Textile color difference analysis method and system based on image processing

Abstract

The present invention belongs to the field of image processing technology, and specifically relates to a method and system for color difference analysis of textiles based on image processing. The method includes automatically collecting textile image data of each batch, automatically analyzing the pre-processed textile image data, and obtaining a textile image acquisition quality index; when the textile image acquisition quality index of the textile to be tested is analyzed to be qualified, the pre-processed textile image data is automatically color analyzed and compared, and the color features in the textile image are automatically obtained; through a color difference analysis method based on color feature matching, the first color difference value between the textile image to be tested and the reference sample image is automatically analyzed, the second color difference value between the textile to be tested and the reference sample is analyzed by a colorimeter, and the textile color difference value of the textile to be tested is analyzed according to the first color difference value and the second color difference value of the textile to be tested. The present invention realizes the improvement of the automation degree of textile color difference analysis.

Description

Technical Field

[0001] The invention belongs to the technical field of image processing, and in particular relates to a textile color difference analysis method and system based on image processing. Background Art

[0002] Textile color difference analysis is the process of scientifically evaluating and analyzing the differences in color of textiles. Such differences usually occur during the production, processing, dyeing or printing of textiles due to different conditions and factors. Textile color difference analysis is mainly used to ensure product quality and color consistency.

[0003] At present, there are many researches in the field of image processing technology. For example, CN117095067A discloses a textile color difference detection method based on artificial intelligence, and CN115100186A discloses a textile color difference detection method based on image data. However, in specific applications, due to the lack of standardization and consistency of the image acquisition equipment used, different devices may cause color differences, thereby affecting the reliability of the textile color difference analysis results; secondly, the existing image processing technology is difficult to operate stably in the automation system in a complex environment, resulting in a low degree of automation in textile color difference analysis. Summary of the invention

[0004] In view of the above deficiencies in the prior art, the purpose of the present invention is to provide a method and system for color difference analysis of textiles based on image processing, thereby improving the degree of automation of color difference analysis of textiles and solving the problem of low degree of automation of color difference analysis of textiles in the prior art.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is a textile color difference analysis method based on image processing, comprising the following steps:

[0006] Automatically collect the textile image data of each batch, perform automatic image preprocessing on the collected textile image data, and automatically analyze the preprocessed textile image data to obtain a textile image collection quality index, which is used to evaluate whether the comprehensive quality of the collected textile image data is qualified;

[0007] When the textile image acquisition quality index of the textile to be tested is found to be qualified, the pre-processed textile image data is automatically subjected to color analysis and comparison to automatically obtain color features in the textile image;

[0008] Through a color difference analysis method based on color feature matching, a first color difference value between the image of the textile to be tested and the image of a reference sample is automatically analyzed according to the color features in the textile image, a second color difference value between the textile to be tested and the reference sample is analyzed by a colorimeter, and a textile color difference value of the textile to be tested is analyzed based on the first color difference value and the second color difference value of the textile to be tested, and the textile color difference value is used to evaluate the color consistency between the textile to be tested and the reference sample.

[0009] Preferably, the specific method of automatically collecting each batch of textile image data is:

[0010] All textiles are transported to the image acquisition space in batches through a fully automatic track to complete the textile image data acquisition of the textiles to be tested;

[0011] Obtain the theoretical light source and theoretical image acquisition space of the reference sample; wherein the image acquisition space has no interference from reflected light, and the light source and image acquisition space used for image acquisition of the product to be tested are consistent with the theoretical light source and theoretical image acquisition space of the reference sample.

[0012] Preferably, the specific method for performing automatic image preprocessing on the collected textile image data is:

[0013] The color textile image is converted into a grayscale image by weighted averaging method;

[0014] The texture and defect information of the textile image is obtained, and the cropping area of ​​the textile image is set through an image cropping algorithm to complete the cropping of the textile image; when cropping, the texture and defect information of the textile image is retained, and the area without texture and defect information is cropped.

