A method, system and storage medium for determining the dyeing uniformity of socks

Abstract

The present invention relates to the technical field of sock dyeing recognition, in particular to a method, system and storage medium for determining the dyeing uniformity of socks; randomly sampling to obtain an initial sock sample; dividing the area of the initial sock sample and taking pictures to obtain an initial sample image set; constructing a color difference recognition model to recognize the initial sample image set to obtain initial difference data; performing a stretching test on the initial sock sample to obtain a stretched sock sample and a stretched sample image set; recognizing the stretched sample image set to obtain stretched difference data; performing a movement test on the stretched sock sample to obtain a movement sample image set; recognizing the movement sample image set to obtain movement difference data; constructing a dyeing uniformity recognition model to recognize the initial difference data, the stretched difference data and the movement difference data to obtain a sock dyeing uniformity coefficient. The present invention realizes the accurate rating of the sock dyeing uniformity through the sock dyeing uniformity coefficient.

Description

Technical Field

[0001] The present invention relates to the technical field of sock dyeing recognition, and specifically to a method, system and storage medium for judging the dyeing uniformity of socks. Background Art

[0002] A sock is a piece of clothing worn on the feet, usually made of soft fabric material; it can reduce the direct friction of shoes on the feet and protect the feet from injury; at the same time, it can absorb the sweat of the feet and reduce the risk of bacterial growth; and it has gradually become an essential daily necessity. With the continuous development of sock production technology, its types have gradually become complex and diverse, including short socks, mid-calf socks, knee-high socks and various functional socks, meeting the different needs of consumers.

[0003] In the production of socks, its dyeing process is an important part; due to factors such as raw materials, dyeing process, equipment performance, operating factors and external environment, it will have a certain impact on the dyeing uniformity of socks, making the actual dyeing effect of socks inconsistent with the color distribution in the design scheme, resulting in uneven dyeing. This will not only affect the quality of the product, but may also have a negative effect on the consumer experience and brand image.

[0004] Therefore, it is necessary to study the dyeing uniformity of sock products to strengthen the quality control of sock production and improve the dyeing uniformity and product quality.

[0005] With the development of image recognition technology, it has gradually been applied to the recognition of sock dyeing uniformity; by obtaining the image of the sock product and performing image recognition to measure its dyeing uniformity to improve efficiency; however, the focus of image recognition technology is to recognize the color distribution on the surface of the sock in the initial state, without considering the changes in fiber density and fiber arrangement caused by fabric stretching during the wearing process of the sock, resulting in inaccurate recognition accuracy.

[0006] Therefore, a method, system and storage medium for judging the dyeing uniformity of socks are proposed. Summary of the Invention

[0007] The object of the present invention is to provide a method, system and storage medium for judging the dyeing uniformity of socks; randomly sampling to obtain an initial sock sample; dividing the initial sock sample into regions and taking pictures to obtain an initial sample image set; constructing a color difference recognition model to recognize the initial sample image set to obtain initial difference data; performing a stretching test on the initial sock sample to obtain a stretched sock sample and a stretched sample image set; recognizing the stretched sample image set to obtain stretched difference data; performing a movement test on the stretched sock sample to obtain a movement sample image set; recognizing the movement sample image set to obtain movement difference data; constructing a dyeing uniformity recognition model to recognize the initial difference data, the stretched difference data and the movement difference data to obtain a sock dyeing uniformity coefficient; and accurately rating the sock dyeing uniformity through the sock dyeing uniformity coefficient.

[0008] To achieve the above object, the present invention provides the following technical solutions;

[0009] A method for judging the dyeing uniformity of socks, comprising:

[0010] S10. Collect sock product data to obtain knitting process data and standard color distribution data; randomly sample to obtain an initial sock sample;

[0011] S20. Divide the initial sock sample into multiple functional regions according to the knitting process data, including a sole region, a toe region, a instep region and a sock mouth region; take pictures of the functional regions to obtain an initial sample image set;

[0012] S30. Construct a color difference recognition model to recognize the initial sample image set, segment and detect the images of the functional regions through the standard color distribution data to obtain initial difference data;

