Constraint correction method for non-uniform illumination multispectral imaging data

Abstract

The invention provides a constraint correction method for non-uniform illumination multispectral imaging data. The method comprises the following steps: step 1, introducing imaging data; and 2, removing dark noise. 4, global correction based on spectral prior constraints: the corrected multi-channel spectral image obtained in the step 3 is used as a whole to be input into an illumination non-uniform global correction model with the spectral prior constraints added, global correction is carried out, and corrected multi-spectral imaging data are obtained. The non-uniform illumination multispectral imaging data constraint correction method has the best effect in subsequent cultural relic spectral image classification, and reliable multispectral data can be provided for research and protection analysis of cultural relics. According to the non-uniform illumination multispectral imaging data constraint correction method, the problem of spectrum drift caused by non-uniform illumination can be effectively corrected, and spectrum correlation and consistency among multiple imaging channels are constrained.

Description

Technical Field

[0001] The invention belongs to the technical field of digital image processing, relates to multispectral imaging data, and in particular to a constraint correction method for non-uniform illumination multispectral imaging data. Background Art

[0002] Due to the complex situation at the cultural relics protection site, in the actual spectral imaging process, the quality of multispectral imaging data is affected by the imaging system equipment and the image acquisition environment. The uneven photoelectric response of the imaging system sensor, the influence of the dark current of the equipment under long-term operation, and the variable and uneven lighting of the shooting environment will cause the reliability of multispectral imaging data to decrease, resulting in the spectral feature shift of the same substance in the multispectral imaging data of multi-component lenses, which ultimately affects the accuracy of pigment detection and classification recognition.

[0003] At present, histogram equalization, function adjustment, frequency domain filtering and Retinex theory methods are mostly used for image correction, but they ignore the correlation information between spatial spectra and there is a phenomenon of grayscale information loss to varying degrees.

[0004] In recent years, deep learning models have also been introduced into the field of image illumination enhancement. Due to the uniqueness of cultural relics, different painted cultural relics are unique individuals. There is no universal data set for the multispectral imaging data of painted cultural relics, and data acquisition is very limited, which makes it difficult to fully train and generalize deep learning models. Summary of the invention

[0005] In view of the deficiencies in the prior art, the object of the present invention is to provide a constrained correction method for non-uniform illumination multispectral imaging data to solve the technical problem that the reliability of multispectral imaging data in the prior art needs to be further improved.

[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions to achieve the above problems:

[0007] A constrained correction method for non-uniform illumination multispectral imaging data, the method comprising the following steps:

[0008] Step 1: Import imaging data.

[0009] Step 2: Dark noise removal.

[0010] The image is subjected to environmental dark noise removal according to the environmental dark noise to obtain the original multi-channel spectral image I(x, y).

[0011] Step 3: Regional adaptive non-uniform illumination image correction:

[0012] Step 301: Based on the original multi-channel spectral image I(x, y) obtained in step 2, a multi-scale Retinex model of guided filtering based on local pixel linear relationship is used to calculate the original illumination component L(x, y) estimated by guided filtering.

[0013] Step 302: Based on the original illumination component L(x, y) estimated by the guided filtering obtained in step 301, the image is segmented into regions. According to the brightness characteristics of each region in the image, a dynamically adjusted gamma correction method is used to achieve balanced processing of over-bright areas and over-dark areas, and the illumination component L after adaptive gamma correction is obtained. g (x,y).

[0014] The dynamically adjusted gamma correction method uses a gamma correction parameter γ(x, y).

[0015] The calculation method of the gamma correction parameter γ(x, y) is:

[0016]

[0017] Where:

[0018] μ i is the mean value of the illumination component in the i-th region;

[0019] σ i is the variance of the illumination component in the ith region.

[0020] Step 303: Use the CLAHE algorithm to enhance the adaptive gamma-corrected illumination component L obtained in step 302. g The contrast and detail clarity of (x, y) are used to obtain the enhanced adaptive gamma-corrected illumination component L c (x,y).