[0015] Preferably, the specific method for analyzing the textile image acquisition quality index is:

[0016] Number all textiles by batch;

[0017] The images of the textile to be tested and the reference sample are converted into the frequency domain by Fourier transform, and the energy of the high-frequency components is analyzed to obtain the image clarity of the textile to be tested and the reference sample images;

[0018] Importing the images of the textile to be tested and the reference sample, directly generating the image histograms of the textile to be tested and the reference sample images through image processing software, and analyzing and obtaining the color restoration degree of the textile to be tested image by comparing the similarity of the histograms of the textile to be tested and the reference sample images;

[0019] Using a median filter to filter the collected textile image to be tested, analyzing the difference between the filtered collected textile image to be tested and the original image, and then obtaining the noise level of the textile image to be tested;

[0020] By comparing the color changes of the collected continuous images of the textile to be tested with a light source stability test instrument, the light source fluctuation of the textile to be tested image is obtained;

[0021] The image clarity, color reproduction, noise level and light source fluctuation of the textile image acquisition data are obtained, and the calculation formula of the textile image acquisition quality index is constructed based on them, which is expressed as:

[0022] ;

[0023] In the formula, FCJ i is the textile image acquisition quality index of the i-th batch of textiles, i is the textile batch number, i=1, 2, ..., N, N is the total number of textile batches, QX i is the image clarity of the images collected for the i-th batch of textiles, HY i is the color restoration degree of the images collected for the i-th batch of textiles, ZS i is the noise level of the images collected for the i-th batch of textiles, GY i is the light source fluctuation degree of the image collected for the i-th batch of textiles, α1 is the image clarity effect on FCJ i The weight ratio of α2 is obtained by objective weight method, and α2 is the color restoration degree to FCJ i The weight ratio of α3 is obtained by objective weight method, and α3 is the noise level to FCJ i The weight ratio of α4 is the effect of light source fluctuation on FCJ i The weight ratio is obtained by objective weight method, e is a natural constant; lg represents the logarithmic function with base 10;

[0024] The textile image acquisition quality index is calculated according to the textile image acquisition quality index calculation formula;

[0025] When the image acquisition quality index of the acquired textile image to be tested is higher than the first threshold, it indicates that the comprehensive quality of the acquired textile image data is qualified;

[0026] When the image acquisition quality index of the acquired textile image to be tested is not higher than the first threshold, it indicates that the comprehensive quality of the acquired textile image data is unqualified.

[0027] Preferably, the specific method of performing automatic color analysis and comparison on the pre-processed textile image data is:

[0028] By using a color space conversion algorithm, the pixels of the textile and the reference sample are classified into different color areas according to the grayscale values ​​of the pixels of the preprocessed grayscale images of the textile to be tested and the reference sample;

[0029] The number of pixels in each color area of ​​the images of the textile to be tested and the reference sample is counted to generate a color histogram of the textile to be tested and the reference sample image. The generated color histogram is used as the color feature of the textile to be tested and the reference sample image.

[0030] Preferably, the specific analysis method for automatically analyzing the first color difference value between the textile image to be tested and the reference sample image is:

[0031] Normalizing the color histograms of the tested textile and the reference sample images;

[0032] The histogram intersection method is used to automatically analyze the similarity value between the color features of the textile image to be tested and the color features of the reference sample image, and the similarity value between the analyzed color features is used as the first color difference value between the textile image to be tested and the reference sample image.

[0033] Preferably, the specific analysis method of analyzing the second color difference value between the textile to be tested and the reference sample by using a colorimeter is:

[0034] Automatically calibrate the colorimeter for black and white, and automatically adjust the aperture of the colorimeter according to the textile to be tested and the reference sample;

[0035] Under the same standard light source and the same image acquisition space, the color data of the textile to be tested and the reference sample are measured by a colorimeter, and the color data measured by the colorimeter is recorded;

[0036] The colorimeter automatically analyzes the second color difference value between the textile to be tested and the reference sample by measuring the color data.

[0037] Preferably, the specific analysis method for analyzing the textile color difference value of the textile to be tested is:

[0038] Number all textiles by batch;

[0039] The first color difference value and the second color difference value of the textile to be tested are obtained, and a calculation formula for the color difference value of the textile is constructed based on the first color difference value and the second color difference value, which is expressed as:

[0040] ;

[0041] In the formula, SC i is the textile color difference value of the i-th batch of textiles, i is the textile batch number, i=1, 2, ..., N, N is the total number of textile batches, TX i is the first color difference value of the i-th batch of textiles, SCJ i is the second color difference value of the i-th batch of textiles, is the first color difference value pair SC i The weight ratio of is the second color difference value pair SC i The weight ratio of ; ln represents the natural logarithm with e as the base;

[0042] Calculate the color difference value of textiles according to the calculation formula of color difference value of textiles;

[0043] The calculated color difference value of the textile to be tested is compared with a preset color difference threshold to determine whether the color of the current textile to be tested is qualified.

