Method for correcting color measurement result of colorimeter

Abstract

The invention relates to the technical field of color difference correction, in particular to a method for correcting a color measurement result of a colorimeter, which comprises the following steps of: acquiring a sample Lab value, calculating a coordinate index correction matrix, acquiring an original coefficient, inputting segmented regression, extracting characteristics, calculating deviation weighting, and acquiring a regional offset; the method comprises the following steps: acquiring a neighborhood compensation channel offset, calling an offset to adaptively correct the neighborhood compensation channel offset to obtain a boundary front value, performing boundary verification limiting amplitude callback processing to obtain an output result, and updating a knowledge base to accumulate deviation incremental counting to optimize the color measurement precision, in the invention, through regionalization analysis and a piecewise linear regression model, the accuracy and flexibility of color difference correction are improved, and the color measurement accuracy is improved. Through the weighting processing and neighborhood compensation technology, correction inaccuracy caused by large errors is effectively reduced, it is ensured that the corrected measurement value does not exceed a reasonable range, an unstable correction result of a traditional method is avoided, adaptability is high, fault tolerance is high, measurement consistency and reliability of the colorimeter are improved, and the method is suitable for popularization and application. The method has obvious advantages especially in a complex chromatic aberration correction scene.

Description

Technical Field

[0001] The present invention relates to the technical field of color difference correction, and in particular to a method for correcting color measurement results of a colorimeter. Background Art

[0002] The technical field of color difference correction involves technologies for detecting and adjusting the difference between the surface color of the measured object and the standard color. Core issues include obtaining color measurement data through a colorimeter, expressing the measurement values ​​using a specific color space, comparing the differences between the sample and the standard in multiple color channels, and analyzing and correcting the measured data based on the comparison results. This technical field systematically covers the principles of optical detection, the application of color space models, the measurement data acquisition process, and the correction method of color difference data.

[0003] Among them, the correction method of the color measurement results of traditional colorimeters refers to correcting the deviation in the measurement results through manual comparison or lookup based on a preset value table after the colorimeter completes the color measurement. It uses multiple repeated measurements under specific light source conditions and manually adjusts the color channel readings based on manually selected reference samples, or uses the least squares method to calculate the linear relationship between the output values ​​of the three channels of the sensor and the human eye perception value to achieve the conversion from sensor readings to color space values. However, after this type of conversion, there will still be large nonlinear error terms.

[0004] Existing technologies rely on manual comparisons and preset value tables for correction, which are susceptible to human interference and have low processing efficiency. Traditional linear regression models fail to effectively address nonlinear errors in color channels, resulting in limited correction effectiveness. The lack of detailed analysis and dynamic adjustment of different measurement areas makes it difficult to flexibly correct based on changes in measured values, which in turn affects the accuracy of the correction results. In high-precision environments, existing technologies cannot meet the changing needs of color correction, resulting in unstable correction results and difficulty achieving ideal results in complex environments. Summary of the Invention

[0005] In order to solve the problem that the existing technology relies on manual comparison and preset numerical tables for correction, which is easily interfered by human factors and has low processing efficiency. The traditional linear regression model fails to effectively deal with nonlinear errors in the color channel, resulting in limited correction effects. The lack of detailed analysis and dynamic adjustment of different measurement areas makes it difficult to make flexible corrections based on changes in measured values, thereby affecting the accuracy of the correction results. In an environment with high precision requirements, the existing technology cannot meet the changing color difference correction needs, resulting in unstable correction effects and difficulty in achieving ideal results in complex environments. The embodiment of the present invention provides a method for correcting the color measurement results of a colorimeter, comprising the following steps: In order to achieve the above object, the present invention adopts the following technical solution: a method for correcting the color measurement results of a colorimeter, comprising the following steps: S1: Obtain the Lab color space value of the sample to be corrected through a colorimeter, calculate the horizontal and vertical coordinates of the grid according to the a channel value and the b channel value, index the pre-constructed L, a, and b correction matrix to analyze the original correction coefficient, and obtain the Lab measurement value to be corrected; S2: Perform regional analysis based on the Lab measurement value to be corrected, input the Lab measurement value to be corrected into a piecewise linear regression model to extract regional features, calculate the deviation between the current measured Lab value and the original standard value in the corresponding grid, and perform weighted processing based on the number of measurements of the grid in the counting matrix to obtain the regional correction offset; S3: Adaptively correcting the measurement results by calling the regionalized correction offset, performing nonlinear weight allocation according to the regionalized correction offset, obtaining correction parameters by using a neighborhood compensation technique when the grid value corresponding to the counting matrix is ​​zero, performing channel offset correction, and obtaining a correction value before boundary verification; S4: Perform color gamut boundary constraint verification processing using the correction value before boundary verification. When the corrected ab value exceeds the current grid boundary, limit the correction amplitude to within the grid boundary range, and perform callback processing on the out-of-bounds value to obtain the output result after boundary verification.

[0006] As a further solution of the present invention, the Lab measurement value to be corrected includes a brightness component, a red-green component, and a yellow-blue component, the regionalized correction offset includes a local difference value, a regional weighted value, and a feature mapping value, the correction value before boundary verification includes a channel offset, a compensation coefficient, and an allocation weight, and the output result after boundary verification includes a boundary constraint value, an amplitude limit value, and a callback correction value.

[0007] As a further solution of the present invention, the specific steps of S1 are: S101: Obtain the Lab color space value of the sample measured by the colorimeter, call the a channel value and the b channel value, calculate the horizontal and vertical coordinate positions of the two in the coordinate plane, and correspond the horizontal and vertical coordinates with the L channel value of the sample to generate a coordinate positioning value; S102: Based on the coordinate positioning value, call the pre-built L, a, and b correction matrices, compare the data boundaries of the cells in the matrix item by item according to the horizontal and vertical coordinate positions, filter the corresponding matrix cell data, and compare the filtered results with the L channel values ​​item by item and unify the format to obtain the matrix index value; S103: According to the matrix index value, for the original correction coefficient in the correction matrix, call the Lab color space value of the sample to perform weighted calculation item by item, combine the correction coefficient after weighted calculation with the original value of the sample to generate the Lab measurement value to be corrected.