[0021] Step 304: The enhanced adaptive gamma-corrected illumination component L obtained in step 303 is c (x, y) and the original illumination component L estimated by the guided filtering g (x, y) is weighted fused to obtain the final illumination correction component L enhancement :

[0022] Step 305: Based on the illumination correction component L enhancement Get the corrected multi-channel spectral image.

[0023] Step 4: Global correction based on spectral prior constraints:

[0024] Step 401 , based on the original multi-channel spectral image I (x, y) obtained in step 2, the standard spectral reflectance of the sampling points is introduced to construct a linear regression model.

[0025] Step 402: The structural features and the spatial domain spectral features are integrated to construct a loss function L.

[0026] The expression of the loss function L is:

[0027]

[0028] Where:

[0029] α1 and α2 are regularization parameters for the spatial structure term and spectral gradient term of the balanced loss function L;

[0030] is the spatial structure term of the loss function L;

[0031] is the spectral gradient term of the loss function L;

[0032] is the 16-dimensional standard spectrum data in the i-th row;

[0033] y i The i-th row of 16-dimensional spectral data output by the linear regression model;

[0034] ‖·‖ F is the Frobenius norm;

[0035] SSIM(·) is the structural similarity calculation;

[0036] is the spectral gradient of the 16-dimensional standard spectral data in the i-th row;

[0037] It is the 16-dimensional spectral gradient of the i-th row output by the linear regression model.

[0038] Step 403 , based on minimization of the loss function L, the linear regression model is optimized to obtain an optimal linear regression model, which is a global correction model for uneven illumination with added spectral prior constraints.

[0039] The global correction model for uneven illumination with added spectral prior constraints is:

[0040]

[0041] Where:

[0042] I(x,y) is the original multi-channel spectral image;

[0043] f c (·) is the CLAHE algorithm operation;

[0044] f gid(·) is the multi-scale guided filtering illumination component estimation operation;

[0045] W is the regression coefficient matrix of size N×N;

[0046] B is an n×N column vector;

[0047] n is the total number of samples;

[0048] N is the total number of channels;

[0049] γ is the gamma correction parameter γ(x,y).

[0050] Step 404 , input the corrected multi-channel spectral image obtained in step 3 as a whole into the uneven illumination global correction model with added spectral prior constraints, perform global correction, and obtain corrected multi-spectral imaging data.

[0051] Compared with the prior art, the present invention has the following technical effects:

[0052] (I) The constrained correction method for non-uniform illumination multispectral imaging data of the present invention is most effective in subsequent classification of cultural relics spectral images and can provide reliable multispectral data for the research and protection analysis of cultural relics.

[0053] (II) The constrained correction method for non-uniform illumination multi-spectral imaging data of the present invention can effectively correct the spectral drift problem caused by uneven illumination and constrain the spectral correlation and consistency between multiple imaging channels.

[0054] (III) In the present invention, the a priori supported multi-channel global correction adds spatial structure and spectral gradient constraints on the basis of non-uniform illumination correction, ensuring that the correlation and consistency of the original spectral features are maintained during the illumination change correction process, thereby maximally retaining the integrity of the multi-channel spectral data during the illumination unevenness correction process, ensuring that the corrected spectral data accurately reflects the true reflectance characteristics of the material.

[0055] (IV) In the present invention, the regional adaptive two-dimensional gamma correction method can effectively improve the local brightness distortion caused by non-uniform illumination and enhance the overall contrast and detail expression of the image. At the same time, in view of the requirement for accurate classification of spectral images, a regression correction model based on spectral prior constraints is constructed, and a loss function is constructed by combining spatial structural features and spectral prior features to globally constrain the original multi-channel spectral image, so as to achieve the purpose of maintaining the spectral correlation and consistency between multiple imaging channels while correcting the spectral information offset of the multi-channel image.

