG. Is generated. As a result, the signal value of the B signal increases, the signal value of the R and G signals decreases, and the signal is out of gamut.

  For each of the RGB signals (R1, G1, B1 signals) shown in FIG. 13 (b), a signal (negative signal) smaller than the reference value (0 value) is clipped to the reference value by the clipping circuit 32. RGB signals (R2, G2, B2 signals) shown in FIG. 13C are generated. As a result, the out-of-gamut signal, which is a negative R, G signal, is clipped to a reference value of 0.

For the RGB signals (R2, G2, B2 signals) shown in FIG. 13C, the linear matrix circuit 40 performs linear matrix color correction processing using the matrices LM1 and LMorg ⁻¹ , and FIG. RGB signals (Rmid, Gmid, Bmid signals) shown in FIG. As a result, the RGB signals (Rmid, Gmid, and Bmid signals) are signals with lower saturation than in FIG.

  The dynamic range compression circuit 102 performs dynamic range compression processing on the RGB signals (Rmid, Gmid, Bmid signals) shown in FIG. 13D, and the RGB signals (Rmid ′, Gmid ′ shown in FIG. 13E). , Bmid ′ signal) is generated. FIG. 13E corresponds to FIGS. 5C and 12C.

  For the RGB signals (Rmid ′, Gmid ′, Bmid ′ signals) shown in FIG. 13E, the linear matrix circuit 11-1 performs color correction processing of the linear matrix using the matrix LM2, and FIG. RGB signals (Rfin, Gfin, Bfin signals) shown in FIG. Thus, the R and G signals (Rfin and Gfin signals) shown in FIG. 13 (f) are different from the R and G signals (Rfin and Gfin signals) shown in FIG. Become. Therefore, it is possible to avoid the signal values of the R and G signals from being hidden, and to suppress the gradation collapse caused by the negative signal value.

(Experimental result of Example 6)
Next, experimental results by computer simulation of Example 6 will be described. FIG. 14 is a diagram for explaining experimental results of Example 6 when a high-saturation RGB signal is input. (1) shows changes in the signal value by the signal processing circuit 100 of the prior art shown in FIG. 7, and (2) shows changes in the video signal in (1). Further, (3) shows the change of the signal value by the signal processing circuit 1 of the first embodiment shown in FIG. 1, and (4) shows the change of the video signal in (3). (5) shows a change in the signal value by the signal processing circuit 5 of the sixth embodiment shown in FIG. 11, and (6) shows a change in the video signal in (5).

  14 (1), (3), and (5), the horizontal axis indicates the incident light intensity, and the vertical axis indicates the signal value of the RGB signal. The RGB signals are output signals of the signal processing circuits 100, 1, 5 respectively, that is, RGB signals (Rfin, Gfin, Bfin signals). 14 (2), (4), and (6), the horizontal axis indicates the same incident light intensity as in FIGS. 14 (1), (3), and (5). FIGS. 14 (2), (4), and (6) show gradations in the entire range of incident light intensity.

  14 (1) to 14 (6) are similar to FIGS. 6 (1) to 6 (4), the matrix LMorg used in the signal processing circuit 100 of the prior art, the signal processing circuit 1 of the first embodiment, and the sixth embodiment. This is an experimental result when the product of the matrices LM1 and LM2 used in the signal processing circuit 5 is the same. Further, the matrix LMorg used in the signal processing circuit 100 of the prior art is the same as the matrix LMorg used in the signal processing circuit 5 of the sixth embodiment.

  It is assumed that the RGB signal of subject light in which blue is dominant is output from the image sensor, and the RGB signal is input to the signal processing circuits 100, 1, and 5, respectively.

  From FIG. 14 (1), in the prior art, the signal value of the B signal is suppressed by the dynamic range compression processing in the range where the incident light intensity is about 80 or more and 600 or less, but the signal value of the R and G signals is It can be seen that the signal value does not increase and is stagnated as a whole. For this reason, as shown in FIG. 14B, the gradation of the video signal is crushed.

  Also, from FIG. 14 (3), the signal value of the B signal of Example 1 is suppressed by the dynamic range compression processing in the range where the incident light intensity is about 80 to 600, and the signal values of the R and G signals are It can be seen that the incident light intensity increases in the range of about 250 or more and 600 or less, and approaches white as the incident light intensity increases.

