A color characterization modeling method for digital cameras based on optimized spectral sensitivity

Abstract

The present invention discloses a digital camera color characterization modeling method based on optimized spectral sensitivity. First, a color chart is photographed using a digital camera under a known light source environment, and the camera's spectral sensitivity is estimated. The spectral sensitivity is used to predict the response values ​​of training samples. The predicted response values ​​are normalized and expanded using a polynomial. Then, a pseudo-inverse algorithm is used to construct a color characterization conversion matrix between the predicted response values ​​of the training samples and the three stimuli. Finally, the constructed color characterization matrix is ​​used to calculate the three stimulus values ​​of the test sample to obtain the final three stimulus values ​​of the test sample.

Description

Technical Field

[0001] The invention belongs to the technical field of computer digital image processing, and in particular relates to a color characterization modeling method for a digital camera based on optimized spectral sensitivity. Background Art

[0002] With the continuous advancement of computer technology, the transmission and reproduction of digital images across various media devices has become increasingly widespread. Most digital image acquisition and display devices use the RGB color space, which is a device-dependent color space. Due to differences in color characteristics between devices, image transmission and reproduction are prone to color distortion. For example, the display quality of the same image may differ between different devices; during the printing process, the color of the display sample and the final print may also be inconsistent. To achieve accurate color reproduction, color management is required during the transmission and reproduction process between devices.

[0003] Color characterization of digital cameras is the process of converting the color information in images captured by a digital camera from the camera's device-dependent color space (such as RGB) to a standardized, device-independent color space (such as CIEXYZ or CIELAB). The RGB color space of a digital camera is often affected by the camera's sensor characteristics, lens optical performance, and image processing algorithms. These factors can cause images captured in the same scene to exhibit different color appearances on different cameras. Through color characterization, digital cameras can accurately map captured color information to a device-independent color space based on a set of standardized reference colors, thereby reducing the impact of color differences between devices. The goal is to ensure that image colors remain consistent and accurate when transferred between different devices. Whether viewed or output on a computer monitor, mobile phone screen, printer, or projector, images that have undergone color characterization can reproduce colors as close to reality as possible.

[0004] However, existing color characterization methods for digital cameras mostly rely on photographing color charts in real-world environments. This involves capturing color charts in real-world scenes to establish a color characterization conversion matrix. This approach struggles to achieve effective color characterization in complex lighting environments. Furthermore, in practical applications, capturing color charts under every different lighting condition is impractical. Currently, no effective solutions to these problems have been proposed, either in academia or industry, both domestically and internationally. Summary of the Invention

[0005] The purpose of the present invention is to solve the problems described in the background art and to propose a color characterization modeling method for a digital camera based on optimized spectral sensitivity.

[0006] In response to the above-mentioned existing research problems, the present invention proposes a method to solve the problem. First, a digital camera is used to shoot a color card under a known light source environment, and the spectral sensitivity of the digital camera is estimated. The response value of the training sample is predicted using the spectral sensitivity. The predicted response value is normalized and expanded using a polynomial. Then, a pseudo-inverse algorithm is used to construct a color characterization conversion matrix between the predicted response value of the training sample and the three stimuli. Finally, the constructed color characterization matrix is ​​used to calculate the three stimulus values ​​of the test sample to obtain the final three stimulus values ​​of the test sample. The technical solution of the present invention is a digital camera color characterization modeling method based on optimized spectral sensitivity, which specifically includes the following steps:

[0007] Step 1: Use a digital camera to shoot the training sample, obtain the original format response value of the training sample, and perform channel-by-channel normalization processing on it to obtain the normalized response value X' of the original format response value of the training sample;

[0008] Step 2: Use a spectrophotometer and an illuminometer to measure the spectral data R' of the training sample set and the spectral power distribution L' of the ambient light source respectively;

[0009] Step 3, estimating the spectral sensitivity of the digital camera to obtain the camera spectral sensitivity S;

[0010] Step 4, using the spectral data of the training sample set to calculate the corresponding tristimulus value W;

[0011] Step 5: Use the estimated camera spectral sensitivity to predict the camera response value of the training sample set and perform polynomial expansion to obtain the expanded response value D train,exp ;

[0012] Step 6: Use the predicted response value D after polynomial expansion of the training sample set train,exp Construct the color characterization conversion matrix Y with the tristimulus values ​​W;

[0013] Step 7: Use a camera to shoot the test sample, obtain the test sample response value and perform polynomial expansion D test,exp ;

[0014] Step 8: Use the color characterization conversion matrix to complete the color characterization conversion of the test sample to obtain the final tristimulus value w of the test sample.