[0056] (V) The method of the present invention can effectively balance the image brightness and suppress the deviation of the image spectral information, which is beneficial to improving the accuracy of subsequent pigment classification and identification and maintaining the structural information of the cultural relics. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is the result of simulated mural correction.

[0058] Figure 2 It is a comparison of the classification results before and after the simulation mural correction.

[0059] Figure 3 This is the comparison of classification results before and after the real mural correction.

[0060] The specific contents of the present invention are further described in detail below in conjunction with the accompanying drawings and embodiments. DETAILED DESCRIPTION

[0061] It should be noted that, unless otherwise specified, all algorithms and methods in the present invention adopt algorithms and methods known in the prior art.

[0062] Based on the current situation introduced in the background technology, the present invention aims to solve the problem of spectral deviation caused by non-uniform illumination in the multispectral imaging process of painted cultural relics. It is necessary to adopt an adaptive multispectral image correction method combined with prior feature constraints to obtain a spectrally corrected image for subsequent research.

[0063] In accordance with the above technical scheme, specific embodiments of the present invention are given below. It should be noted that the present invention is not limited to the following specific embodiments, and all equivalent changes made on the basis of the technical scheme of this application fall within the protection scope of the present invention.

[0064] Example:

[0065] This embodiment provides a constraint correction method for non-uniform illumination multi-spectral imaging data, which includes the following steps.

[0066] Step 1: Import imaging data:

[0067] Multispectral images of cultural relics are collected, and the multispectral images are registered and cropped to obtain images of murals with non-uniform illumination; dark frame images under no-light conditions are introduced as reference data.

[0068] In step one, the differences in optical transmittance of various parts of the optical elements in the imaging device, the differences in the sensor's response to the incident light, and the differences in the spectral distribution of the ambient light source will directly affect the radiation intensity distribution of the incident light reaching the photoreceptor surface, causing dark current noise spectral errors on the resulting image.

[0069] Step 2: Dark noise removal:

[0070] The ambient dark noise is estimated using the laboratory calibration method according to the dark frame image introduced in step 1, and the ambient dark noise is removed from the non-uniformly illuminated mural image obtained in step 1 according to the ambient dark noise to obtain the original multi-channel spectral image I(x,y).

[0071] In step 2, in order to correct the dark noise problem caused by the characteristics of the imaging system equipment, a dark frame image under no light conditions is introduced, and the dark noise correction is performed using the laboratory calibration method to eliminate the random noise that exists in a short period of time, and finally obtain the average dark current value of the imaging device in the absence of external light sources.

[0072] In step 2, the dark frame output image of the imaging device that continuously acquires images for a long time is averaged, and the measured dark current value is deducted during the data processing process. This reduces the impact of dark current on image quality and reduces image noise caused by dark current non-uniformity. The calculation method for removing environmental dark noise is:

[0073]

[0074] Where:

[0075] DN (i,j,λ) is the response value of the camera in the i-th row and j-th column after removing the dark noise in the spectral channel with the central wavelength λ;

[0076] is the response value of the camera in the i-th row and j-th column in the spectral channel with the central wavelength λ;

[0077] DN andianliu(i,j,λ,k) is the dark current value of the pixel in the i-th row and j-th column in the spectral channel with the central wavelength λ at the k-th time point;

[0078] N is the total number of observation time points.

[0079] Step 3: Regional adaptive non-uniform illumination image correction:

[0080] Step three specifically includes the following steps:

[0081] Step 301: Based on the original multi-channel spectral image I(x, y) obtained in step 2, a multi-scale Retinex model of guided filtering based on local pixel linear relationship is used to calculate the original illumination component L(x, y) estimated by guided filtering.

[0082] In step 301, the multi-scale Retinex model is:

[0083]

[0084] Where:

[0085] L(x,y) is the original illumination component estimated by guided filtering;

[0086] i is the filtering scale;

[0087] N is the number of filter scales,

[0088] w i The weights of different scales are all 1 / 3;

[0089] I(x,y) is the guiding image;

[0090] q i (x,y) is the guided filter function.