[0008] As a further solution of the present invention, the specific steps of S2 are: S201: Based on the Lab measurement values ​​to be corrected, the measurement values ​​are sequentially input into a piecewise linear regression model, the values ​​are fitted according to the piecewise function coefficients within the interval, corresponding parameters are extracted according to the slope and intercept of the fitting curve in the differentiation interval, and regional segmental characteristic coefficients are generated; The regional segmentation characteristic coefficient refers to the slope and intercept obtained by piecewise linear regression fitting, which is used to characterize the corresponding relationship between the measured value and the standard value within the differentiation interval; S202: calling the regional segment characteristic coefficients and comparing them with the original standard values ​​in the grid one by one, summarizing the results using a difference accumulation method based on the difference between the current measured Lab value and the standard value, and summarizing the accumulated differences into a calculable indicator to obtain a grid difference value; The grid difference refers to the result of accumulating the difference between the measured value and the standard value, which is used to reflect the overall deviation degree within a single grid; S203: According to the grid difference amount, a weight is set in combination with the number of measurements corresponding to the grid in the counting matrix, the difference amount is weightedly calculated using the weight, and the weighted result is converted into an overall correction value to obtain a regionalized correction offset.

[0009] As a further solution of the present invention, the specific steps of S3 are: S301: performing point-by-point comparison of the measurement results based on the regionalized correction offset, performing difference calculation between the measurement matrix grid point values ​​and the regionalized correction offset, extracting values ​​according to the position coordinates and weightedly accumulating them in the difference calculation to generate a difference offset coefficient; S302: Calling the difference offset coefficient to perform nonlinear weight allocation, compensating cells in the count matrix with zero grid points according to the values ​​of adjacent non-zero grid points in the weight allocation, superimposing the compensated value with the original allocated value and performing normalization processing to obtain a compensated weight value; S303: calling the compensation weight value to perform channel offset correction, superimposing the compensation weight and the channel offset parameter one by one in the correction, and redistributing the channel offset parameter in the superposition matrix to obtain the correction value before boundary verification.

[0010] As a further solution of the present invention, the difference offset coefficient refers to a difference quantization coefficient obtained by weighted accumulation based on a point-by-point comparison of the measurement matrix grid point value and the regionalized correction offset; The normalization process adopts a proportional normalization method, that is, the compensation weight value is divided by the sum of the compensation weight values; The compensation weight value refers to a weight parameter obtained after performing neighboring value compensation and normalization processing on the zero-value grid point in nonlinear weight distribution.

[0011] As a further solution of the present invention, the specific steps of S4 are: S401: Based on the pre-boundary verification correction value and the current grid boundary range, determine whether the corrected ab value exceeds the boundary, compare the corrected ab value with the upper and lower limits of the grid boundary, mark the out-of-bounds data points, extract their corresponding coordinate information, and generate an out-of-bounds coordinate set; S402: Calling the grid position coordinates in the out-of-bounds coordinate set, re-limiting the ab value that exceeds the boundary according to the upper and lower limits of the boundary, adjusting the difference of the out-of-bounds part to the critical position, and recalculating the limited ab value distribution range to obtain the boundary interception value interval; S403: According to the correction result in the boundary interception value interval, the ab values ​​within the boundary and the critical point are integrated, and the adjusted data and the data that do not cross the boundary are merged to obtain the output result after boundary verification.

[0012] As a further embodiment of the present invention, the method further comprises step S5: S5: Based on the output result after the boundary verification, the correction knowledge base is updated and maintained, the newly added Lab deviation is calculated and accumulated to the Lab correction matrix of the corresponding grid coordinate, and the value of the corresponding position of the count matrix is ​​incremented to obtain the colorimeter color measurement accuracy optimization result; The colorimeter color measurement accuracy optimization result includes the deviation accumulation amount, the correction matrix increment and the counting matrix update value.

[0013] As a further solution of the present invention, the specific steps of S5 are: S501: Based on the output result after the boundary verification, obtain the value of the corresponding color block in the correction knowledge base, detect the Lab deviation of the newly added measurement point, calculate the difference between the newly added deviation value and the existing reference Lab value, record the deviation value according to the difference, update the correction knowledge base, and generate a newly added Lab deviation value; S502: calling the newly added Lab deviation value, updating the Lab correction matrix for the corresponding grid coordinate in the correction knowledge base, and adding the deviation value to the value of the corresponding position in the matrix to generate a grid coordinate cumulative correction value; S503: According to the accumulated correction value of the grid coordinate, the corresponding coordinate point is searched in the counting matrix, the counting value is incremented, and after the counting matrix is ​​updated, the correction matrix and the counting matrix are stored together to generate the colorimeter color measurement accuracy optimization result.

[0014] As a further solution of the present invention, the newly added Lab deviation value refers to the difference between the newly added measurement point and the reference Lab value in the revised knowledge base; The lattice coordinate cumulative correction value refers to the accumulation of the original Lab deviation values ​​of the corresponding lattice coordinates in the correction matrix.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are: This method improves the accuracy and flexibility of color correction through regional analysis and a piecewise linear regression model. Weighted processing and neighborhood compensation techniques effectively reduce correction inaccuracies caused by large errors and ensure that corrected measurements remain within a reasonable range, avoiding the unstable correction results seen with traditional methods. Furthermore, the method is highly adaptable to diverse light source conditions and highly tolerant to changes in the measurement environment, improving the colorimeter's measurement consistency and reliability, offering significant advantages in complex color correction scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 Schematic diagram of the steps of the present invention; Figure 2 This is a schematic diagram of the refinement of S1 of the present invention; Figure 3 This is a schematic diagram of the refinement of S2 of the present invention; Figure 4 This is a schematic diagram of the refinement of S3 of the present invention; Figure 5 This is a schematic diagram of the refinement of S4 of the present invention; Figure 6 This is a schematic diagram of the refinement of S5 of the present invention. DETAILED DESCRIPTION

[0018] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0019] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0020] In the embodiments of the present invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same. The terms "of," "corresponding," and "corresponding" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same.