[0015] Furthermore, the calculation formula of the normalized response value X' of the original format response value of the training sample obtained in step 1 is as follows:

[0016]

[0017] Where, X j Represents the j-channel response value of all training samples, min(Xj ) represents the minimum response value among the j-channel response values ​​of all training samples, max(X j ) represents the maximum response value among the j-channel response values ​​of all training samples, X truth Represents the response value of all training samples after three-channel normalization.

[0018] Furthermore, the specific implementation of step 3 is as follows:

[0019] Step 31: collect a public spectral sensitivity database S, which contains the spectral responses of three channels of multiple cameras, and perform normalization processing on each channel;

[0020] Step 32: Use principal component analysis to calculate the eigenvector Q of each channel in the spectral sensitivity database S. j , where j represents the jth channel;

[0021] Step 33: Combine the spectral data R' of the training sample set, the normalized response value X' and the spectral power distribution L' of the ambient light source, and the characteristic vector Q of each channel obtained by principal component analysis. j , use the pseudo-inverse algorithm to calculate the initial spectral sensitivity S initial ;

[0022] Step 34, defining an optimization function with the goal of optimizing imaging color quality and a non-negative constraint condition to ensure the inherent properties of spectral sensitivity;

[0023] Step 35: Set the initial spectral sensitivity S initial Introduce interior point method for iterative optimization;

[0024] Step 36: The interior point method is iterated until convergence to complete the spectral sensitivity estimation.

[0025] Furthermore, in step 31, a public spectral sensitivity database S (containing the spectral responses of three channels of multiple cameras) is collected, and then S is normalized by channel, as shown in formula (2):

[0026]

[0027] Where S ij represents the spectral response value of channel j of the ith camera in the spectral sensitivity database, min(S ij ) represents the minimum spectral response value in the j channel of the i-th camera, max(S ij ) represents the maximum spectral response value in the j channel of the i-th camera, S i represents the spectral sensitivity of the i-th camera after normalization.

[0028] Furthermore, in step 32, principal component analysis (PCA) is used to extract the principal components of the spectral sensitivity database by channel, and the method is shown in formula (3):

[0029] S j =Q j d j ,j∈{R,G,B}.(3)

[0030] Where S j is the j-channel spectral response of all cameras in the public spectral sensitivity database, Q j is the j-channel eigenvector obtained by PCA, d j is the weight coefficient corresponding to channel j.

[0031] Furthermore, in step 33, the initial spectral sensitivity S is calculated using a pseudo-inverse algorithm. initial , and the method is shown in formula (4):

[0032] S initial =Q[(MQ) T (MQ)] -1 (MQ) T X'(4)

[0033] Where M is the product of the spectral reflectance R' of the training sample and the relative spectral power distribution L' of the light source, Q is the eigenvector obtained by PCA, the superscript 'T' is the transpose symbol, the superscript '-1' indicates the inverse operation, X' is the original format response value of the training sample after normalization, S initial is the initial spectral sensitivity obtained by the initial estimation of the PCA method.