[0044] Preferably, the specific method of comparing the calculated textile color difference value of the textile to be tested with a preset color difference threshold is:

[0045] Compare the calculated color difference value of the textile to be tested with the preset color difference threshold value to automatically determine whether the color of the current textile to be tested is qualified;

[0046] If the color difference value of the textile to be tested is lower than the preset color difference threshold, it means that the color of the current textile to be tested is qualified and meets the color requirements of textile production;

[0047] If the color difference value of the textile to be tested is not lower than the preset color difference threshold, it means that the color of the current textile to be tested is unqualified and does not meet the color requirements of textile production;

[0048] The color difference value of the textile to be tested is compared with the preset color difference threshold value, and the color difference analysis result is output.

[0049] A textile color difference analysis system based on image processing, used to implement the above-mentioned textile color difference analysis method based on image processing, including an image acquisition module, a color feature extraction module, and a color difference analysis module;

[0050] An image acquisition module is used to automatically acquire image data of each batch of textiles, automatically preprocess the acquired textile image data, and automatically analyze the preprocessed textile image data to obtain a textile image acquisition quality index, which is used to evaluate whether the comprehensive quality of the acquired textile image data is qualified;

[0051] A color feature extraction module is used to automatically analyze and compare the pre-processed textile image data after analyzing that the textile image acquisition quality index of the textile to be tested is qualified, and automatically obtain the color features in the textile image; when the textile image acquisition quality index of the textile to be tested is analyzed to be qualified, the textile to be tested is re-transported to the image acquisition location through a fully automatic track to perform image acquisition again;

[0052] The color difference analysis module is used to automatically analyze the first color difference value between the textile image to be tested and the reference sample image according to the color features in the textile image through a color difference analysis method based on color feature matching, analyze the second color difference value between the textile to be tested and the reference sample through a colorimeter, and analyze the textile color difference value of the textile to be tested according to the first color difference value and the second color difference value of the textile to be tested, and the textile color difference value is used to evaluate the color consistency between the textile to be tested and the reference sample.

[0053] The present invention has the following beneficial effects:

[0054] The present invention automatically collects and preprocesses textile image data, performs automatic color analysis and comparison on the preprocessed textile image data to obtain color features in the textile image, and then automatically analyzes the color difference value between the textile image to be tested and the reference sample image through a color difference analysis method based on color feature matching, thereby improving the automation level of textile color difference analysis and solving the problem of low automation level of textile color difference analysis in the prior art.

[0055] The present invention analyzes the textile color difference value of the textile to be tested by analyzing the first color difference value obtained after image processing analysis of the textile to be tested and the second color difference value obtained by colorimeter analysis, thereby making full use of the respective advantages of image processing and colorimeter to achieve more accurate color difference analysis.

[0056] The present invention automatically analyzes the pre-processed textile image data to obtain a textile image acquisition quality index, thereby reducing the color difference value result error caused by the quality problem of the image acquired by the textile to be tested. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 Schematic diagram of the process of the textile color difference analysis method based on image processing of the present invention;

[0058] Figure 2 It is a module schematic diagram of the textile color difference analysis system based on image processing of the present invention. DETAILED DESCRIPTION

[0059] The embodiments of the present invention are further described below in conjunction with the accompanying drawings.

[0060] Example 1: Figure 1 As shown, the textile color difference analysis method based on image processing includes the following steps:

[0061] Automatically collect the textile image data of each batch, perform automatic image preprocessing on the collected textile image data, and automatically analyze the preprocessed textile image data to obtain a textile image collection quality index, which is used to evaluate whether the comprehensive quality of the collected textile image data is qualified;

[0062] When the textile image acquisition quality index of the textile to be tested is found to be qualified, the pre-processed textile image data is automatically subjected to color analysis and comparison to automatically obtain color features in the textile image;

[0063] Through a color difference analysis method based on color feature matching, a first color difference value between the image of the textile to be tested and the image of a reference sample is automatically analyzed according to the color features in the textile image, a second color difference value between the textile to be tested and the reference sample is analyzed by a colorimeter, and a textile color difference value of the textile to be tested is analyzed based on the first color difference value and the second color difference value of the textile to be tested, and the textile color difference value is used to evaluate the color consistency between the textile to be tested and the reference sample.