[0013] S40. Determine stretching data according to the size data of the initial sock sample, perform a stretching test on the initial sock sample according to the stretching data to obtain a stretched sock sample; take pictures of the functional regions of the stretched sock sample to obtain a stretched sample image set;

[0014] S50. Recognize the stretched sample image set through the color difference recognition model to obtain stretched difference data;

[0015] S60. Screen users according to the size data of the initial sock sample to determine a first user; perform a movement test with the first user and the stretched sock sample to obtain a movement sock sample; take pictures of the functional regions of the movement sock sample to obtain a movement sample image set;

[0016] S70. Recognize the movement sample image set through the color difference recognition model to obtain movement difference data;

[0017] S80. Construct a model for identifying uniform dyeing, identify the initial difference data, stretching difference data, and motion difference data of the same functional area, and obtain the color anomaly coefficient; then identify the sock dyeing uniformity coefficient through the color distribution anomaly coefficient and conduct grade evaluation.

[0018] The knitting process data includes the three-dimensional shape data and textile structure data of the socks; the initial sock samples are divided into a sole area, a toe area, a instep area, and a cuff area through the knitting process data;

[0019] The initial sample image set includes a first sole area image, a first toe area image, a first instep area image, and a first cuff area image; the stretching sample image set includes a second sole area image, a second toe area image, a second instep area image, and a second cuff area image; the motion sample image set includes a third sole area image, a third toe area image, a third instep area image, and a third cuff area image.

[0020] The color difference identification model includes a sock data acquisition layer, a sock image segmentation layer, and a sock color identification layer;

[0021] The sock data acquisition layer is used to acquire the standard color distribution data of the socks and the sample image set, where the sample image set includes sole area images, toe area images, instep area images, and cuff area images;

[0022] The sock image segmentation layer segments the functional area images in the sample image set according to the standard color distribution data, divides the sub-regions belonging to the same color distribution in the area images into one category, and obtains the area image subset;

[0023] The sock color identification layer is used to identify the color distribution data in the area image subset to obtain the area subset color data; identify the difference area according to the area subset color data and the standard color distribution data; obtain the difference coefficient according to the data difference between the color distribution data of the difference area and the corresponding standard color distribution data; output the difference area and the difference coefficient as difference data.

[0024] The process of the stretching test is as follows:

[0025] Obtain the size data of the initial sock samples; determine the foot shape data of the user according to the size data, and use the foot shape data as the stretching data;

[0026] Determine the foot mold parameters according to the stretching data, and obtain an experimental foot mold according to the foot mold parameters; wear the initial sock samples on the experimental foot mold to obtain stretched sock samples;

[0027] Take pictures of the stretched sock samples through a photographing device to obtain a stretching sample image set.

[0028] The process of the motion test is as follows:

[0029] Determine the foot type data of the user according to the size data; identify the user according to the foot type data, and take the user who meets the foot type data as the first user;

[0030] Identify the process of the first user wearing the stretch sock sample during exercise, and take the sock sample during exercise as the exercise sock sample; the exercise types include walking, running, jumping and spinning;

[0031] Take pictures of the functional areas of the exercise sock sample to obtain a set of exercise sample images;

[0032] In the set of exercise sample images, there are a walking image subset, a running image subset, a jumping image subset and a spinning image subset; according to the identification of each image subset, obtain the walking difference data, running difference data, jumping difference data and spinning difference data; then, according to the comprehensive identification of the walking difference data, running difference data, jumping difference data and spinning difference data, obtain the exercise difference data.

[0033] The dyeing uniformity identification model includes a first difference identification layer, a second difference identification layer, a third difference identification layer and a dyeing uniformity identification layer;

[0034] The first difference identification layer is used to identify the initial difference data of the functional area to obtain the initial difference area and initial difference coefficient of the functional area; according to the ratio of the area of the initial difference area to the area of the corresponding functional area, weight the initial difference coefficient to obtain the initial color anomaly coefficient of the functional area;

[0035] The second difference identification layer is used to identify the stretch difference data of the functional area to obtain the stretch difference area and stretch difference coefficient of the functional area; according to the ratio of the area of the stretch difference area to the area of the corresponding functional area, weight the stretch difference coefficient to obtain the stretch color anomaly coefficient of the functional area;

[0036] The third difference identification layer is used to identify the exercise difference data of the functional area to obtain the exercise difference area and exercise difference coefficient of the functional area; according to the ratio of the area of the exercise difference area to the area of the corresponding functional area, weight the exercise difference coefficient to obtain the exercise color anomaly coefficient of the functional area;

[0037] The dyeing uniformity identification layer weights and identifies according to the initial color anomaly coefficient, stretch color anomaly coefficient and exercise color anomaly coefficient of each functional area to obtain the sock dyeing uniformity coefficient.