[0034] Furthermore, in step 34, the interior point method objective function and constraints are defined: the objective function mainly includes two parts, where the μ-factor represents the camera's imaging color quality, and the DE color difference represents the color difference between the current spectral sensitivity synthetic response value and the actual shooting response value. Both parts can effectively ensure the color quality of the spectral sensitivity synthetic response value. Spectral sensitivity has the inherent property of being non-negative, so a non-negative constraint is added. The corresponding methods are shown in Equations (5) and (6):

[0035]

[0036] f(s)=μ A (S)+DE (6)

[0037] Where, It represents finding a value of variable s so that the objective function f(s) reaches the minimum value. S is the spectral sensitivity to be optimized. S ≥ 0 represents that the constraint in the optimization process is a non-negative constraint. f(S) is the objective function to be optimized, which is specifically shown in formula (6).A (S) is the μ-factor of the current spectral sensitivity to be optimized. The specific calculation method is shown in formula (7): , where the superscript ‘T’ is the transpose symbol, the superscript ‘-1’ indicates the inverse operation, trace represents the sum of the diagonal of the returned matrix, and A is the CIE1931 standard observer function.

[0038]

[0039] DE is the predicted response value X pre The color difference between the normalized response value X' and the response value X' is obtained by using Equation (8), where S is the spectral sensitivity to be optimized and M is the product of the spectral reflectance of the training sample and the relative spectral power distribution of the light source.

[0040] X pre =MS.(8)

[0041] The color difference calculation method is shown in formula (9), where L truth , a truth , b truth , L pre , a pre , b pre They represent the normalized response value and the chromaticity value of the predicted response value in Lab space respectively.

[0042]

[0043] Among them, to calculate the color difference, the response value needs to be converted from the RGB color space to the XYZ color space, and then converted to the Lab color space. The calculation method is shown in Equations (10) and (11).

[0044]

[0045] Among them, L, a, and b are the brightness, red, green, and yellow and blue color values ​​of the sample in the CIELab color space; X, Y, and Z are the tristimulus color data of the sample; X n 、Y n 、Z n are the tristimulus color data of the reference light source; H, H n are the CIEXYZ tristimulus values ​​of the sample and reference light sources, respectively.

[0046] Furthermore, in step 35, the initial spectral sensitivity S initial The interior point method is used for iterative optimization, as shown in formula (12):

[0047]

[0048] Finally, the initial spectral sensitivity Sinitial The interior point method is used for iterative optimization, and the spectral sensitivity estimation is completed until convergence, and the final spectral sensitivity is obtained.

[0049] Furthermore, in step 4, the spectral data of the training sample set is used to calculate the CIEXYZ tristimulus value W, and the calculation method is shown in formulas (13) and (14):

[0050]

[0051] in,

[0052]

[0053] Where x(λ), y(λ), and z(λ) are the standard observer color matching functions, E(λ) is the spectral reflectance of the object, L(λ) is the relative spectral power distribution function of the light source, λ is the wavelength, η is the adjustment factor, and X, Y, and Z are the tristimulus value data of the training sample set.

[0054] Furthermore, in step 5, the spectral sensitivity of the digital camera is used to predict the camera response value of the training sample set. The calculation method is shown in formula (15):

[0055] D train,pre =R'L"S (15)

[0056] Where R' is the spectral reflectance of the training sample set, L' is the relative spectral power distribution matrix of the ambient light source after diagonal processing, S is the spectral sensitivity of the digital camera, and D train,pre is the predicted response value of the training samples using the spectral sensitivity.

[0057] Furthermore, in steps 5 and 7, the predicted response value of the training sample and the response value of the test sample are expanded using a third-order polynomial to obtain the expanded response value D train,exp , its expanded form is shown in formula (16):

[0058] d exp =(r,g,b,rg,rb,gb,r 2 ,g 2 ,b 2 ,rgb,r 2 g,r 2 b,g 2 r,g 2 b,b 2 r,b 2 g,r 3 ,g 3 ,b 3 ,1)(16)

[0059] Where r, g, and b are the original format response values ​​of the R channel, G channel, and B channel of the sample, respectively, and d exp Represents a vector of original format response values ​​after sample expansion.

[0060] Furthermore, in step 6, the predicted response values ​​and tristimulus values ​​W of the training sample set are used to construct the color characterization conversion matrix Y, and the calculation method is shown in formula (17):

[0061] Y=(D train,exp T D train,exp ) -1 D train,exp T W (17)

[0062] Where D train,exp It is an expanded form of the predicted response value of the training sample, W is the tristimulus value of the training sample, and Y is the color characterization conversion matrix.