  However, the signal values of all the RGB signals of Example 1 do not change when the incident light intensity is in the range of about 80 to about 250 (the R and G signals in this range are as shown in FIG. 12D). The signal is out of gamut). For this reason, as shown in FIG. 14 (4), the video signal shows a change in gradation over the entire range, but a part of the range (incident light intensity is a range of about 80 or more and about 250 or less). In, there is no change in gradation, and partial gradation collapse occurs.

  On the other hand, from FIG. 14 (5), the signal value of the B signal of Example 6 is suppressed by the dynamic range compression processing in the range where the incident light intensity is about 80 or more and 600 or less. It can be seen that the values increase in the range where the incident light intensity is about 110 or more and 600 or less, respectively, and approach white as the incident light intensity increases. For this reason, as shown in FIG. 14 (6), the video signal shows a change in gradation in a wider range than that in FIG. 14 (4), and a natural gradation without collapse is maintained.

  As described above, the first embodiment can suppress the gradation collapse compared to the conventional technique, and the sixth embodiment can further suppress the gradation collapse compared to the first embodiment.

  Note that a normal computer can be used as the hardware configuration of the signal processing circuits 1, 2, 3, 4, and 5 according to the first, second, third, fifth, and sixth embodiments of the present invention. The signal processing circuits 1, 2, 3, 4, and 5 are configured by a computer including a volatile storage medium such as a CPU and a RAM, a non-volatile storage medium such as a ROM, and an interface. Each function of the linear matrix circuits 10-1 and 11-1 and the dynamic range compression circuit 102 provided in the signal processing circuit 1 is realized by causing the CPU to execute a program describing these functions. Each function of the linear matrix circuits 10-2 and 11-2 and the dynamic range compression circuit 102 provided in the signal processing circuit 2 is realized by causing the CPU to execute a program describing these functions. Each function of the linear matrix circuits 10-1 and 11-1 and the level diagram adjusting circuit 103 provided in the signal processing circuit 3 is realized by causing the CPU to execute a program describing these functions.

  Each function of the linear matrix circuits 31 and 33 and the clip circuit 32 provided in the signal processing circuit 4 is realized by causing the CPU to execute a program describing these functions. Furthermore, each function of the linear matrix circuits 31 and 40 and the clip circuit 32 provided in the signal processing circuit 5 is realized by causing the CPU to execute a program describing these functions.

  These programs are stored in the storage medium and read out and executed by the CPU. These programs can also be stored and distributed in a storage medium such as a magnetic disk (floppy (registered trademark) disk, hard disk, etc.), optical disk (CD-ROM, DVD, etc.), semiconductor memory, etc. You can also send and receive.

  In the first, second, third, fifth, and sixth embodiments of the present invention, each component from the linear matrix circuit 10-1 to the linear matrix circuit 11-1 of the signal processing circuit 1 according to the first embodiment shown in FIG. 2, the processing of each component from the linear matrix circuit 10-2 to the linear matrix circuit 11-2 of the signal processing circuit 2 of the second embodiment shown in FIG. 2, the signal processing circuit of the third embodiment shown in FIG. 3, the processing of each component from the linear matrix circuit 10-1 to the linear matrix circuit 11-1, the signal processing circuits 1, 2, 3 from the linear matrix circuit 31 of the signal processing circuit 4 of the fifth embodiment shown in FIG. The processing of each component up to the linear matrix circuits 11-1 and 11-2 and the linear matrix circuit 31 to the linear matrix circuit 11- of the signal processing circuit 5 of the sixth embodiment shown in FIG. The processing of each component up to 11-2 may also be realized by LSI chip is an integrated circuit mounted to the signal processing circuit 1,2,3,4,5. These may be individually made into one chip, or a part or all of them may be made into one chip.

  Further, instead of the LSI, it may be realized by a chip such as a VLSI or ULSI having a different degree of integration. Furthermore, the present invention is not limited to a chip such as an LSI, and a dedicated circuit or a general-purpose processor may be used, or an FPGA (Field Programmable Gate Array) may be used.

1, 2, 3, 4, 5, 100 Signal processing circuit 10, 11, 31, 33, 40, 101 Linear matrix circuit 20 Imaging device 21 Imaging element 22 Black balance / white balance circuit 23 Video processing circuit 30 Out-of-gamut signal processing Part 32 clip circuit 102 dynamic range compression circuit 103 level diagram adjustment circuit

Claims (12)