[0064] Using the same standard light source and the same image acquisition space to acquire textile images can eliminate the impact of light source differences on image acquisition, avoid the influence of ambient light on the color difference analysis results of textile images, and improve the stability and reliability of textile image acquisition; the image processing analysis value is used to determine the specific numerical value of the color difference between the image of the textile to be tested and the reference sample image analyzed by the color difference analysis method based on color feature matching; the colorimeter analysis value is used to determine the specific numerical value of the color difference between the textile to be tested and the reference sample analyzed by the colorimeter.

[0065] The specific method for automatically collecting image data of each batch of textiles is: transporting all textiles in batches to the image collection space through a fully automatic track to complete the textile image data collection of the textiles to be tested; obtaining the theoretical light source and theoretical image collection space of the reference sample; wherein the image collection space has no interference from reflected light, and the light source and image collection space used for image collection of the product to be tested are consistent with the theoretical light source and theoretical image collection space of the reference sample.

[0066] The image acquisition space is a closed environment that can reduce the interference of external light and dust, which helps to ensure that the collected images of the textiles to be tested are clear and clean. In order to eliminate the impact of reflected light on the quality of textile images, the inner walls and floors of the image acquisition space are usually made of matte materials to avoid light reflection on these surfaces; the image acquisition space should be equipped with a professional lighting system, such as Canon Speedlite 600EX II-RT, Nikon SB-5000, Godox AD200 Pro and other brands and models of LED flash to ensure that the light is evenly irradiated on the textiles to be tested, which helps to reduce shadows and reflections and improve the clarity of the collected textile images; the image acquisition equipment in the image acquisition space uses professional image acquisition equipment with a resolution of or more than 40 million pixels and a color reproduction error of less than 5%, such as Canon EOS R5 or Nikon Z9 and other high-end cameras to complete the image acquisition of the textiles to be tested.

[0067] The specific method for automatic image preprocessing of collected textile image data is as follows: converting color textile images into grayscale images by weighted averaging method; obtaining texture and defect information of textile images, setting the cropping area of ​​textile images by image cropping algorithm, and completing textile image cropping; when cropping, retaining the texture and defect information of textile images, and cropping areas without texture and defect information.

[0068] Converting color images into grayscale images through weighted averaging can simplify image data and reduce computational complexity, while retaining sufficient texture and structural information for subsequent color difference analysis, improving analysis efficiency and accuracy, and facilitating subsequent color feature extraction and comparison. In order to retain the texture and existing defect information of the textile image, the cropping area is set to crop areas without texture and defect information, which can remove unnecessary edge and background information in the textile image, making the target area of ​​the textile image data more prominent.

[0069] The specific method for analyzing the textile image acquisition quality index is as follows: all textiles are numbered according to batches; the images of the textile to be tested and the reference sample are converted to the frequency domain by Fourier transform, and the energy of the high-frequency components is analyzed to obtain the image clarity of the textile to be tested and the reference sample images; the images of the textile to be tested and the reference sample are imported, and the image histograms of the textile to be tested and the reference sample images are directly generated by image processing software, and the color reproduction degree of the textile to be tested image is analyzed and obtained by comparing the similarity of the histograms of the textile to be tested and the reference sample images.

[0070] The collected textile image to be tested is filtered by using a median filter, and the difference between the filtered collected textile image to be tested and the original image is analyzed to obtain the noise level of the textile image to be tested; the color change of the collected continuous image of the textile to be tested is compared by a light source stability test instrument to obtain the light source fluctuation of the textile image to be tested; the image clarity, color reproduction, noise level and light source fluctuation degree of the textile image acquisition data are obtained, and a textile image acquisition quality index calculation formula is constructed based on them; the textile image acquisition quality index is calculated according to the textile image acquisition quality index calculation formula; the textile image acquisition quality index calculation formula is:

[0071] ;

[0072] In the formula, FCJ i is the textile image acquisition quality index of the i-th batch of textiles, i is the textile batch number, i=1, 2, ..., N, N is the total number of textile batches, QX i is the image clarity of the images collected for the i-th batch of textiles, HY i is the color restoration degree of the images collected for the i-th batch of textiles, ZS i is the noise level of the images collected for the i-th batch of textiles, GY i is the light source fluctuation degree of the image collected for the i-th batch of textiles, α1 is the image clarity effect on FCJ i The weight ratio of α2 is obtained by objective weight method, and α2 is the color restoration degree to FCJ i The weight ratio of α3 is obtained by objective weight method, and α3 is the noise level to FCJ i The weight ratio of α4 is the effect of light source fluctuation on FCJ i The weight ratio value is obtained through the objective weight method, and e is a natural constant.