[0091] Step 302, based on the original illumination component L(x,y) estimated by the guided filtering obtained in step 301, the image is segmented into regions, such as Figure 1 According to the brightness characteristics of each area in the image, a dynamically adjusted gamma correction method is used to achieve balanced processing of over-bright areas and over-dark areas, and the adaptive gamma-corrected illumination component L is obtained. g (x,y).

[0092] In step 302, the dynamically adjusted gamma correction method is:

[0093]

[0094] Where:

[0095] L g (x, y) is the illumination component after adaptive gamma correction;

[0096] L(x,y) is the original illumination component estimated by guided filtering;

[0097] γ(x,y) is the gamma correction parameter.

[0098] In step 302, the gamma correction parameters are calculated as follows:

[0099]

[0100] Where:

[0101] μ i is the mean value of the illumination component in the i-th region;

[0102] σ i is the variance of the illumination component in the ith region.

[0103] In this embodiment, the mean and variance characteristics of the region are introduced to calculate the gamma correction parameter γ(x, y); the gamma correction parameter γ(x, y) controls the intensity of brightness correction, and its value is crucial to the effect of adjusting the brightness of the image region.

[0104] Step 303: Use the CLAHE algorithm to enhance the adaptive gamma-corrected illumination component L obtained in step 302. gThe contrast and detail clarity of (x, y) are further improved to further enhance the brightness of the illumination component and maintain the texture details, thus obtaining the enhanced adaptive gamma-corrected illumination component L c (x, y).

[0105] In step 303, the calculation formula enhanced by the CLAHE algorithm is:

[0106] L c (x, y) = f c (L g (x, y)

[0107] Where:

[0108] L c (x, y) is the enhanced illumination component after adaptive gamma correction;

[0109] L g (x, y) is the illumination component after adaptive gamma correction;

[0110] f c (·) is the CLAHE algorithm operation.

[0111] Step 304: The enhanced adaptive gamma-corrected illumination component L obtained in step 303 is c (x, y) and the original illumination component L estimated by guided filtering g (x, y) is weighted fused to obtain the final illumination correction component L enhancement :

[0112] In step 304, the calculation method of weighted fusion is:

[0113] L enhancement =mL(x,y)+(1-m)L c (x, y).

[0114] Where:

[0115] m is the weight value, ranging from 0 to 1.

[0116] In step 304, an optimal weight value of m is determined by experimental trial and error. The experimental trial and error method is: taking values ​​one by one within the range at intervals of 0.1, comparing the spectral distance between the corrected spectrum and the standard spectral reflectance each time, and taking the weight value with the smallest spectral distance as the optimal weight that meets the correction model in this scenario.

[0117] Step 305: Based on the illumination correction component L enhancement Get the corrected multi-channel spectral image.

[0118] Step 4: Global correction based on spectral prior constraints:

[0119] In step 4, a linear regression model based on prior spectral constraints is introduced to deal with the gray value loss problem caused by single-channel processing, and the correction coefficient and bias term are obtained by minimizing the objective function to correct the multi-channel spectral image. Standard spectral data is used as prior knowledge to provide the regression model with correlation information between spectral channels, and the multi-channel spectral information is input into the regression model as a complete whole, so that the model can simultaneously characterize the spectral channel characteristics and spatial structure characteristics, and maintain the overall information of the multi-channel spectrum while performing regional adaptive correction.

[0120] Step 401 , based on the original multi-channel spectral image I (x, y) obtained in step 2, the standard spectral reflectance of the sampling points is introduced to construct a linear regression model.

[0121] In step 401, the linear regression model is expressed as: Y=XW+B.

[0122] Where:

[0123] Y is an n×N output matrix, and Y represents all output values ​​of all samples.