[0021] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0022] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0023] See also Figure 1 , an embodiment of the present invention provides a method for correcting color measurement results of a colorimeter, comprising the following steps: S1: Obtain the Lab color space value of the sample to be corrected through a colorimeter, calculate the horizontal and vertical coordinates of the grid according to the a channel value and the b channel value, index the pre-constructed L, a, and b correction matrix to analyze the original correction coefficient, and obtain the Lab measurement value to be corrected; S2: Perform regional analysis based on the Lab measurement value to be corrected. Input the Lab measurement value to be corrected into the piecewise linear regression model to extract regional features. Calculate the deviation between the current measured Lab value and the original standard value in the corresponding grid. Combined with the number of measurements of the grid in the counting matrix, perform weighted processing to obtain the regional correction offset. S3: Call the regionalized correction offset to perform adaptive correction processing on the measurement results, perform nonlinear weight distribution according to the regionalized correction offset, use the neighborhood compensation technology to obtain the correction parameters when the corresponding grid value of the counting matrix is ​​zero, perform channel offset correction, and obtain the correction value before boundary verification; S4: Performing color gamut boundary constraint verification using the correction value before boundary verification. When the corrected ab value exceeds the current grid boundary, the correction amplitude is limited to the grid boundary range, and a callback is performed on the out-of-bounds value to obtain the output result after boundary verification. S5: Based on the output results after boundary verification, the correction knowledge base is updated and maintained, the newly added Lab deviation is calculated and accumulated to the Lab correction matrix of the corresponding grid coordinate, and the corresponding position value of the count matrix is ​​incremented to obtain the colorimeter color measurement accuracy optimization result; The Lab measurement values ​​to be corrected include the brightness component, red-green component, and yellow-blue component. The regionalized correction offset includes the local difference value, regional weighted value, and feature mapping value. The correction value before boundary verification includes the channel offset, compensation coefficient, and allocation weight. The output result after boundary verification includes the boundary constraint value, amplitude limit value, and callback correction value. The colorimeter color measurement accuracy optimization result includes the deviation accumulation value, correction matrix increment, and count matrix update value.

[0024] See also Figure 2 , the specific steps of S1 are: S101: Obtain the Lab color space value of the sample measured by the colorimeter, call the a channel value and the b channel value, calculate the horizontal and vertical coordinate positions of the two in the coordinate plane, and correspond the horizontal and vertical coordinates with the L channel value of the sample to generate a coordinate positioning value; When obtaining the Lab color space value of a sample from a colorimeter, the colorimeter first reads the sample surface reflectance spectrum data and converts it into tristimulus values. The colorimeter's internal calculation logic then calculates the corresponding L channel (brightness value), a channel (color axis component value from green to red), and b channel (color axis component value from blue to yellow). In this example, the Lab value of a certain fabric sample is measured to be L=62.38, a=-3.52, and b=18.77. The values ​​of the a and b channels are then called, with the a channel as the horizontal coordinate value Xa and the b channel as the vertical coordinate value Yb. When performing coordinate conversion, Xa=-3.52 and Yb=18.77 are mapped to the given unit ratio. The coordinate grid position is in units of 0.1. The horizontal coordinate is determined by the negative value of the a channel to be located in the left half plane, and the vertical coordinate is determined by the positive value of the b channel to be located in the upper half plane. In this mapping, the two-dimensional coordinate of the sample is located at (-3.5, 18.8). Then, the two-dimensional coordinate is arranged in a one-to-one correspondence with the L channel value of the sample L=62.38 to form a set of triplet data {L, Xa, Yb}={62.38, -3.5, 18.8}. In order to ensure the subsequent positioning accuracy, the same sample needs to be tested three times for average processing. For example, the L values ​​measured three times are 62.38, 62.42, and 62.36 respectively, then the mean value of L is Lmean=(62.3 8+62.42+62.36) / 3=62.387. The mean calculation method of Xa and Yb is the same, thus obtaining the mean coordinate positioning value {62.387, -3.51, 18.78}. In this sorting operation, the specific steps for calculating the average value are to first call the corresponding channel values ​​one by one, perform floating-point addition operation on the corresponding channel values ​​measured each time and record the total, and then call the value of the measurement number n and perform division operation on the total to obtain the final mean. For example, in the above example, the mean of channel a = (-3.52) + (-3.50) + (-3.51) / 3 = -3.51, and the mean of channel b = (18.77+18.80+18.78) / 3 = 18. .783. The judgment link involved in the above calculation is: if a channel value exceeds the preset physical reasonable range (L value 0~100, a value -128~127, b value -128~127), the measurement data will be directly discarded, and re-measured to make up the number of valid data groups. For the judgment of "higher" and "lower", for example, for the L channel value, 0≤L<33 can be set as the low brightness range, 33≤L<66 as the medium brightness range, and 66≤L≤100 as the high brightness range. In this example, L=62.387 belongs to the medium brightness range. Finally, when generating the coordinate positioning value, the sorted triplet mean data is called to output the positioning value table item in a unified format for subsequent matrix comparison.

[0025] Table 1: Sample Lab original measurement and mean calculation table As shown in Table 1, by performing floating-point addition and division calculations on the three measurement data, the mean of the three channels L, a, and b is obtained. The mean is the stable positioning value of the sample in the Lab color space, and the final coordinate positioning value is formed by combining the two-dimensional coordinates (a, b) and the corresponding L value.

[0026] S102: Based on the coordinate positioning value, call the pre-built L, a, and b correction matrices, compare the data boundaries of the cells in the matrix item by item according to the horizontal and vertical coordinate positions, filter the corresponding matrix cell data, and compare the filtered results with the L channel values ​​item by item and unify the format to obtain the matrix index value; Based on the coordinate positioning value {62.387, -3.51, 18.783}, first call the pre-built L channel correction matrix, a channel correction matrix and b channel correction matrix three sets of data, these three matrices are partitioned and stored according to the measurement range of L value, a value and b value respectively. For example, the L channel matrix is ​​indexed with every 5 units as a row, the a channel matrix is ​​indexed with every 2 units as a column, and the b channel matrix is ​​indexed with every 2 units as a column. The first step of calling the correction matrix data is to directly read the horizontal coordinate Xa=-3.51 and the vertical coordinate Yb=18.783 in the positioning value, perform interval position judgment on Xa, and compare it with the matrix column index. The boundary values ​​of the reference are compared item by item. The column index boundary of the matrix a channel starts from -128 to 127 and is arranged in sequence with a step size of 2. -4 to -2 corresponds to a column, then -3.51 is greater than -4 and less than -2, and it is determined to fall within the range of the column. At the same time, the same action is performed on Yb, and 18.783 is compared with the boundary value of the matrix b channel row index. For example, the row index 18 to 20 is an interval, and 18.783 is between them, then it is determined to fall within the range of the row. In this way, the specific unit coordinates of the matrix in the a and b dimensions are determined. Then the L value = 62.387 is called for the same judgment. For example, the L channel matrix has 60 to 65 as a row, 65 To 70 is the next row. Among these values, 62.387 is in the range of 60 to 65. The corresponding row index of the L matrix is ​​determined. When filtering the corresponding matrix unit data, the system will retrieve the cells that meet the above a, b, and L interval conditions from the three-dimensional matrix structure, and temporarily store the original correction vector in the cell for subsequent calculation. In the entire screening process, the comparison operation is specifically manifested as: the currently measured value is compared with the lower limit of each interval of the matrix to perform a greater than or equal judgment, and then with the upper limit to perform a less than judgment. Only cells that meet both judgment conditions can be selected. If the value is exactly equal to the upper limit of a certain interval, then It is classified into the next interval during the judgment. After the horizontal and vertical coordinates are judged, the filtered matrix unit data is compared with the current L value with the same data accuracy. If the matrix unit contains multiple correction coefficient groups, the group of data needs to be formatted into a unified array representation, such as arranging it in the form of a ternary array [Lc, ac, bc]. At this time, the matrix index value is generated by calling the determined row index and column index, and combining them into a set of unique index identifiers, such as (L row index = 13, a column index = 64, b column index = 69). The identifier is stored as the matrix index value as a direct call entry for subsequent weighted correction operations.