[0073] When the image acquisition quality index of the collected textile image to be tested is higher than the first threshold, it indicates that the comprehensive quality of the collected textile image data is qualified; when the image acquisition quality index of the collected textile image to be tested is not higher than the first threshold, it indicates that the comprehensive quality of the collected textile image data is unqualified.

[0074] The textile image acquisition quality index can also use denoising algorithm, gray value algorithm and binarization algorithm to pre-process and extract features of the collected textile images. By comparing the image quality parameters before and after processing and combining with a specific quality assessment model, the quality index of textile image acquisition can be obtained; the higher the textile image acquisition quality index, the higher the comprehensive quality of the collected textile image data; the statistical table of textile image acquisition quality index is shown in Table 1:

[0075] Table 1 Statistical data related to textile image acquisition quality index

[0076]

[0077] Among them, α1=0.5, α2=0.2, α3=0.2, and α4=0.1 are set. From Table 1, we can see that the textile image acquisition quality index FCJ i With the image clarity QX i And color reproduction HY i This indicates that with the increase of image clarity and color reproduction, the quality of textile image acquisition will gradually deteriorate.

[0078] The specific method for automatically performing color analysis and comparison on the preprocessed textile image data is as follows: using a color space conversion algorithm, the pixels of the textile and the reference sample are classified into different color areas according to the grayscale values ​​of the pixels of the preprocessed grayscale images of the textile to be tested and the reference sample; the number of pixels in each color area in the images of the textile to be tested and the reference sample is counted, and a color histogram of the textile to be tested and the reference sample image is generated, and the generated color histogram is used as the color feature of the textile to be tested and the reference sample image.

[0079] When selecting representative reference samples and textile samples to be tested, it is necessary to ensure that the color values ​​of the reference samples are known and stable, without wear and contamination, as a reliable reference. The surface of the textile samples to be tested should be ensured to be clean. Color histogram, as a common representation of color features, can intuitively show the color area distribution of textile images, and clearly show the proportion and distribution of each color area. It is an effective tool for analyzing color features.

[0080] The specific analysis method for automatically analyzing the first color difference value between the textile image to be tested and the reference sample image is as follows: normalizing the color histograms of the textile image to be tested and the reference sample image; using the histogram intersection method to automatically analyze the similarity value between the color features of the textile image to be tested and the color features of the reference sample image, and using the similarity value between the analyzed color features as the first color difference value between the textile image to be tested and the reference sample image.

[0081] Normalization processing can achieve comparison of the color histograms of the tested textile and the reference sample image on the same scale, eliminate the color intensity differences caused by factors such as lighting conditions, shooting equipment differences or inconsistent image preprocessing, and make the color features of the tested textile and the reference sample image easier to compare; the histogram intersection method is used to automatically analyze the similarity value between the color features of the tested textile image and the color features of the reference sample image, avoiding the subjectivity of manual judgment, and can more objectively and accurately analyze the color difference between the tested textile and the reference sample, thereby improving the degree of automation of the textile color difference analysis method based on image processing, thereby improving the accuracy and efficiency of textile color difference analysis.

[0082] The specific analysis method of analyzing the second color difference value between the textile to be tested and the reference sample by using a colorimeter is as follows: automatically calibrating the colorimeter in black and white, and automatically adjusting the measuring aperture of the colorimeter according to the textile to be tested and the reference sample; under the same standard light source and the same image acquisition space, measuring the color data of the textile to be tested and the reference sample by using the colorimeter, and recording the color data measured by the colorimeter; the colorimeter automatically analyzes the second color difference value between the textile to be tested and the reference sample by measuring the color data.

[0083] Automatically calibrating the colorimeter in black and white can ensure that the colorimeter is in the best working condition before measurement, eliminating measurement errors caused by equipment deviations; automatically recording the color data measured by the colorimeter can save a lot of manpower and time, while reducing the possibility of human errors; automatically analyzing the color difference between the textile to be tested and the reference sample can quickly obtain analysis results and improve measurement efficiency.