[0124] The input matrix X is n×N, which is the multi-channel spectral data after black and white plate correction;

[0125] W is a regression coefficient matrix of size N×N; each column w in W j represents the regression coefficient of the j-th output variable;

[0126] B is an n×N column vector, B represents the bias error, and each row is the same bias vector β j ;

[0127] n is the total number of samples;

[0128] N is the total number of channels;

[0129] j is the sequence of channels, j=1,2,…,N.

[0130] In step 401, in the linear regression model, the jth output value y of the i-th sample ij It is a linear combination of the N input features of the sample, plus the bias vector β j The result is:

[0131] y ij =x i1 w 1j +x i2 w 2j +…+x iN w Nj +β j; i is the sequence of samples, i = 1, 2,…, n; for ease of calculation, the linear regression model is expressed as: Y = XW + B.

[0132] Step 402: The structural features and the spatial domain spectral features are integrated to construct a loss function L.

[0133] The expression of the loss function L is:

[0134]

[0135] Where:

[0136] α1 and α2 are regularization parameters for the spatial structure term and spectral gradient term of the balanced loss function L;

[0137] is the spatial structure term of the loss function L;

[0138] is the spectral gradient term of the loss function L;

[0139] is the 16-dimensional standard spectrum data in the i-th row;

[0140] y i The i-th row of 16-dimensional spectral data output by the linear regression model;

[0141] ‖·‖ F is the Frobenius norm;

[0142] SSIM(·) is the structural similarity calculation;

[0143] is the spectral gradient of the 16-dimensional standard spectral data in the i-th row;

[0144] It is the 16-dimensional spectral gradient of the i-th row output by the linear regression model.

[0145] In step 402, the construction process of the loss function L is:

[0146] Step 40201, using the spectral structure loss function l structure To measure the difference between the multispectral image output by the linear regression model and the standard uniform illumination multispectral image.

[0147] Spectral structure loss function l structure for:

[0148]

[0149] Step 40202, adding a gradient information constraint that reflects the change trend between adjacent wavelengths, adding a structural perception smoothness constraint on the multispectral image, and obtaining the spectral gradient loss function l gradient for:

[0150]

[0151] Step 40203, in order to simultaneously maintain the spatial structure information and suppress spectral distortion, a loss function L is defined that fuses the structural features with the spatial domain spectral features.

[0152] Step 403 , based on minimization of the loss function L, the linear regression model is optimized to obtain an optimal linear regression model, which is a global correction model for uneven illumination with added spectral prior constraints.

[0153] The global correction model for uneven illumination with added spectral prior constraints is:

[0154]

[0155] Where:

[0156] I(x,y) is the original multi-channel spectral image;

[0157] f c (·) is the CLAHE algorithm operation;

[0158] f gid (·) is the multi-scale guided filtering illumination component estimation operation;

[0159] W is the regression coefficient matrix of size N×N;

[0160] B is an n×N column vector;

[0161] n is the total number of samples;

[0162] N is the total number of channels;

[0163] γ is the gamma correction parameter γ(x,y).

[0164] In step 403, the spatial structure term and the spectral gradient term of the loss function are adjusted to guide the regression model to maintain specific standard spectral data structure characteristics, and the regression coefficient matrix and the bias term are calculated by minimizing the loss function.

[0165] In step 403, the multi-channel global correction of spectral prior constraints adds spatial structure and spectral gradient constraints on the basis of non-uniform illumination correction, ensuring that the correlation and consistency of the original spectral characteristics are maintained during the illumination change correction process, thereby maximally retaining the integrity of the multi-channel spectral data during the correction of uneven illumination, ensuring that the corrected spectral data accurately reflects the true reflectance characteristics of the material.

[0166] Step 404 , taking the corrected multi-channel spectral image obtained in step 3 as the overall input, adding a global correction model of uneven illumination with spectral prior constraints, and performing global correction to obtain corrected multi-spectral imaging data.