[0038] A sock dyeing uniformity determination system includes:

[0039] A sample sampling module collects sock product data to obtain knitting process data and standard color distribution data; through random sampling, an initial sock sample is obtained;

[0040] An initial image acquisition module divides the initial sock sample into multiple functional regions according to the knitting process data, including a sole region, a toe region, a dorsal region, and a cuff region; takes pictures of the functional regions to obtain an initial sample image set;

[0041] An initial image recognition module constructs a color difference recognition model to recognize the initial sample image set, segments and detects the images of the functional regions through the standard color distribution data, and obtains initial difference data;

[0042] A stretched image acquisition module determines stretching data according to the size data of the initial sock sample, performs a stretching test on the initial sock sample according to the stretching data to obtain a stretched sock sample; takes pictures of the functional regions of the stretched sock sample to obtain a stretched sample image set;

[0043] A stretched image recognition module recognizes the stretched sample image set through the color difference recognition model to obtain stretched difference data;

[0044] A motion image acquisition module screens users according to the size data of the initial sock sample to determine a first user; performs a motion test with the first user and the stretched sock sample to obtain a motion sock sample; takes pictures of the functional regions of the motion sock sample to obtain a motion sample image set;

[0045] A motion image recognition module recognizes the motion sample image set through the color difference recognition model to obtain motion difference data;

[0046] A dyeing grade evaluation module constructs a dyeing uniformity recognition model, recognizes the initial difference data, stretched difference data, and motion difference data of the same functional region to obtain a color anomaly coefficient; then obtains a sock dyeing uniformity coefficient through the color distribution anomaly coefficient recognition for grade evaluation.

[0047] The knitting process data includes three-dimensional shape data and textile structure data of the sock; the initial sock sample is divided into a sole region, a toe region, a dorsal region, and a cuff region through the knitting process data;

[0048] The initial sample image set includes a first sole region image, a first toe region image, a first dorsal region image, and a first cuff region image; the stretched sample image set includes a second sole region image, a second toe region image, a second dorsal region image, and a second cuff region image; the motion sample image set includes a third sole region image, a third toe region image, a third dorsal region image, and a third cuff region image.

[0049] The color difference recognition model includes a sock data acquisition layer, a sock image segmentation layer, and a sock color recognition layer;

[0050] The sock data acquisition layer is used to obtain the standard color distribution data of the sock and a sample image set, where the sample image set includes plantar region images, toe region images, dorsal region images, and sock mouth region images;

[0051] The sock image segmentation layer segments the functional region images in the sample image set according to the standard color distribution data, divides the sub-regions belonging to the same color distribution in the region images into one category, and obtains a region image subset;

[0052] The sock color recognition layer is used to recognize the color distribution data in the region image subset to obtain region subset color data; recognize the difference region according to the region subset color data and the standard color distribution data; obtain a difference coefficient according to the data difference between the color distribution data of the difference region and the corresponding standard color distribution data; and output the difference region and the difference coefficient as difference data.

[0053] A storage medium stores a computer program thereon, and when the computer program is executed by a processor, it implements the described method for determining the dyeing uniformity of socks.

[0054] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0055] 1. The present invention constructs a color difference recognition model to recognize the obtained sample image set; obtains the images of each functional region in the sample image set, divides the sub-regions of the functional region according to the standard color distribution data, divides the sub-regions with the same color distribution into one category, and obtains a region image subset; recognizes the difference region and the difference coefficient through the difference between the color distribution in the region image subset and the standard color distribution data, and accurately recognizes the uneven regions in the initial state.