In a signal processing circuit that performs predetermined processing on RGB signals,
A first linear matrix circuit that performs color correction processing of a linear matrix by multiplying the RGB signal by a first matrix of 3 rows and 3 columns set in advance;
A dynamic range compression circuit that performs dynamic range compression processing on the RGB signal that has undergone the color correction processing by the first linear matrix circuit;
A second color correction process for a linear matrix is performed by multiplying the RGB signal that has been subjected to the dynamic range compression process by the dynamic range compression circuit by a preset second matrix of 3 rows and 3 columns. A linear matrix circuit,
An RGB signal, which has been subjected to the color correction processing using the first matrix by the first linear matrix circuit, is a 3 × 3 matrix that is the product of the first matrix and the second matrix. A signal processing circuit characterized in that the signal has a lower saturation than the RGB signal in the case where the color correction processing is performed using. The signal processing circuit according to claim 1,
The first linear matrix circuit includes:
By multiplying the RGB signal by the first matrix and performing a linear matrix color correction process, the saturation of the RGB signal is reduced,
The second linear matrix circuit includes:
The RGB signal subjected to the dynamic range compression processing by the dynamic range compression circuit is multiplied by the second matrix to perform a linear matrix color correction process, thereby increasing the saturation of the RGB signal, A signal processing circuit for adjusting a color balance of the RGB signals. The signal processing circuit according to claim 1,
The first linear matrix circuit includes:
By multiplying the RGB signal by the first matrix and performing color correction processing of a linear matrix, the color balance of the RGB signal is adjusted,
The second linear matrix circuit includes:
Increasing the saturation of the RGB signal by multiplying the RGB signal that has been subjected to the dynamic range compression processing by the dynamic range compression circuit with the second matrix and performing linear matrix color correction processing; A signal processing circuit. The signal processing circuit according to claim 1,
The first matrix LM1 and the second matrix are defined as a 3-by-3 matrix, which is the product of the first matrix and the second matrix, with LMorg and α as a preset positive real parameter. LM2
A signal processing circuit characterized by that. The signal processing circuit according to claim 1,
A matrix of 3 rows and 3 columns, which is the product of the first matrix and the second matrix, is set to LMorg, and α _R , α _G , and α _B are set to positive real numbers set in advance corresponding to the RGB signals, respectively. As parameters, the first matrix LM1 and the second matrix LM2 are
A signal processing circuit characterized by that. The signal processing circuit according to claim 1,
The second matrix LM2 and the first matrix with LMorg as a matrix of 3 rows and 3 columns, which is the product of the first matrix and the second matrix, and β as a preset positive real parameter LM1
A signal processing circuit characterized by that. The signal processing circuit according to claim 1,
A matrix of 3 rows and 3 columns, which is the product of the first matrix and the second matrix, is set to LMorg, β _R , β _G , and β _B are set to positive real numbers set in advance corresponding to the RGB signals, respectively. As parameters, the second matrix LM2 and the first matrix LM1,
A signal processing circuit characterized by that. In the signal processing circuit according to any one of claims 1 to 7,
The signal processing circuit, wherein the second matrix is an inverse matrix of the first matrix. In the signal processing circuit according to any one of claims 1 to 8,
The first linear matrix circuit includes:
A signal processing circuit that performs color correction processing of the linear matrix by multiplying the first matrix by the RGB signal obtained by an image sensor. In the signal processing circuit according to any one of claims 1 to 9,
Further, by multiplying the RGB signal by a preset third matrix of 3 rows and 3 columns, which is the product of the second matrix and the first matrix, color correction processing of a linear matrix is performed. A third linear matrix circuit to perform;
A clip that clips a signal smaller than a preset reference value to the reference value for each of the RGB signals subjected to the color correction processing by the third linear matrix circuit, and generates a clipped RGB signal Circuit,
A fourth linear matrix circuit that performs linear matrix color correction processing by multiplying the RGB signal generated by the clipping circuit by a preset fourth matrix that is an inverse matrix of the third matrix. And comprising
The first linear matrix circuit includes:
A signal processing circuit that performs color correction processing of a linear matrix by multiplying the RGB signal subjected to the color correction processing by the fourth linear matrix circuit with the first matrix. The signal processing circuit according to claim 10.
In place of the fourth linear matrix circuit and the first linear matrix circuit, a fifth linear matrix circuit is provided,
The fifth linear matrix circuit includes:
By multiplying the RGB signal generated by the clipping circuit by a preset fifth matrix of 3 rows and 3 columns, which is the product of the first matrix and the fourth matrix, Perform color correction processing,
The dynamic range compression circuit includes:
A signal processing circuit, wherein dynamic range compression processing is performed on an RGB signal that has been subjected to the color correction processing by the fifth linear matrix circuit.   The program for functioning a computer as a signal processing circuit as described in any one of Claim 1-11.