[0027] S103: Based on the matrix index value, the original correction coefficient in the correction matrix is ​​called to perform weighted calculation on the Lab color space value of the sample item by item, and the correction coefficient after weighted calculation is combined with the original value of the sample to generate the Lab measurement value to be corrected; According to the matrix index value, the original correction coefficient group corresponding to the position in the correction matrix is ​​first called. The coefficient group consists of three correction factors for the L, a, and b channels respectively. The original correction coefficient group retrieved from the matrix is ​​{ΔL=0.85, Δa=-0.12, Δb=0.34}, and then the Lab color space values ​​of the sample {62.387, -3.51, 18.783} are taken out in turn to participate in the calculation. The specific process of weighted calculation is to first assign a preset weight value to each channel. The setting of the weight value refers to the ratio of color difference to visual perception sensitivity. For example, the L channel weight is 0.6, the a channel weight is 0.25, and the b channel weight is 0.15. In this example, the weight allocation is based on the statistical results of the sample's sensitivity to the differentiated channels, and the sum of the weights in the range of 0 to 1 is determined to be 1 through the judgment condition. When the calculation is performed, the weighted correction amount for the L channel is , the weighted correction value of channel a is , the weighted correction value of channel b is , the weighted correction value maintains the same decimal precision as the original channel value, and then the original measurement value of the channel is directly added to the corresponding weighted correction value, that is, the correction result of the L channel is , the correction result of channel a is , the correction result of b channel is In this process, the numerical range judgment must be performed on each addition operation. For example, the L value needs to be kept in the range of 0 to 100. If it exceeds the upper limit of the interval after correction, it needs to be truncated to the maximum value of 100, and if it is lower than the lower limit, it needs to be truncated to 0. The a value and b value need to be kept in the range of -128 to 127. The correction results of this example all fall within the valid range, so no truncation operation is required. Finally, the three corrected channel values ​​are combined into a new Lab data group {62.897, -3.54, 18.834}. This data group is the Lab measurement value to be corrected after the weighted correction operation. This group of values ​​will enter the subsequent links for further data verification and comparison to determine the stability and consistency of the correction.

[0028] See also Figure 3 , the specific steps of S2 are: S201: Based on the Lab measurement values ​​to be corrected, the measurement values ​​are sequentially input into the piecewise linear regression model, the values ​​are fitted according to the piecewise function coefficients within the interval, and the corresponding parameters are extracted according to the slope and intercept of the fitting curve in the differentiation interval to generate the regional segmentation characteristic coefficients; The regional segmental characteristic coefficient refers to the slope and intercept obtained by piecewise linear regression fitting, which is used to characterize the corresponding relationship between the measured value and the standard value in the differentiation interval; Based on the Lab measurement value to be corrected, first input L=62.897, a=-3.54, and b=18.834 into the calculation process of the piecewise linear regression. The piecewise linear regression structure here is a continuous interval divided based on the channel value range. For example, the L channel is set to 0 to 33 as the low brightness segment, 33 to 66 as the medium brightness segment, and 66 to 100 as the high brightness segment. For the a channel, -128 to -64 is set as the strong green segment, -64 to 0 is the green segment, and 0 to 64 is the red segment. 64 to 127 is a strong reddish segment, the b channel is set to -128 to -64 as a strong blue segment, -64 to 0 as a blue segment, 0 to 64 as a yellow segment, and 64 to 127 as a strong yellow segment. Each channel value to be corrected is positioned within the established interval according to its range. For example, L=62.897 falls in the medium brightness segment, a=-3.54 falls in the green segment, and b=18.834 falls in the yellow segment. Then, the piecewise function coefficient group under the corresponding interval is called respectively. For example, the coefficient group of the middle segment of L is set to the slope ,intercept , the coefficient group of the segment where the a value is located is ,intercept , the coefficient group of the segment where the b value is located is ,intercept , during the fitting process, for each channel value Linear operation of is the corresponding measurement channel value, 、 The stored coefficient group from the segment where the channel is located, such as the L channel fitting value , a channel fitting value , b channel fitting value To ensure that the interval selection is correct, a boundary comparison needs to be performed before calling each coefficient group. The input value is double-judged to be greater than or equal to and less than the upper and lower limits of the interval. For example, for a=-3.54, it is judged whether it is ≥-64 and <0. If the condition is met, the green segment coefficient group is selected. After completing the fitting value calculation of the three channels, the channel characteristics are directly extracted according to the slope and intercept of the fitting curve. For example, the characteristic coefficient of the L segment corresponds to the slope of this segment. =0.045, intercept =60.1, the characteristic coefficients of segment a are 0.011 and -4.0, and the characteristic coefficients of segment b are 0.024 and 16.5. Combining these three pairs of coefficients, we can finally generate the regional segmentation characteristic coefficient set {(0.045, 60.1), (0.011, -4.0), (0.024, 16.5)}.