[0056] 2. The present invention determines the foot type data of the user according to the size of the sock, and uses the foot type data as stretching data to perform a stretching test on the initial sock sample to obtain a stretched sock sample and a stretched sample image set; then determines the first user according to the size data of the sock, and performs a motion test according to the first user and the stretched sock sample to obtain a motion sample image set; accurately obtains the color distribution data of the sock in the stretched state and the motion state.

[0057] 3. The present invention constructs a dyeing uniformity recognition model to recognize the initial difference data of each functional area, the stretching difference data in the stretching state, and the motion difference data in the motion state. The difference coefficient is weighted by the ratio of the area of the difference region to the area of the corresponding functional area to obtain the initial color abnormality coefficient, the stretching color abnormality coefficient, and the motion color abnormality coefficient of each functional area. Then, through recognition, the dyeing uniformity coefficient of the sock is obtained, which accurately measures the dyeing uniformity of the sock. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a schematic flow chart of a method for determining the dyeing uniformity of socks according to the present invention;

[0059] Figure 2 It is a schematic structural diagram of the color difference recognition model of the present invention;

[0060] Figure 3 It is a schematic structural diagram of the dyeing uniformity recognition model of the present invention;

[0061] Figure 4 It is a schematic structural diagram of a system for determining the dyeing uniformity of socks according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0062] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0063] During the wearing process of socks, the socks will be stretched to a certain extent, and with the changes in the user's movement behaviors and patterns, the stretching amplitude and stretching state of the socks will also change; during the stretching process of the socks, the fiber density and arrangement pattern of the sock fabric will change, resulting in changes in the overall color distribution of the fabric; therefore, it is necessary to recognize the color distribution during the wearing process of the socks.

[0064] For this reason, a method, a system, and a storage medium for determining the dyeing uniformity of socks are proposed.

[0065] Embodiment 1

[0066] A method for determining the dyeing uniformity of socks according to the present invention has a process as Figure 1 shown and includes:

[0067] S10. Collect sock product data to obtain knitting process data and standard color distribution data; through random sampling, obtain initial sock samples.

[0068] S20. Divide the initial sock sample into multiple functional regions according to the knitting process data, including the sole region, toe region, instep region, and cuff region; photograph the functional regions to obtain an initial sample image set.

[0069] Divide the initial sock sample into a sole region, a toe region, an instep region, and a cuff region through the knitting process data; the knitting process data includes the three-dimensional shape data and textile structure data of the sock;

[0070] The initial sample image set includes a first sole region image, a first toe region image, a first instep region image, and a first cuff region image.

[0071] Among them, the same style means that the knitting process data and standard color distribution data of the sock are consistent; the textile structure data refers to the fabric structure formed by the combination of yarns. The standard color distribution data is obtained through the dyeing design scheme of the sock product.

[0072] The present invention obtains the knitting process data of the sock product, including the three-dimensional shape data and textile structure data of the sock; divides the sock into regions according to the knitting process data to obtain a sole region, a toe region, an instep region, and a cuff region; and performs differential identification according to different regions of the sock to improve the accuracy of identification.

[0073] S30. Construct a color difference recognition model to recognize the initial sample image set, segment and detect the images of the functional regions through the standard color distribution data to obtain initial difference data.

[0074] The color difference recognition model is constructed according to a deep neural network model, and its structure is as Figure 2 shown, including a sock data acquisition layer, a sock image segmentation layer, and a sock color recognition layer;

[0075] The sock data acquisition layer is used to obtain the standard color distribution data and sample image set of the sock, where the sample image set includes a sole region image, a toe region image, an instep region image, and a cuff region image;

[0076] The sock image segmentation layer segments the images of the functional regions in the sample image set according to the standard color distribution data, divides the sub-regions belonging to the same color distribution in the region image into one category to obtain a region image subset;

[0077] The sock color recognition layer is used to recognize the color distribution data in the regional image subset to obtain the regional subset color data; identify the difference region according to the regional subset color data and the standard color distribution data; obtain the difference coefficient according to the data difference between the color distribution data of the difference region and the corresponding standard color distribution data; and output the difference region and the difference coefficient as difference data.