[0063] Furthermore, in step 8, the color characterization conversion matrix is ​​used to complete the color characterization conversion of the test sample to obtain the final tristimulus value of the test sample;

[0064] w=D test,exp Y(18)

[0065] Among them, w is the tristimulus value of the converted test sample, D test,exp is the response value vector of the test sample after polynomial expansion.

[0066] The present invention addresses the limitations of existing digital camera color characterization methods and is based on an optimized spectral sensitivity estimation method. First, the spectral sensitivity of a digital camera is estimated using training samples under a known light source environment. Then, the camera response values ​​of the training samples are predicted, and a color characterization conversion matrix between the training sample response values ​​and the three stimuli is constructed. Finally, the color characterization conversion matrix is ​​used to perform color characterization on the test sample. Traditional methods require the use of a digital camera to simultaneously shoot the training samples and test samples under the same light source environment to obtain response values ​​when performing digital camera color characterization. However, in actual applications, the spectral data of the training samples cannot be obtained in any scenario. The method proposed in the present invention effectively solves the problem of difficulty in obtaining training samples in actual applications. By using a response value prediction method instead of a digital camera to obtain the response values ​​of the training samples, the method proposed in the present invention has greater advantages in actual application scenarios compared to other traditional methods, and has great application prospects in aspects such as mobile phone ISP processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 Flowchart of an embodiment of the present invention. DETAILED DESCRIPTION

[0068] When the technical solution of the present invention is specifically implemented, those skilled in the art can use computer software technology to run it.

[0069] like Figure 1 As shown, the present invention provides a digital camera color characterization modeling method based on optimized spectral sensitivity, which specifically includes the following steps:

[0070] Step 1: Use a digital camera to shoot the training sample, obtain the original format response value of the training sample, and perform channel-by-channel normalization processing on it to obtain the normalized response value X' of the original format response value of the training sample;

[0071] In step 1, a Nikon D7200 digital camera is used to photograph the ColorCheck 140 color chart, and the original format response value of the ColorCheck 140 color chart is extracted from the .NEF format image.

[0072] Furthermore, the original format response values ​​of the training samples are normalized by channel. The normalization method is shown in formula (1);

[0073]

[0074] Where, X j Represents the j-channel response value of all training samples, min(X j ) represents the minimum response value among the j-channel response values ​​of all training samples, max(X j ) represents the maximum response value among the j-channel response values ​​of all training samples, and X' represents the response value of all training samples after three-channel normalization.

[0075] Step 2: Use a spectrophotometer and an illuminometer to measure the spectral data R' of the training sample set and the spectral power distribution L' of the ambient light source respectively;

[0076] Step 3, estimating the spectral sensitivity of the digital camera to obtain the camera spectral sensitivity S;

[0077] In step 3, the spectral sensitivity of the digital camera is estimated. The specific method for obtaining the camera spectral sensitivity S is as follows:

[0078] Step 31: First, collect the public spectral sensitivity database S and perform normalization processing on each channel. The spectral sensitivity channel normalization method is shown in formula (2).

[0079]

[0080] Where S ij represents the spectral response value of channel j of the ith camera in the spectral sensitivity database, min(Sij ) represents the minimum spectral response value in the j channel of the i-th camera, max(S ij ) represents the maximum spectral response value in the j channel of the i-th camera, S i represents the spectral sensitivity of the i-th camera after normalization.

[0081] The public database includes the following two spectral sensitivity databases:

[0082] [1] https: / / www.image-engineering.de / library / data-and-tools .

[0083] [2]Tominaga S,Nishi S,Ohtera R.Measurement and estimation of spectralsensitivity functions for mobile phone cameras[J].Sensors,2021,21(15):4985.

[0084] Step 32: Calculate the principal components of the spectral sensitivity database by channel by using principal component analysis (PCA), as shown in formula (3):

[0085] S j =Q j d j ,j∈{R,G,B}.(3)

[0086] Where S j is the j-channel spectral response of all cameras in the public spectral sensitivity database, Q j is the j-channel eigenvector obtained by PCA, d j is the weight coefficient corresponding to channel j.