[0167] Application examples:

[0168] This application example is based on a constrained correction method for non-uniform illumination multi-spectral imaging data given in the above-mentioned embodiment. Specifically, in order to verify the application effect of the constrained correction method for non-uniform illumination multi-spectral imaging data proposed in the present invention in actual situations, application effect tests were carried out on simulated murals and the mixed pigment murals of the Thirteenth Venerable of Dule Temple.

[0169] The regional adaptive two-dimensional gamma correction method is used to balance the non-uniform illumination area, and then the CLAHE algorithm is used to improve the overall contrast and detail expression of the image. The original multi-channel spectral image is globally constrained by the regression correction model based on spectral prior constraints to ensure that the relevance and consistency of the original spectral features are maintained during the correction process of illumination changes.

[0170] The results obtained by the method of the present invention are as follows: Figure 2 As shown, for the non-uniformity correction of the simulated mural scene, the method of the present invention still maintains a good illumination correction effect and the global correction method based on the prior spectral feature constraints can better restore the spectral features of the original channel.

[0171] The verification effect of mixed pigments on real murals and cultural relics is as follows Figure 3 As shown in the figure, even in the mixed pigment area, the edges of the corrected classification results are clearer, and the similarity between the corrected spectral data and the standard reflectance is significantly improved, the distinction between categories is significantly enhanced, and the overall classification visual effect is natural. In general, the constrained correction method for non-uniform illumination multispectral imaging data has a good effect.

Claims

1. A constrained correction method for non-uniform illumination multispectral imaging data, characterized in that: The method comprises the following steps: Step 1: Import imaging data; Step 2: dark noise removal; According to the environmental dark noise, the image is subjected to environmental dark noise removal to obtain the original multi-channel spectral image I(x,y); Step 3: Regional adaptive non-uniform illumination image correction: Step 301, according to the original multi-channel spectral image I (x, y) obtained in step 2, a multi-scale Retinex model of guided filtering based on local pixel linear relationship is used to calculate the original illumination component L (x, y) estimated by guided filtering; Step 302: Based on the original illumination component L(x, y) estimated by the guided filtering obtained in step 301, the image is segmented into regions. According to the brightness characteristics of each region in the image, a dynamically adjusted gamma correction method is used to achieve balanced processing of over-bright areas and over-dark areas, and the illumination component L after adaptive gamma correction is obtained. g (x,y); The gamma correction parameter γ(x,y) is used in the dynamically adjusted gamma correction method; The calculation method of the gamma correction parameter γ(x, y) is: Where: μ i is the mean value of the illumination component in the i-th region; σ i is the variance of the illumination component in the i-th region; Step 303: Use the CLAHE algorithm to enhance the adaptive gamma-corrected illumination component L obtained in step 302. g The contrast and detail clarity of (x, y) are used to obtain the enhanced adaptive gamma-corrected illumination component L c (x,y); Step 304: The enhanced adaptive gamma-corrected illumination component L obtained in step 303 is c (x, y) and the original illumination component L estimated by the guided filtering g (x, y) is weighted fused to obtain the final illumination correction component L enhancement : Step 305: Based on the illumination correction component L enhancement Obtain a corrected multi-channel spectral image; Step 4: Global correction based on spectral prior constraints: Step 401, based on the original multi-channel spectral image I (x, y) obtained in step 2, the standard spectral reflectance of the sampling points is introduced to construct a linear regression model; Step 402, integrating the structural features and the spatial domain spectral features to construct a loss function L; The expression of the loss function L is: Where: α1 and α2 are regularization parameters for the spatial structure term and spectral gradient term of the balanced loss function L; is the spatial structure term of the loss function L; is the spectral gradient term of the loss function L; is the 16-dimensional standard spectrum data in the i-th row; y i The i-th row of 16-dimensional spectral data output by the linear regression model; ‖·‖ F is the Frobenius norm; SSIM(·) is the structural similarity calculation; is the spectral gradient of the 16-dimensional standard spectral data in the i-th row; The i-th row of 16-dimensional spectral gradient output by the linear regression model; Step 403, based on minimization of the loss function L, the linear regression model is optimized to obtain an optimal linear regression model, that is, a global correction model for uneven illumination with added spectral prior constraints; The global correction model for uneven illumination with added spectral prior constraints is: Where: I(x,y) is the original multi-channel spectral image; f c (·) is the CLAHE algorithm operation; f gid (·) is the multi-scale guided filtering illumination component estimation operation; W is the regression coefficient matrix of size N×N; B is an n×N column vector; n is the total number of samples; N is the total number of channels; γ is the gamma correction parameter γ(x,y); Step 404 , input the corrected multi-channel spectral image obtained in step 3 as a whole into the uneven illumination global correction model with added spectral prior constraints, perform global correction, and obtain corrected multi-spectral imaging data.