[0078] Among them, the calculation formula of the difference coefficient is:

[0079] ;

[0080] Wherein, represents the difference coefficient of the difference region; represents the pixel points of the difference region; 、 and represent the color data of the pixel points; 、 and represent the standard color distribution data corresponding to the pixel points; and are RGB color model data.

[0081] The present invention constructs a color difference recognition model to recognize the obtained sample image set; obtains the images of each functional region in the sample image set, divides the functional region into sub-regions according to the standard color distribution data, divides the sub-regions with the same color distribution into one category to obtain a regional image subset; and identifies the difference region and the difference coefficient through the difference between the color distribution in the regional image subset and the standard color distribution data.

[0082] S40. Determine the stretching data according to the size data of the initial sock sample, perform a stretching test on the initial sock sample according to the stretching data to obtain a stretched sock sample; photograph the functional region of the stretched sock sample to obtain a stretched sample image set.

[0083] The process of the stretching test is as follows:

[0084] Obtain the size data of the initial sock sample; determine the foot shape data of the user according to the size data, and use the foot shape data as the stretching data;

[0085] Determine the foot mold parameters according to the stretching data, and obtain an experimental foot mold according to the foot mold parameters; wear the initial sock sample on the experimental foot mold to obtain a stretched sock sample;

[0086] Photograph the stretched sock sample through a photographing device to obtain a stretched sample image set.

[0087] The present invention determines the foot shape data of a user based on the size data of the socks, and uses the foot shape data as stretching data; determines the matching foot mold parameters according to the stretching data to obtain an experimental foot mold; performs a stretching test on the initial sock sample according to the experimental foot mold, and captures a set of stretched sample images to accurately obtain the color distribution data of the socks during wearing.

[0088] S50. The stretching difference data is obtained by identifying the set of stretched sample images through a color difference identification model.

[0089] S60. Screen users according to the size data of the initial sock sample to determine the first user; conduct a motion test with the first user and the stretched sock sample to obtain a motion sock sample; capture the functional areas of the motion sock sample to obtain a set of motion sample images;

[0090] The process of the motion test is as follows:

[0091] Determine the foot shape data of the user according to the size data; identify the user according to the foot shape data, and use the user who conforms to the foot shape data as the first user;

[0092] Identify the process of the first user wearing the stretched sock sample during motion, and use the sock sample during motion as the motion sock sample; wherein, the motion types include walking, running, jumping, and spinning;

[0093] Capture the functional areas of the motion sock sample to obtain a set of motion sample images.

[0094] The present invention determines the foot shape data of the user according to the size data of the socks, screens the user group according to the foot shape data to obtain the first user; identifies the process of the first user wearing the stretched sock sample during motion, and captures a set of motion sample images to accurately obtain the color distribution data of the socks during movement.

[0095] S70. Identify the set of motion sample images through a color difference identification model to obtain motion difference data;

[0096] In the set of motion sample images, there are a walking image subset, a running image subset, a jumping image subset, and a spinning image subset; according to the identification of each image subset, walking difference data, running difference data, jumping difference data, and spinning difference data are obtained; then, based on the comprehensive identification of the walking difference data, running difference data, jumping difference data, and spinning difference data, motion difference data is obtained.

[0097] S80. Construct a dyeing uniformity identification model to identify the initial difference data, stretching difference data, and motion difference data of the same functional area to obtain a color anomaly coefficient; then, identify the sock dyeing uniformity coefficient through the color distribution anomaly coefficient for grade evaluation.

[0098] The described dyeing uniformity recognition model is constructed based on a deep neural network model, and its structure is as Figure 3 shown, including a first difference recognition layer, a second difference recognition layer, a third difference recognition layer, and a dyeing uniformity recognition layer;

[0099] The first difference recognition layer is used to recognize the initial difference data of the functional area, obtaining the initial difference area and the initial difference coefficient of the functional area; according to the ratio of the area of the initial difference area to the area of the corresponding functional area, the initial difference coefficient is weighted to obtain the initial color abnormality coefficient of the functional area;