[0084] The specific analysis method for analyzing the color difference value of the textile to be tested is: numbering all textiles according to batches; obtaining the first color difference value and the first color difference value of the textile to be tested, and constructing a textile color difference value calculation formula based on them; calculating the textile color difference value according to the textile color difference value calculation formula; the textile color difference value calculation formula is:

[0085] ;

[0086] In the formula, SC i is the textile color difference value of the i-th batch of textiles, i is the textile batch number, TX i is the first color difference value of the i-th batch of textiles, SCJ i is the second color difference value of the i-th batch of textiles, is the first color difference value pair SC i The weight ratio of is the second color difference value pair SC iThe weighted proportion value of the textile color difference value to be tested is compared with the preset color difference threshold value to determine whether the color of the current textile to be tested is qualified.

[0087] Obtained through expert evaluation , The color difference value of textiles can also be obtained through the principle of colorimetry by comparing the color coordinates of the textile to be tested under a specific light source and combining it with the color difference formula for quantitative analysis; the higher the color difference value of the textile, the lower the color consistency between the textile to be tested and the reference sample.

[0088] The specific method for comparing the calculated textile color difference value of the textile to be tested with the preset color difference threshold is as follows: comparing the calculated textile color difference value of the textile to be tested with the preset color difference threshold, automatically judging whether the color of the current textile to be tested is qualified; when the textile color difference value of the textile to be tested is lower than the preset color difference threshold, it indicates that the color of the current textile to be tested is qualified and meets the color requirements of textile production; when the textile color difference value of the textile to be tested is not lower than the preset color difference threshold, it indicates that the color of the current textile to be tested is unqualified and does not meet the color requirements of textile production; outputting the color difference analysis result according to the comparison result between the calculated textile color difference value of the textile to be tested and the preset color difference threshold.

[0089] In the textile industry, color difference range control follows specific industry standards. The textile color difference industry standards clearly stipulate the allowable color difference ranges for various types of textiles, usually expressed as specific values ​​of ΔE, where ΔE is the color difference unit; in order to accurately evaluate the color quality of the textiles to be tested, it is first necessary to consult the relevant industry standards for textiles to obtain the specific color difference range of the textiles to be tested for the category standards, and then communicate in depth with the customers to obtain the specific color difference acceptance range of the customers. A comprehensive analysis is conducted based on the specific color difference range allowed by the standards for the corresponding categories of the textiles to be tested and the specific color difference acceptance range of the customers, and the color difference threshold of the textiles to be tested is preset; the color difference value of the textile to be tested obtained by analysis is automatically compared with the preset color difference threshold.

[0090] Example 2: Figure 2 As shown, a textile color difference analysis system based on image processing is used to implement the method in Example 1, including an image acquisition module, a color feature extraction module, and a color difference analysis module;

[0091] An image acquisition module is used to automatically acquire image data of each batch of textiles, automatically preprocess the acquired textile image data, and automatically analyze the preprocessed textile image data to obtain a textile image acquisition quality index, which is used to evaluate whether the comprehensive quality of the acquired textile image data is qualified;

[0092] A color feature extraction module is used to automatically analyze and compare the pre-processed textile image data after analyzing that the textile image acquisition quality index of the textile to be tested is qualified, and automatically obtain the color features in the textile image; when the textile image acquisition quality index of the textile to be tested is analyzed to be qualified, the textile to be tested is re-transported to the image acquisition location through a fully automatic track to perform image acquisition again;

[0093] The color difference analysis module is used to automatically analyze the first color difference value between the textile image to be tested and the reference sample image according to the color features in the textile image through a color difference analysis method based on color feature matching, analyze the second color difference value between the textile to be tested and the reference sample through a colorimeter, and analyze the textile color difference value of the textile to be tested according to the first color difference value and the second color difference value of the textile to be tested, and the textile color difference value is used to evaluate the color consistency between the textile to be tested and the reference sample.

[0094] In summary, with the help of the above-mentioned technical scheme of the present invention, by automatically collecting and preprocessing textile image data, the preprocessed textile image data is automatically color analyzed and compared to obtain the color features in the textile image, and then the color difference value between the textile image to be tested and the reference sample image is automatically analyzed through a color difference analysis method based on color feature matching, thereby achieving an improvement in the degree of automation of textile color difference analysis.