[0087] Step 33: Calculate the initial spectral sensitivity S using the pseudo-inverse algorithm initial , and the method is shown in formula (4):

[0088] S initial =Q[(MQ) T (MQ)] -1 (MQ) T X'.(4)

[0089] Where M is the product of the spectral reflectance R' of the training sample and the relative spectral power distribution L' of the light source, Q is the eigenvector obtained by PCA, the superscript 'T' is the transpose symbol, the superscript '-1' indicates the inverse operation, X' is the response value of the training sample after normalization, S initial is the initial spectral sensitivity obtained by the initial estimation of the PCA method.

[0090] Step 34 defines an optimization function with the goal of optimizing the imaging color quality and a non-negative constraint condition to ensure the inherent properties of spectral sensitivity. The objective function contains two parts: the imaging color quality of the camera and the color difference between the current spectral sensitivity composite response value and the actual shooting response value. Both parts can effectively ensure the color quality of the spectral sensitivity composite response value. Spectral sensitivity has an inherent non-negative property, so a non-negative constraint is added. The method is shown in Equations (5) and (6):

[0091]

[0092] f(s)=μ A (S)+DE (6)

[0093] Where, It represents finding a value of variable s so that the objective function f(s) reaches the minimum value. S is the spectral sensitivity to be optimized. S ≥ 0 represents that the constraint in the optimization process is a non-negative constraint. f(S) is the objective function to be optimized, which is specifically shown in formula (6). A (S) is the μ-factor of the current spectral sensitivity to be optimized, and the specific calculation method is shown in formula (7): where the superscript ‘T’ is the transpose symbol, the superscript ‘-1’ indicates the inverse operation, trace represents the sum of the diagonal of the returned matrix, and A is the CIE1931 standard observer function.

[0094]

[0095] DE is the predicted response value X pre The color difference between the normalized response value X' and the response value X' is obtained by using Equation (8), where S is the spectral sensitivity to be optimized and M is the product of the spectral reflectance of the training sample and the relative spectral power distribution of the light source.

[0096] X pre =MS.(8)

[0097] The color difference calculation method is shown in formula (9), where L truth , a truth , b truth , L pre , a pre , b pre They represent the normalized response value and the chromaticity value of the predicted response value in Lab space respectively.

[0098]

[0099] Among them, to calculate the color difference, the response value needs to be converted from the RGB color space to the XYZ color space, and then converted to the Lab color space. The calculation method is shown in Equations (10) and (11).

[0100]

[0101]

[0102] Among them, L, a, and b are the brightness, red, green, and yellow and blue color values ​​of the sample in the CIELab color space; X, Y, and Z are the tristimulus color data of the sample; X n 、Y n 、Z n are the tristimulus color data of the reference light source; H, H n are the CIEXYZ tristimulus values ​​of the sample and reference light sources, respectively.

[0103] Step 35: Set the initial spectral sensitivity S initial The interior point method is used for iterative optimization, as shown in formula (12):

[0104]

[0105] The optimization uses the fmincon optimization function provided by matlab2021b. The interior point method is used for optimization by default, and the number of iterations is set to 30.

[0106] Step 36, finally, the initial spectral sensitivity S initial The interior point method is used for iterative optimization, and the spectral sensitivity estimation is completed until convergence, and the final spectral sensitivity is obtained.

[0107] Step 4: Use the spectral data of the training sample set to calculate the corresponding tristimulus value D train,pre ;

[0108] In step 4, the spectral data of the training sample set is used to calculate the CIEXYZ tristimulus value W, and the calculation method is shown in formulas (13) and (14):

[0109]

[0110] in,

[0111]

[0112] Where x(λ), y(λ), and z(λ) are the standard observer color matching functions, E(λ) is the spectral reflectance of the object, L(λ) is the relative spectral power distribution function of the light source, λ is the wavelength, η is the adjustment factor, and X, Y, and Z are the tristimulus value data of the training sample set.