[0029] S202: calling the regional segment characteristic coefficients and comparing them with the original standard values ​​in the grid one by one. Based on the difference between the current measured Lab value and the standard value, the results are summarized using the difference accumulation method. The accumulated differences are uniformly summarized as a calculable indicator to obtain the grid difference value. The grid difference refers to the cumulative difference between the measured value and the standard value, which is used to reflect the overall deviation degree within a single grid; Call the regional segmented feature coefficient set {(0.045, 60.1), (0.011, -4.0), (0.024, 16.5)}, call the current Lab measurement value {62.897, -3.54, 18.834} one by one and compare the difference with the original standard value stored in the grid. Assuming that the standard Lab value of a certain area is {63.120, -3.60, 18.920}, first calculate the difference of the L channel respectively , a channel difference , b channel difference The calculation of each difference follows the principle of taking the current measured value as the minuend and then taking the standard value as the subtrahend, performing subtraction to get the result and accurate to three decimal places, and then entering the difference accumulation step, adding the three channel differences to the corresponding accumulation variables respectively. For example, when the initial accumulation values ​​are all 0, execute 、 、 If there are multiple sets of standard values ​​in the current grid, it is necessary to loop through each set of values ​​and calculate the difference between the measured values ​​and add them up in sequence. For example, compare another set of standard values ​​{63.050, -3.50, 18.850} and get the difference 、 、 , continue execution 、 、 The judgment condition in the entire accumulation process is: if the absolute value of the difference is less than 0.005, it is considered negligible and does not participate in the accumulation to avoid interference caused by small measurement fluctuations. After completing all comparisons, the cumulative differences of the three channels are sorted into a unified format indicator vector {-0.376, 0.020, -0.102}. This vector is the grid difference between the grid and the current measured Lab value in all comparisons.

[0030] S203: setting weights based on the grid differences and the number of measurements corresponding to the grids in the counting matrix, performing weighted calculation on the differences using the weights, and converting the weighted results into overall correction values ​​to obtain regionalized correction offsets; After obtaining the grid difference vector {-0.376, 0.020, -0.102}, it is necessary to set the channel weight based on the number of times the grid is measured in the counting matrix. Assuming that the cumulative number of measurements for the L channel is 5, the a channel is 3, and the b channel is 2, first calculate the weight coefficient of each channel The specific setting method of weight is: divide the number of measurements of each channel by the total number of measurements of the three channels, for example, the total number of , then the L channel weight , a channel weight , b channel weight In the weighted calculation, the grid difference of the corresponding channel is directly multiplied by its weight to obtain the weighted difference: , , After completing the weighted difference calculation of the three channels, add them together to get the overall correction value This value represents the net offset of the weighted difference of the grid to the overall color measurement, which is then defined as the regionalized correction offset. If the value is negative, it indicates that the overall correction should be reduced in the corresponding direction. If it is positive, it indicates that the value should be increased. In this example, the result is -0.2024, indicating that the corresponding color parameter should be negatively adjusted in the subsequent correction step.

[0031] Table 2: Weighted calculation table of grid difference As shown in Table 2, by converting the difference of each channel into a weight according to the number of measurements of the channel and summing the products, the final regionalized correction offset of -0.2024 can be obtained, which is used for numerical adjustment in subsequent correction steps.

[0032] See also Figure 4 , the specific steps of S3 are: S301: performing point-by-point comparison of the measurement results based on the regionalized correction offset, performing difference calculation between the measurement matrix grid point values ​​and the regionalized correction offset, extracting values ​​according to the position coordinates and weightedly accumulating them in the difference calculation to generate a difference offset coefficient; Correcting offsets based on regionalization To compare the measurement results point by point, first call the position coordinate value of each grid point in the measurement matrix in turn And the corresponding actual measured Lab data For each grid point, first extract the three-channel values ​​in the measurement matrix. For example, at a certain grid point , the corresponding L, a, b values ​​are measured as , then regionalize the corrected offset In the difference calculation step applied to the grid point, the specific approach is: perform the difference value , , , the action in the difference value calculation is to use the original value of the current grid point as the minuend and the offset as the subtrahend to perform subtraction operation to obtain the new value. If the matrix grid point is at the intersection of multiple regions, it is necessary to determine the priority region offset based on the coordinate segment to which the grid point belongs, to ensure that only a single region parameter is involved in the calculation, and then perform the value extraction and weighted accumulation operation according to the position coordinates, for example, for The weight of the grid point is set based on the ratio of the cumulative number of measurements of the row and column in the matrix. Assume that the statistical weight of the grid point is , then multiply by the weight when accumulating, and get , , , traverse the matrix grid points in this way, and sum the weighted difference values ​​of each grid point in the three channels L, a, and b respectively. For example, assuming that the measurement matrix has 100 grid points, the accumulated result after traversal is , , , and finally the weighted sum of these three channels is combined into a set of three-dimensional vectors , this set of vectors is the difference offset coefficient formed by weighted accumulation according to position, which is used for the next step of nonlinear weight distribution calculation.

[0033] S302: Calling the difference offset coefficient to perform nonlinear weight allocation. In the weight allocation, cells in the count matrix with zero grid points are compensated according to the values ​​of adjacent non-zero grid points. The compensated value is superimposed on the original allocated value and normalized to obtain a compensated weight value. Call the difference offset coefficient vector , enter the nonlinear weight distribution step, first scan the measurement matrix grid by grid, and detect the measurement times of the grid points in the counting matrix ,when When , the grid point is determined to be an empty measurement unit and needs to be compensated for the adjacent grid point. This judgment action is to perform an equal value judgment operation with zero after reading the measurement number of each grid point. If they are equal, the coordinates are recorded and enter the compensation queue. In the compensation calculation, for each empty point, the non-zero grid point value in the adjacent grid point set with a Manhattan distance of 1 (i.e., the upper, lower, left, and right adjacent grid points) is extracted, that is: ; The average of the difference offset coefficients of the non-zero grid points is taken as the compensation value. For example, if there is no measurement value at the coordinate (5, 6), the weighted difference offset values ​​of the non-zero grid points found from (4, 6), (6, 6), (5, 5), and (5, 7) are 2.53, 2.49, 2.60, and 2.58 respectively. The compensation value is , this compensation value will be added to the original assigned difference offset weight, that is, for the original weight The grid points are adjusted to After completing the compensation of zero-measurement grid points, a set of compensation weight matrices is obtained , in order to perform the ratio normalization, it is necessary to calculate The sum of , for example, the sum calculation result is , and then perform for each compensation weight value in the matrix The division operation of makes the compensation weights of the grid points between 0 and 1 and the sum is exactly 1, for example The grid normalization result is , obtained after completing full matrix normalization The set is the compensation weight value, which is used for the next channel offset correction operation.