Claims

1. A textile color difference analysis method based on image processing, characterized in that: The following steps are involved: Automatically collect the textile image data of each batch, perform automatic image preprocessing on the collected textile image data, and automatically analyze the preprocessed textile image data to obtain a textile image collection quality index, which is used to evaluate whether the comprehensive quality of the collected textile image data is qualified; When the textile image acquisition quality index of the textile to be tested is found to be qualified, the pre-processed textile image data is automatically subjected to color analysis and comparison to automatically obtain color features in the textile image; By using a color difference analysis method based on color feature matching, a first color difference value between the image of the textile to be tested and the image of the reference sample is automatically analyzed according to the color features in the textile image, a second color difference value between the textile to be tested and the reference sample is analyzed by a colorimeter, and a textile color difference value of the textile to be tested is analyzed according to the first color difference value and the second color difference value of the textile to be tested, and the textile color difference value is used to evaluate the color consistency between the textile to be tested and the reference sample; The specific method for analyzing the textile image acquisition quality index is: Number all textiles by batch; The images of the textile to be tested and the reference sample are converted into the frequency domain by Fourier transform, and the energy of the high-frequency components is analyzed to obtain the image clarity of the textile to be tested and the reference sample images; Importing the images of the textile to be tested and the reference sample, directly generating the image histograms of the textile to be tested and the reference sample images through image processing software, and analyzing and obtaining the color restoration degree of the textile to be tested image by comparing the similarity of the histograms of the textile to be tested and the reference sample images; Using a median filter to filter the collected textile image to be tested, analyzing the difference between the filtered collected textile image to be tested and the original image, and then obtaining the noise level of the textile image to be tested; By comparing the color changes of the collected continuous images of the textile to be tested with a light source stability test instrument, the light source fluctuation of the textile to be tested image is obtained; The image clarity, color reproduction, noise level and light source fluctuation of the textile image acquisition data are obtained, and the calculation formula of the textile image acquisition quality index is constructed based on them, which is expressed as: ； In the formula, FCJ i is the textile image acquisition quality index of the i-th batch of textiles, i is the textile batch number, i=1, 2, ..., N, N is the total number of textile batches, QX i is the image clarity of the images collected for the i-th batch of textiles, HY i is the color restoration degree of the images collected for the i-th batch of textiles, ZS i is the noise level of the images collected for the i-th batch of textiles, GY i is the light source fluctuation degree of the image collected for the i-th batch of textiles, α1 is the image clarity effect on FCJ i The weight ratio of α2 is obtained by objective weight method, and α2 is the color restoration degree to FCJ i The weight ratio of α3 is obtained by objective weight method, and α3 is the noise level to FCJ i The weight ratio of α4 is the effect of light source fluctuation on FCJ i The weight ratio of is obtained by objective weight method, and e is a natural constant; The textile image acquisition quality index is calculated according to the textile image acquisition quality index calculation formula; When the image acquisition quality index of the acquired textile image to be tested is higher than the first threshold, it indicates that the comprehensive quality of the acquired textile image data is qualified; When the image acquisition quality index of the acquired textile image to be tested is not higher than the first threshold, it indicates that the comprehensive quality of the acquired textile image data is unqualified.

2. The textile color difference analysis method based on image processing according to claim 1, characterized in that: The specific method of automatically collecting each batch of textile image data is as follows: All textiles are transported to the image acquisition space in batches through a fully automatic track to complete the textile image data acquisition of the textiles to be tested; Obtain the theoretical light source and theoretical image acquisition space of the reference sample; wherein the image acquisition space has no interference from reflected light, and the light source and image acquisition space used for image acquisition of the product to be tested are consistent with the theoretical light source and theoretical image acquisition space of the reference sample.

3. The textile color difference analysis method based on image processing according to claim 1, characterized in that: The specific method for performing automatic image preprocessing on the collected textile image data is: The color textile image is converted into a grayscale image by weighted averaging method; The texture and defect information of the textile image is obtained, and the cropping area of ​​the textile image is set through an image cropping algorithm to complete the cropping of the textile image; when cropping, the texture and defect information of the textile image is retained, and the area without texture and defect information is cropped.