[0113] Step 5: Use the estimated camera spectral sensitivity to predict the camera response value of the training sample set and perform polynomial expansion to obtain the expanded response value D train,exp ;

[0114] In step 5, the estimated camera spectral sensitivity is used to predict the camera response value of the training sample set and perform polynomial expansion to obtain the expanded response value D train,exp , the calculation method is shown in formula (15):

[0115] D train,pre =R'L"S (15)

[0116] Where R' is the spectral reflectance of the training sample set, L' is the relative spectral power distribution matrix of the ambient light source after diagonal processing, S is the spectral sensitivity of the digital camera, and D train,pre is the predicted response value of the training samples using the spectral sensitivity.

[0117] To D train,pre Perform polynomial expansion, and its expanded form is shown in Equation 16:

[0118] d exp =(r,g,b,rg,rb,gb,r 2 ,g 2 ,b 2 ,rgb,r 2 g,r 2 b,g 2 r,g 2 b,b 2 r,b 2 g,r 3 ,g 3 ,b 3 ,1)(16)

[0119] Where r, g, and b are the original format response values ​​of the R channel, G channel, and B channel of the sample, respectively, and d exp Represents a vector of original format response values ​​after sample expansion.

[0120] Step 6: Use the predicted response value D after polynomial expansion of the training sample set train,exp Construct the color characterization conversion matrix Y with the tristimulus values ​​W;

[0121] In step 6, the predicted response value and tristimulus value W of the training sample set are used to construct the color characterization conversion matrix Y. The calculation method is shown in formula (17):

[0122] Y=(D train,exp T D train,exp ) -1 D train,exp T W (17)

[0123] Where D train,expIt is an expanded form of the predicted response value of the training sample, W is the tristimulus value of the training sample, and Y is the color characterization conversion matrix.

[0124] Step 7: Use a camera to shoot the test sample, obtain the test sample response value and perform polynomial expansion D train,exp ;

[0125] In step 7, the ColorCheck24 color chart is used as the test sample and photographed using a Nikon D7200. The original response value in the .NEF format image is extracted and polynomial expansion is performed using formula (16).

[0126] Step 8: Use the color characterization conversion matrix to complete the color characterization conversion of the test sample to obtain the final tristimulus value w of the test sample;

[0127] In step 8, the color characterization conversion matrix is ​​used to complete the color characterization conversion of the test sample to obtain the final tristimulus value of the test sample;

[0128] w=D test,exp Y(18)

[0129] Among them, w is the tristimulus value of the converted test sample, D test,exp is the response value vector of the test sample after polynomial expansion.

[0130] The following is a specific example to illustrate the practical application effect of the present invention:

[0131] In the embodiment, a Nikon D7200 digital camera, the CIED65 standard light source recommended by the International Commission on Illumination, and the internationally used ColorCheck 140 and 24-color color chart were used to form experimental conditions for actual experimental testing of the method of the present invention.

[0132] The experimental evaluation indicators mainly use DE1976 color difference, DE2000 color difference and root mean square error for evaluation, where the root mean square error includes the root mean square error RMSE of the three stimulus values XYZ and Lab root mean square error RMSE Lab , the calculation method is shown in formula (19): where C i and Represent the true value and the converted value after characterization, respectively, and n is the number of color samples.

[0133]

[0134] In the experiment, we used the experimental results (Truth) of shooting a 140-color color chart in a real environment to characterize and correct a 24-color color chart as a comparison. The experimental results are shown in Table 1.

[0135] Table 1 Actual experimental results of color characterization

[0136]

[0137] Experimental results demonstrate the effectiveness of the proposed method for digital camera color characterization. While some aspects of the proposed method still require improvement compared to traditional characterization correction methods that rely on color charts, it demonstrates significant advantages in practical applications. In particular, it eliminates the need for training samples captured in a real environment, which is particularly advantageous when processing complex computational imaging scenarios. Furthermore, the proposed method has broad application prospects in areas such as mobile phone image signal processing (ISP) and multispectral image reconstruction for open environments.

[0138] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0139] The specific embodiments described herein are merely illustrative of the spirit of the present invention. Persons skilled in the art may make various modifications, additions, or substitutions to the described specific embodiments without departing from the spirit of the present invention or exceeding the scope of the appended claims.