[0034] S303: Calling the compensation weight value to perform channel offset correction, in the correction, the compensation weight and the channel offset parameter are superimposed one by one, and the channel offset parameter is redistributed in the superposition matrix to obtain the correction value before boundary verification; Call the compensation weight value and perform the channel offset correction operation. First, traverse each grid point in the matrix one by one, and perform a one-to-one multiplication operation on the compensation weight value and the channel offset parameter at the corresponding position to obtain the weighted offset. For example, in the L, a, and b three-channel offset parameter matrix, a grid point The offset parameters are 、 、 , and its normalized compensation weight is , then the weighted result of the grid point is (L channel), (a channel), (b channel), this operation is performed on each grid point in the matrix in turn, and the weighted offset of the grid point is added in the corresponding channel. For example, after traversing the grid points, the weighted sum of the L channel can be obtained , a channel weighted sum , b channel weighted sum , the weighted sum will be defined as the overall compensation value of the channel, and then the overall compensation value will be superimposed on the original total offset parameter of the channel. For example, the original total offset parameter is 、 、 , then the corrected channel value is 、 、 Finally, the corrected values ​​of the three channels are redistributed back to the corresponding grid points in the matrix, so that each grid point in the superposition matrix is ​​updated to the corrected channel value set. For example, the original value of grid point (5, 6) {62.91, -3.55, 18.84} is updated to {63.422, -3.714, 19.078} after the corresponding weighted compensation superposition and overall compensation correction rules are adjusted. After the grid points are updated, the resulting matrix is ​​the corrected value matrix before entering the boundary verification phase.

[0035] Table 3: Channel offset correction calculation example As shown in Table 3, by multiplying the normalized compensation weights with the channel offset parameters one by one, adding the results to the original channel values, and then redistributing the entire matrix, the correction result matrix used before boundary verification can be obtained.

[0036] See also Figure 5 , the specific steps of S4 are: S401: Based on the correction value before boundary verification and the current grid boundary range, determine whether the corrected ab value exceeds the boundary, compare the corrected ab value with the upper and lower limits of the grid boundary, mark the out-of-bounds data points, extract their corresponding coordinate information, and generate an out-of-bounds coordinate set; When judging based on the correction value before boundary verification and the current grid ab boundary range, read the corrected a and b channel values ​​of each grid point in the matrix one by one, and call the corresponding grid preset upper and lower limits. For example, the current grid a channel allowable range is [-5.00, 4.00], and the b channel allowable range is [15.00, 19.00]. Then, the comparison action is performed for each grid point in turn. The comparison action is divided into two steps: the first step is to perform a greater than judgment, and compare the a value with the a upper limit. If Then it is judged as the upper bound is exceeded; the second step is to perform the less than judgment, compare the value of a with the lower limit of a, if The lower bound is determined to be exceeded. Similarly, the b value is compared with the greater-than and less-than values ​​to detect the situation of exceeding the b channel boundary. For example, at the grid point (5, 6), the a value of -3.714 is greater than the lower limit of -5.00 and less than the upper limit of 4.00, so it is not in the a channel out-of-bounds set. The b value of 19.078 is greater than the b upper limit of 19.00 and the difference 19.078-19.00=0.078>0, which is determined to be the b channel upper bound exceeded. The coordinates of the grid point (5, 6) are recorded in the out-of-bounds coordinate set and the channel type is marked as b. Upper bound out of bounds. During the entire scan, coordinate information extraction is performed for each out-of-bounds grid point, and the (i, j) two-dimensional index and out-of-bounds channel identifier are stored in groups, for example, [(5, 6, b+), (8, 9, a−), (3, 4, a+), (3, 4, b−)]. In this process, if the a and b values ​​of a grid point are both out of bounds, they are recorded once in the out-of-bounds coordinate set so that the subsequent processing stage can limit them by channel. After the matrix grid point scan is completed, the out-of-bounds judgment and coordinate extraction are completed to generate the out-of-bounds coordinate set.

[0037] S402: Calling the grid position coordinates in the out-of-bounds coordinate set, re-limiting the ab value that exceeds the boundary according to the upper and lower limits of the boundary, adjusting the difference of the out-of-bounds part to the critical position, and recalculating the limited ab value distribution range to obtain the boundary interception value interval; Call the out-of-bounds coordinate set [(5, 6, b+), (8, 9, a−), (3, 4, a+), (3, 4, b−)], call the grid point positions corresponding to the coordinates in turn, and perform amplitude limiting processing on the a and b values ​​that exceed the boundaries. The limiting process is performed on each channel separately. First, the original corrected value of the grid point and the corresponding upper and lower limits are taken. For example, for the grid point (5, 6, b+), its b value is 19.078, and the upper limit of the b channel is 19.000. Perform the difference calculation If the result is greater than 0, it means it exceeds the upper limit, then adjust the b value to the upper limit. For the grid point (8, 9, a−), the a value is -5.346, the a channel lower limit is -5.000, and the difference calculation is performed Since the result is less than 0 and the absolute value is greater than 0, it is determined to be out of bounds, and the value of a is adjusted to the lower limit. ; For the grid point (3, 4, a+), the a value is 4.128, the a channel upper limit is 4.000, execute If the result is greater than 0, adjust the a value to 4.000; for the grid point (3, 4, b−), the b value is 14.968, the b channel lower limit is 15.000, execute , if the result is less than 0, increase the b value to 15.000 to meet the boundary conditions. Each adjustment only affects the corresponding channel value, and the other channel values ​​remain unchanged. After completing the amplitude adjustment of the out-of-bounds grid points, re-count the value ranges of the two channels a and b of the full matrix. The statistical method is to traverse the grid points and record the minimum and maximum values ​​of the two channels respectively. For example, after adjustment, the minimum value of channel a is -5.000 and the maximum value is 4.000, and the minimum value of channel b is 15.000 and the maximum value is 19.000. Finally, these two ranges constitute the new boundary interception value interval for subsequent boundary data integration processing.

[0038] S403: Based on the correction results in the boundary interception value interval, the ab values ​​within the boundary and the critical point are integrated, and the adjusted data and the data that do not cross the boundary are merged to obtain the output result after boundary verification; According to the boundary interception value interval (a channel [-5.000, 4.000], b channel [15.000, 19.000]), the correction results of the grid points are integrated according to the interval. First, each grid point in the matrix is ​​traversed to determine whether its a and b values ​​belong to the result set after the out-of-bounds adjustment. If it belongs to the data within the boundary, the original correction value is directly retained. If it belongs to the critical point (that is, the grid point with a value equal to -5.000 or 4.000, and b value equal to 15.000 or 19.000), the limited critical value is retained and merged with the non-out-of-bounds data in the new matrix. For example, the adjusted value of grid point (5, 6) is {L=63.422, a=-3.714, b=19.000}, because the b value The value of grid point (8, 9) after adjustment is {L=61.235, a=-5.000, b=16.742}, which becomes a critical point because the a value is equal to the lower bound, and is retained in the output matrix; grid point (3, 4) is adjusted to the critical value {L=64.017, a=4.000, b=15.000} within the boundary in both a channel and b channel, and is also directly incorporated into the final matrix, while grid points that do not appear in the out-of-bounds coordinate set (such as (6, 5)={L=62.881, a=-1.352, b=17.641}) are fully retained to ensure that the matrix after boundary verification is composed of the adjusted critical point data and the non-out-of-bounds data to form a complete output result matrix.