4. The textile color difference analysis method based on image processing according to claim 1, characterized in that: The specific method for automatically analyzing and comparing the pre-processed textile image data is as follows: By using a color space conversion algorithm, the pixels of the textile and the reference sample are classified into different color areas according to the grayscale values ​​of the pixels of the preprocessed grayscale images of the textile to be tested and the reference sample; The number of pixels in each color area of ​​the images of the textile to be tested and the reference sample is counted to generate a color histogram of the textile to be tested and the reference sample image. The generated color histogram is used as the color feature of the textile to be tested and the reference sample image.

5. The textile color difference analysis method based on image processing according to claim 1, characterized in that: The specific analysis method for automatically analyzing the first color difference value between the textile image to be tested and the reference sample image is: Normalizing the color histograms of the tested textile and the reference sample images; The histogram intersection method is used to automatically analyze the similarity value between the color features of the textile image to be tested and the color features of the reference sample image, and the similarity value between the analyzed color features is used as the first color difference value between the textile image to be tested and the reference sample image.

6. The textile color difference analysis method based on image processing according to claim 1, characterized in that: The specific analysis method of analyzing the second color difference value between the textile to be tested and the reference sample by using a colorimeter is: Automatically calibrate the colorimeter for black and white, and automatically adjust the aperture of the colorimeter according to the textile to be tested and the reference sample; Under the same standard light source and the same image acquisition space, the color data of the textile to be tested and the reference sample are measured by a colorimeter, and the color data measured by the colorimeter is recorded; The colorimeter automatically analyzes the second color difference value between the textile to be tested and the reference sample by measuring the color data.

7. The textile color difference analysis method based on image processing according to claim 1, characterized in that: The specific analysis method for analyzing the color difference value of the textile to be tested is: Number all textiles by batch; The first color difference value and the second color difference value of the textile to be tested are obtained, and a calculation formula for the color difference value of the textile is constructed based on the first color difference value and the second color difference value, which is expressed as: ； In the formula, SC i is the textile color difference value of the i-th batch of textiles, i is the textile batch number, i=1, 2, ..., N, N is the total number of textile batches, TX i is the first color difference value of the i-th batch of textiles, SCJ i is the second color difference value of the i-th batch of textiles, is the first color difference value pair SC i The weight ratio of is the second color difference value pair SC i The weight ratio of Calculate the color difference value of textiles according to the calculation formula of color difference value of textiles; The calculated color difference value of the textile to be tested is compared with a preset color difference threshold to determine whether the color of the current textile to be tested is qualified.

8. The textile color difference analysis method based on image processing according to claim 7, characterized in that: The specific method of comparing the calculated textile color difference value of the textile to be tested with the preset color difference threshold is: Compare the calculated color difference value of the textile to be tested with the preset color difference threshold value to automatically determine whether the color of the current textile to be tested is qualified; If the color difference value of the textile to be tested is lower than the preset color difference threshold, it means that the color of the current textile to be tested is qualified and meets the color requirements of textile production; If the color difference value of the textile to be tested is not lower than the preset color difference threshold, it means that the color of the current textile to be tested is unqualified and does not meet the color requirements of textile production; The color difference value of the textile to be tested is compared with the preset color difference threshold value, and the color difference analysis result is output.

9. A textile color difference analysis system based on image processing, used to implement the textile color difference analysis method based on image processing according to any one of claims 1 to 8, characterized in that: It includes image acquisition module, color feature extraction module and color difference analysis module; An image acquisition module is used to automatically acquire image data of each batch of textiles, automatically preprocess the acquired textile image data, and automatically analyze the preprocessed textile image data to obtain a textile image acquisition quality index, which is used to evaluate whether the comprehensive quality of the acquired textile image data is qualified; A color feature extraction module is used to automatically analyze and compare the pre-processed textile image data after analyzing that the textile image acquisition quality index of the textile to be tested is qualified, and automatically obtain the color features in the textile image; when the textile image acquisition quality index of the textile to be tested is analyzed to be qualified, the textile to be tested is re-transported to the image acquisition location through a fully automatic track to perform image acquisition again; The color difference analysis module is used to automatically analyze the first color difference value between the textile image to be tested and the reference sample image according to the color features in the textile image through a color difference analysis method based on color feature matching, analyze the second color difference value between the textile to be tested and the reference sample through a colorimeter, and analyze the textile color difference value of the textile to be tested according to the first color difference value and the second color difference value of the textile to be tested, and the textile color difference value is used to evaluate the color consistency between the textile to be tested and the reference sample.

Images