[0100] The second difference recognition layer is used to recognize the stretching difference data of the functional area, obtaining the stretching difference area and the stretching difference coefficient of the functional area; according to the ratio of the area of the stretching difference area to the area of the corresponding functional area, the stretching difference coefficient is weighted to obtain the stretching color abnormality coefficient of the functional area;

[0101] The third difference recognition layer is used to recognize the motion difference data of the functional area, obtaining the motion difference area and the motion difference coefficient of the functional area; according to the ratio of the area of the motion difference area to the area of the corresponding functional area, the motion difference coefficient is weighted to obtain the motion color abnormality coefficient of the functional area;

[0102] The dyeing uniformity recognition layer performs weighted recognition based on the initial color abnormality coefficient, the stretching color abnormality coefficient, and the motion color abnormality coefficient of each functional area to obtain the sock dyeing uniformity coefficient.

[0103] The present invention constructs a dyeing uniformity recognition model to recognize the initial difference data of each functional area, the stretching difference data in the stretching state, and the motion difference data in the motion state, and weights its difference coefficient according to the ratio of the area of the difference area to the area of the corresponding functional area to obtain the initial color abnormality coefficient, the stretching color abnormality coefficient, and the motion color abnormality coefficient of each functional area; and then obtains the sock dyeing uniformity coefficient through recognition, accurately measuring the dyeing uniformity of the socks.

[0104] The present invention proposes a storage medium, on which a computer program is stored for implementing the described method for determining the dyeing uniformity of socks.

[0105] The present invention randomly samples sock products of the same style to obtain an initial sock sample; divides and photographs the initial sock sample to obtain an initial sample image set; constructs a color difference recognition model to recognize the initial sample image set to obtain initial difference data; conducts a stretching test on the initial sock sample to obtain a stretched sock sample and a stretched sample image set; recognizes the stretched sample image set to obtain stretching difference data; conducts a movement test on the stretched sock sample to obtain a movement sample image set; recognizes the movement sample image set to obtain movement difference data; constructs a dyeing uniformity recognition model to recognize the initial difference data, stretching difference data, and movement difference data to obtain a sock dyeing uniformity coefficient. The present invention accurately rates the sock dyeing uniformity through the sock dyeing uniformity coefficient.

[0106] Embodiment 2

[0107] Sock processing factory R is engaged in the production and sales of socks. Among them, the dyeing uniformity of socks is an important factor affecting the quality and sales volume of socks; in order to understand the dyeing uniformity of sock products for the update and iteration of production technology; sock processing factory R uses a sock dyeing uniformity determination system described in the present invention for recognition.

[0108] A sock dyeing uniformity determination system proposed by the present invention has a structure as Figure 4 shown, including: a sample sampling module, an initial image acquisition module, an initial image recognition module, a stretching image acquisition module, a stretching image recognition module, a movement image acquisition module, a movement image recognition module, and a dyeing grade evaluation module.

[0109] The sample sampling module collects sock product data to obtain knitting process data and standard color distribution data; through random sampling, an initial sock sample is obtained.

[0110] Sock processing factory R takes sock product A as the research object, randomly samples products A produced in the same batch and with consistent size data to obtain an initial sock sample.

[0111] The initial image acquisition module divides the initial sock sample into multiple functional areas according to the knitting process data, including a sole area, a toe area, a dorsal area, and a cuff area; photographs the functional areas to obtain an initial sample image set.

[0112] Among them, the same style means that the knitting process data and color distribution data of the socks are consistent; the textile structure data refers to the fabric structure formed by the combination of yarns.

[0113] Divide the initial sock sample into a sole area, a toe area, a dorsal area, and a cuff area according to knitting process data; wherein, the sole area refers to the bottom of the sock, covering the front half of the sole and the heel; the toe area refers to the front end of the sock, covering the toe area; the dorsal area refers to the middle part of the sock, covering the top of the foot; the cuff area refers to the top of the sock, covering the calf and the ankle.

[0114] Take pictures of the sole area, toe area, dorsal area, and cuff area of the initial sock sample to obtain an initial sample image set; the initial sample image set includes a first sole area image, a first toe area image, a first dorsal area image, and a first cuff area image.

[0115] An initial image recognition module constructs a color difference recognition model to recognize the initial sample image set, and segments and detects the images of the functional areas through standard color distribution data to obtain initial difference data.