2. The constrained correction method for non-uniform illumination multispectral imaging data according to claim 1, characterized in that: The specific process of step one is: collect multispectral images of cultural relics, align and crop the multispectral images to obtain non-uniformly illuminated mural images; introduce dark frame images under no-light conditions as reference data.

3. The constrained correction method for non-uniform illumination multispectral imaging data according to claim 2, characterized in that: The ambient dark noise is estimated using the laboratory calibration method according to the dark frame image introduced in step 1, and the ambient dark noise is removed from the non-uniformly illuminated mural image obtained in step 1 according to the ambient dark noise to obtain the original multi-channel spectral image I(x, y).

4. The constrained correction method for non-uniform illumination multispectral imaging data according to claim 1, characterized in that: In step 2, the calculation method for removing the ambient dark noise is: Where: DN (i,j,λ) is the response value of the camera in the i-th row and j-th column after removing the dark noise in the spectral channel with the central wavelength λ; DH raw(i,j,λ) is the response value of the camera in the i-th row and j-th column in the spectral channel with the central wavelength λ; DN andianliu(i,j,λ,k) is the dark current value of the pixel in the i-th row and j-th column in the spectral channel with the central wavelength λ at the k-th time point; N is the total number of observation time points.

5. The constrained correction method for non-uniform illumination multispectral imaging data according to claim 1, characterized in that: In step 301, the multi-scale Retinex model is: Where: L(x,y) is the original illumination component estimated by guided filtering; i is the filtering scale; N is the number of filter scales, w i The weights of different scales are all 1 / 3; I(x,y) is the guiding image; q i (x,y) is the guided filter function.

6. The constrained correction method for non-uniform illumination multispectral imaging data according to claim 1, characterized in that: In step 302, the dynamically adjusted gamma correction method is: Where: L g (x, y) is the illumination component after adaptive gamma correction; L(x,y) is the original illumination component estimated by guided filtering; γ(x,y) is the gamma correction parameter.

7. The constrained correction method for non-uniform illumination multispectral imaging data according to claim 1, characterized in that: In step 303, the calculation formula enhanced by the CLAHE algorithm is: L c (x,y)=f c (L g (x,y)) Where: L c (x, y) is the enhanced illumination component after adaptive gamma correction; L g (x, y) is the illumination component after adaptive gamma correction; f c (·) is the CLAHE algorithm operation.

8. The constrained correction method for non-uniform illumination multispectral imaging data according to claim 1, characterized in that: In step 304, the weighted fusion calculation method is: L enhancement =mL(x,y)+(1-m)L c (x,y); Where: m is the weight value, ranging from 0 to 1.

9. The constrained correction method for non-uniform illumination multispectral imaging data according to claim 1, characterized in that: In step 401, the linear regression model is expressed as: Y=XW+B; Where: Y is the n×N output matrix, and Y represents all output values ​​of all samples; The input matrix X is n×N, which is the multi-channel spectral data after black and white plate correction; W is a regression coefficient matrix of size N×N; each column w in W j represents the regression coefficient of the j-th output variable; B is an n×N column vector, B represents the bias error, and each row is the same bias vector β j ; n is the total number of samples; N is the total number of channels; j is the sequence of channels, j=1,2,…,N.

Images