[0039] Table 4: Example of partial grid point output after boundary verification As shown in Table 4, after the unified merging of the critical points and the data within the boundary, the obtained matrix is ​​the final output result matrix after boundary verification, which can be directly called by subsequent links.

[0040] See also Figure 6 , the specific steps of S5 are: S501: Based on the output result after boundary verification, obtain the value of the corresponding color block in the correction knowledge base, detect the Lab deviation of the newly added measurement point, calculate the difference between the newly added deviation value and the existing reference Lab value, record the deviation value based on the difference, update the correction knowledge base, and generate a newly added Lab deviation value; Output the results based on boundary verification, first according to the coordinates of each grid point Read the corrected Lab values ​​in turn, and then query the current reference Lab value of the corresponding color block in the correction knowledge base. This query process locates the corresponding record entry in the knowledge base index table according to the grid coordinates. For example, for coordinates (5, 6), the reference Lab value recorded in the correction knowledge base is , add the Lab value of the new measurement point Deviation detection is performed with this record. The order of deviation calculation follows the method of processing three channels one by one. The L channel deviation is executed first. , then execute channel a deviation , then perform the b channel bias , the channel deviation is retained to three decimal places to ensure the same accuracy in subsequent accumulation, and then the difference between the new deviation and the existing reference value is compared. Here, the "comparison" action is to perform absolute value calculation and zero threshold judgment. If If the value is greater than 0, then record it, otherwise ignore it. In this example, all three channels are greater than the threshold 0, so they are all recorded. The new deviation value is structured as Temporarily store and attach grid coordinate labels (5, 6) to each channel. When processing new measurement points in the entire matrix, continue to write multiple points' {deviation vector, grid coordinates} into the cache queue, and finally summarize them to form a new Lab deviation value set. , for example, some of the data is shown in Table 5 below.

[0041] Table 5: New Lab deviation value example table As shown in Table 5, the newly added Lab deviation value is obtained based on the grid-by-grid comparison between the corrected output matrix and the knowledge base benchmark value, and each deviation exceeding the zero threshold is recorded to form a deviation set for subsequent correction matrix update.

[0042] S502: calling the newly added Lab deviation value, updating the Lab correction matrix for the corresponding grid coordinate in the correction knowledge base, and adding the deviation value to the value of the corresponding position in the matrix to generate the grid coordinate cumulative correction value; Call the newly added Lab deviation value set , call the grid coordinates recorded by each element in the collection in turn , find the corresponding Lab correction matrix position in the correction knowledge base and perform the cumulative update operation. This process is performed separately for the three channels L, a, and b. For example, for the corresponding deviation (5, 6) recorded in the set First, read the comprehensive cumulative correction value currently stored at position (5, 6) in the correction matrix , assuming its current value is , then perform the addition operation in sequence: , , After completing the update of the three-channel values ​​of the coordinate, directly write it back to the storage unit corresponding to the correction matrix. For the corresponding deviation of (8, 9) recorded in the set , also read the original value And accumulate them separately, for example, the original value Updated to , which is applied one by one For coordinate points in the set, if a channel has no stored value during the update process (when recording for the first time), the deviation value is automatically written directly into the matrix unit as the initial value without accumulation. Finally, after traversing the entire set of newly added Lab deviation values, the correction matrix obtained is the grid coordinate cumulative correction value matrix after accumulating and integrating the deviation records of this batch.

[0043] S503: Based on the accumulated correction values ​​of the grid coordinates, the corresponding coordinate points are searched in the counting matrix, the counting values ​​are incremented, and after the counting matrix is ​​updated, the correction matrix and the counting matrix are stored together to generate the colorimeter color measurement accuracy optimization result; According to the lattice coordinate cumulative correction value, the counting matrix is ​​updated synchronously according to the coordinate value. First, the non-zero coordinate positions in the cumulative correction matrix are read one by one. , and then locate the same coordinate unit in the counting matrix and perform the count increment operation, which is to add 1 to the current count value to record that the corresponding grid point has experienced a new deviation accumulation correction. For example, when the original count value of the position (5, 6) in the counting matrix is When executing Update, if the original count value of position (8, 9) is , then update to This point-by-point increment method will be completed in a loop according to the coordinate list of the newly added deviation records. When a coordinate that has never appeared in the counting matrix is ​​processed, its count value is initialized to 1 until the coordinates involved in the cumulative correction matrix are counted and updated. Finally, the correction matrix and the counting matrix after this round of update are stored together in the persistent storage area of ​​the system to form a complete data snapshot file, in which the correction matrix reflects the cumulative numerical offset of the grid points on the three channels L, a, and b, and the counting matrix reflects the number of times the grid points have been corrected. This storage result is the final formation of the colorimeter's color measurement accuracy optimization result.

[0044] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for correcting color measurement results of a colorimeter, characterized in that: The following steps are involved: S1: Obtain the Lab color space value of the sample to be corrected through a colorimeter, calculate the horizontal and vertical coordinates of the grid according to the a channel value and the b channel value, index the pre-constructed L, a, and b correction matrix to analyze the original correction coefficient, and obtain the Lab measurement value to be corrected; S2: Perform regional analysis based on the Lab measurement value to be corrected, input the Lab measurement value to be corrected into a piecewise linear regression model to extract regional features, calculate the deviation between the current measured Lab value and the original standard value in the corresponding grid, and perform weighted processing based on the number of measurements of the grid in the counting matrix to obtain the regional correction offset; S3: Adaptively correcting the measurement results by calling the regionalized correction offset, performing nonlinear weight allocation according to the regionalized correction offset, obtaining correction parameters by using a neighborhood compensation technique when the grid value corresponding to the counting matrix is ​​zero, performing channel offset correction, and obtaining a correction value before boundary verification; S4: Perform color gamut boundary constraint verification processing using the correction value before boundary verification. When the corrected ab value exceeds the current grid boundary, limit the correction amplitude to within the grid boundary range, and perform callback processing on the out-of-bounds value to obtain the output result after boundary verification.