Claims

1. A color characterization modeling method for a digital camera based on optimized spectral sensitivity, characterized in that: The steps include: Step 1: Use a digital camera to shoot the training sample, obtain the original format response value of the training sample, and perform channel-by-channel normalization processing on it to obtain the normalized response value X' of the original format response value of the training sample; Step 2: Use a spectrophotometer and an illuminometer to measure the spectral data R' of the training sample set and the spectral power distribution L' of the ambient light source respectively; Step 3, estimating the spectral sensitivity of the digital camera to obtain the camera spectral sensitivity S; Step 4, using the spectral data of the training sample set to calculate the corresponding tristimulus value W; Step 5: Use the estimated camera spectral sensitivity to predict the camera response value of the training sample set and perform polynomial expansion to obtain the expanded response value D train,exp ; Step 6: Use the predicted response value D after polynomial expansion of the training sample set train,exp Construct the color characterization conversion matrix Y with the tristimulus values ​​W; Step 7: Use a camera to shoot the test sample, obtain the test sample response value and perform polynomial expansion D test,exp ; Step 8: Use the color characterization conversion matrix to complete the color characterization conversion of the test sample to obtain the final tristimulus value w of the test sample.

2. The color characterization modeling method for a digital camera based on optimized spectral sensitivity according to claim 1, wherein: The specific implementation of step 3 is as follows: Step 31: collect a public spectral sensitivity database S, which contains the spectral responses of three channels of multiple cameras, and perform normalization processing on each channel; Step 32: Use principal component analysis to calculate the eigenvector Q of each channel in the spectral sensitivity database S. j , where j represents the jth channel; Step 33: Combine the spectral data R' of the training sample set, the normalized response value X' and the spectral power distribution L' of the ambient light source, and the characteristic vector Q of each channel obtained by principal component analysis. j , use the pseudo-inverse algorithm to calculate the initial spectral sensitivity S initial ; Step 34, defining an optimization function with the goal of optimizing imaging color quality and a non-negative constraint condition to ensure the inherent properties of spectral sensitivity; Step 35: Set the initial spectral sensitivity S initial Introduce interior point method for iterative optimization; Step 36: The interior point method is iterated until convergence to complete the spectral sensitivity estimation.

3. The method for color characterization modeling of a digital camera based on optimized spectral sensitivity according to claim 1, wherein: In step 31, the public spectral sensitivity database is normalized by channel, and the method is shown in formula (2): Where S ij represents the spectral response value of channel j of the ith camera in the spectral sensitivity database, min(S ij ) represents the minimum spectral response value in the j channel of the i-th camera, max(S ij ) represents the maximum spectral response value in the j channel of the i-th camera, S i Represents the normalized spectral sensitivity of the i-th camera, where R, G, and B represent the red, green, and blue channels, respectively.

4. The method for color characterization modeling of a digital camera based on optimized spectral sensitivity according to claim 1, wherein: In step 33, the initial spectral sensitivity S is calculated using the pseudo-inverse algorithm. initial , and the method is shown in formula (4): S initial =Q[(MQ) T (MQ)] -1 (MQ) T X' (4) Where M is the product of the spectral reflectance R' of the training sample and the relative spectral power distribution L' of the light source, Q is the eigenvector obtained by PCA, the superscript 'T' is the transpose symbol, the superscript '-1' indicates the inverse operation, X' is the response value of the training sample after normalization, S initial is the initial spectral sensitivity obtained by the initial estimation of the PCA method.

5. The method for color characterization modeling of a digital camera based on optimized spectral sensitivity according to claim 1, wherein: The objective function in step 34 includes two parts: the camera's imaging color quality and the color difference between the current spectral sensitivity synthetic response value and the actual shooting response value. Both parts can effectively guarantee the color quality of the spectral sensitivity synthetic response value. Spectral sensitivity has the inherent property of being non-negative, so a non-negative constraint is added; specifically, as shown in Equations (5) and (6): f(s) = μ A (S)+DE (6) Where, It represents finding a value of variable s so that the objective function f(s) reaches the minimum value. S is the spectral sensitivity to be optimized. S ≥ 0 represents that the constraint in the optimization process is a non-negative constraint. f(S) is the objective function to be optimized, which is specifically shown in formula (6). A (S) is the μ-factor of the current spectral sensitivity to be optimized, and the specific calculation method is shown in formula (7): where the superscript ‘T’ is the transpose symbol, the superscript ‘-1’ indicates the inversion operation, trace represents the sum of the diagonal of the returned matrix, and A is the CIE1931 standard observer function; DE is the predicted response value X pre The color difference between the normalized response value X'.

6. The color characterization modeling method for a digital camera based on optimized spectral sensitivity according to claim 5, wherein: The predicted response value is realized by formula (8), where S is the spectral sensitivity to be optimized, M is the product of the spectral reflectance R' of the training sample and the relative spectral power distribution L' of the light source; X pre =MS(8) The color difference calculation method is shown in formula (9), where L truth , a truth , b truth , L pre , a pre , b pre Represent the chromaticity values ​​of the normalized response value and the predicted response value in Lab space respectively; Among them, to calculate the color difference, the response value needs to be converted from RGB color space to XYZ color space, and then converted to Lab color space. The calculation method is shown in formulas (10) and (11); Among them, L, a, and b are the brightness, red, green, and yellow and blue color values ​​of the sample in the CIELab color space; X, Y, and Z are the tristimulus color data of the sample; X n 、Y n 、Z n are the tristimulus color data of the reference light source; H, H n are the CIEXYZ tristimulus values ​​of the sample and reference light sources, respectively.

7. The method for color characterization modeling of a digital camera based on optimized spectral sensitivity according to claim 1, wherein: In step 4, the spectral data of the training sample set is used to calculate the CIEXYZ tristimulus value W, and the calculation method is shown in formulas (13) and (14): in, Where x(λ), y(λ), and z(λ) are the standard observer color matching functions, E(λ) is the spectral reflectance of the object, L(λ) is the relative spectral power distribution function of the light source, λ is the wavelength, η is the adjustment factor, and X, Y, and Z are the tristimulus value data of the training sample set.

8. The method for color characterization modeling of a digital camera based on optimized spectral sensitivity according to claim 1, wherein: In step 5, the spectral sensitivity of the digital camera is used to predict the camera response value of the training sample set. The calculation method is shown in formula (15): D train,pre =R'L”S(15) Where R' is the spectral reflectance of the training sample set, L' is the relative spectral power distribution matrix of the ambient light source after diagonal processing, S is the spectral sensitivity of the digital camera, and D train,pre is the predicted response value of the training samples using the spectral sensitivity.

9. The method for color characterization modeling of a digital camera based on optimized spectral sensitivity according to claim 1, wherein: In steps 5 and 7, the predicted response value of the training sample and the response value of the test sample are expanded using a third-order polynomial to obtain the expanded response value D train,exp , its expanded form is shown in formula (16): d exp =(r,g,b,rg,rb,gb,r 2 ,g 2 ,b 2 ,rgb,r 2 g,r 2 b,g 2 r,g 2 b,b 2 r,b 2 g,r 3 ,g 3 ,b 3 ,1)(16)where r, g and b are the original format response values ​​of the R channel, G channel and B channel of the sample respectively, d exp Represents a sample expanded original format response value vector; Furthermore, in step 6, the predicted response values ​​and tristimulus values ​​W of the training sample set are used to construct the color characterization conversion matrix Y, and the calculation method is shown in formula (17): Y=(D train,exp T D train,exp ) -1 D train,exp T W(17) Where D train,exp It is an expanded form of the predicted response value of the training sample, W is the tristimulus value of the training sample, and Y is the color characterization conversion matrix.

10. The color characterization modeling method for a digital camera based on optimized spectral sensitivity according to claim 1, wherein: In step 8, the color characterization conversion matrix is ​​used to complete the color characterization conversion of the test sample to obtain the final tristimulus value of the test sample; w=D test,exp Y(18) Among them, w is the tristimulus value of the converted test sample, D test,exp is the response value vector of the test sample after polynomial expansion.

Images