2. The method for correcting colorimetric results according to claim 1, wherein: The Lab measurement value to be corrected includes a brightness component, a red-green hue component, and a yellow-blue hue component; the regionalized correction offset includes a local difference value, a regional weighted value, and a feature mapping value; the correction value before boundary verification includes a channel offset, a compensation coefficient, and an allocation weight; the output result after boundary verification includes a boundary constraint value, an amplitude limit value, and a callback correction value.

3. The method for correcting colorimetric results according to claim 1, wherein: The specific steps of S1 are: S101: Obtain the Lab color space value of the sample measured by the colorimeter, call the a channel value and the b channel value, calculate the horizontal and vertical coordinate positions of the two in the coordinate plane, and correspond the horizontal and vertical coordinates with the L channel value of the sample to generate a coordinate positioning value; S102: Based on the coordinate positioning value, call the pre-built L, a, and b correction matrices, compare the data boundaries of the cells in the matrix item by item according to the horizontal and vertical coordinate positions, filter the corresponding matrix cell data, and compare the filtered results with the L channel values ​​item by item and unify the format to obtain the matrix index value; S103: According to the matrix index value, for the original correction coefficient in the correction matrix, call the Lab color space value of the sample to perform weighted calculation item by item, combine the correction coefficient after weighted calculation with the original value of the sample to generate the Lab measurement value to be corrected.

4. The method for correcting colorimetric results according to claim 3, wherein: The specific steps of S2 are: S201: Based on the Lab measurement values ​​to be corrected, the measurement values ​​are sequentially input into a piecewise linear regression model, the values ​​are fitted according to the piecewise function coefficients within the interval, corresponding parameters are extracted according to the slope and intercept of the fitting curve in the differentiation interval, and regional segmental characteristic coefficients are generated; S202: calling the regional segment characteristic coefficients and comparing them with the original standard values ​​in the grid one by one, summarizing the results using a difference accumulation method based on the difference between the current measured Lab value and the standard value, and summarizing the accumulated differences into a calculable indicator to obtain a grid difference value; S203: According to the grid difference amount, a weight is set in combination with the number of measurements corresponding to the grid in the counting matrix, the difference amount is weightedly calculated using the weight, and the weighted result is converted into an overall correction value to obtain a regionalized correction offset.

5. The method for correcting colorimetric results of a colorimeter according to claim 4, wherein: The specific steps of S3 are: S301: performing point-by-point comparison of the measurement results based on the regionalized correction offset, performing difference calculation between the measurement matrix grid point values ​​and the regionalized correction offset, extracting values ​​according to the position coordinates and weightedly accumulating them in the difference calculation to generate a difference offset coefficient; S302: Calling the difference offset coefficient to perform nonlinear weight allocation, compensating cells in the count matrix with zero grid points according to the values ​​of adjacent non-zero grid points in the weight allocation, superimposing the compensated value with the original allocated value and performing normalization processing to obtain a compensated weight value; S303: calling the compensation weight value to perform channel offset correction, superimposing the compensation weight and the channel offset parameter one by one in the correction, and redistributing the channel offset parameter in the superposition matrix to obtain the correction value before boundary verification.

6. The method for correcting colorimetric results according to claim 5, wherein: The difference offset coefficient refers to the difference quantization coefficient obtained by comparing the measurement matrix grid point value with the regionalized correction offset point by point and then weighted accumulation; The normalization process adopts a proportional normalization method, that is, dividing the compensation weight value by the sum of the compensation weight values; The compensation weight value refers to a weight parameter obtained after performing neighboring value compensation and normalization processing on the zero-value grid point in nonlinear weight distribution.

7. The method for correcting colorimetric results according to claim 5, wherein: The specific steps of S4 are: S401: Based on the pre-boundary verification correction value and the current grid boundary range, determine whether the corrected ab value exceeds the boundary, compare the corrected ab value with the upper and lower limits of the grid boundary, mark the out-of-bounds data points, extract their corresponding coordinate information, and generate an out-of-bounds coordinate set; S402: Calling the grid position coordinates in the out-of-bounds coordinate set, re-limiting the ab value that exceeds the boundary according to the upper and lower limits of the boundary, adjusting the difference of the out-of-bounds part to the critical position, and recalculating the limited ab value distribution range to obtain the boundary interception value interval; S403: According to the correction result in the boundary interception value interval, the ab values ​​within the boundary and the critical point are integrated, and the adjusted data and the data that do not cross the boundary are merged to obtain the output result after boundary verification.

8. The method for correcting colorimetric results according to claim 1, wherein: The method further comprises step S5: S5: Based on the output result after the boundary verification, the correction knowledge base is updated and maintained, the newly added Lab deviation is calculated and accumulated to the Lab correction matrix of the corresponding grid coordinate, and the value of the corresponding position of the count matrix is ​​incremented to obtain the colorimeter color measurement accuracy optimization result; The colorimeter color measurement accuracy optimization result includes the deviation accumulation amount, the correction matrix increment and the counting matrix update value.

9. The method for correcting colorimetric results according to claim 8, wherein: The specific steps of S5 are: S501: Based on the output result after the boundary verification, obtain the value of the corresponding color block in the correction knowledge base, detect the Lab deviation of the newly added measurement point, calculate the difference between the newly added deviation value and the existing reference Lab value, record the deviation value according to the difference, update the correction knowledge base, and generate a newly added Lab deviation value; S502: calling the newly added Lab deviation value, updating the Lab correction matrix for the corresponding grid coordinate in the correction knowledge base, and adding the deviation value to the value of the corresponding position in the matrix to generate a grid coordinate cumulative correction value; S503: According to the accumulated correction value of the grid coordinates, the corresponding coordinate point is searched in the counting matrix, the counting value is incremented, and after the counting matrix is ​​updated, the correction matrix and the counting matrix are stored together to generate the colorimeter color measurement accuracy optimization result.

10. The method for correcting colorimetric results of a colorimeter according to claim 9, wherein: The newly added Lab deviation value refers to the difference between the newly added measurement point and the benchmark Lab value in the revised knowledge base; The lattice coordinate cumulative correction value refers to the accumulation of the original Lab deviation values ​​of the corresponding lattice coordinates in the correction matrix.

Images