[0116] The color difference recognition model includes a sock data acquisition layer, a sock image segmentation layer, and a sock color recognition layer;

[0117] The sock data acquisition layer is used to obtain the standard color distribution data of the sock and the sample image set, wherein the sample image set includes a sole area image, a toe area image, a dorsal area image, and a cuff area image;

[0118] The sock image segmentation layer segments the images of the functional areas in the sample image set according to the standard color distribution data, and divides the sub-regions belonging to the same color distribution in the area image into one category to obtain an area image subset;

[0119] The sock color recognition layer is used to recognize the color distribution data in the area image subset to obtain area subset color data; recognize the difference area according to the area subset color data and the standard color distribution data; obtain a difference coefficient according to the data difference between the color distribution data of the difference area and the corresponding standard color distribution data; output the difference area and the difference coefficient as difference data.

[0120] The present invention constructs a color difference recognition model to recognize the obtained sample image set; obtains the images of each functional area in the sample image set, divides the sub-regions of the functional areas according to the standard color distribution data, and divides the sub-regions with the same color distribution into one category to obtain an area image subset; recognizes the difference area and the difference coefficient through the difference between the color distribution in the area image subset and the standard color distribution data.

[0121] The calculation formula of the difference coefficient is as follows:

[0122] ;

[0123] Among them, represents the difference coefficient of the difference region; represents the pixel points of the difference region; , and represent the color data of the pixel points; , and represent the standard color distribution data corresponding to the pixel points, which are RGB color model data.

[0124] Based on the standard color distribution data of sock product A, the color distribution of each functional area is identified and regionally divided, and the classification results are shown in Table 1.

[0125] Table 1 Data table for regional division of color distribution of sock product A

[0126]

[0127] From Table 1, it can be seen that the sole area of sock product A includes a total of two color distributions, which are respectively distributed in the first sub-region set of the sole and the second sub-region set of the sole; there is only one color distribution in the toe area of sock product A; there are three colors distributed in the instep area of sock product A; there is one color distributed in the cuff area of sock product A.

[0128] The stretching image acquisition module determines the stretching data according to the size data of the initial sock sample, conducts a stretching test on the initial sock sample according to the stretching data, and obtains a stretched sock sample; takes pictures of the functional areas of the stretched sock sample to obtain a set of stretched sample images.

[0129] The stretching test process of sock product A is as follows:

[0130] Obtain the size data of the initial sock sample; determine the foot shape data of the user according to the size data, and use the foot shape data as the stretching data;

[0131] Determine the foot mold parameters according to the stretching data, and obtain an experimental foot mold according to the foot mold parameters; wear the initial sock sample on the experimental foot mold to obtain a stretched sock sample;

[0132] Take pictures of the stretched sock sample through the photographing device to obtain a set of stretched sample images.

[0133] Among them, obtain the division method of each functional area of the initial sock sample and identify the division boundary data; after obtaining the stretched sock sample through the stretching test, there is a stretching change in the stretched sock sample relative to the initial sock sample; at this time, divide the second sample according to the division boundary data of the first sample.

[0134] The stretching image recognition module uses a color difference recognition model to recognize the stretching sample image set and obtain stretching difference data.

[0135] The motion image acquisition module screens users according to the size data of the initial sock samples to determine the first user; conducts motion tests with the first user and the stretching sock samples to obtain motion sock samples; and takes pictures of the functional areas of the motion sock samples to obtain a set of motion sample images.

[0136] The process of the motion test for sock product A is as follows:

[0137] Determine the foot shape data of the user according to the size data; identify the user according to the foot shape data, and take the user who meets the foot shape data as the first user;

[0138] Identify the process of the first user wearing the stretching sock sample during motion, and take the sock sample during motion as the motion sock sample; the motion types include walking, running, jumping, and spinning;

[0139] Take pictures of the functional areas of the motion sock samples to obtain a set of motion sample images.

[0140] For the images of the sole area of the sock, they are obtained by having the first user move on a transparent glass plate.

[0141] The motion image recognition module uses a color difference recognition model to recognize the set of motion sample images and obtain motion difference data.

[0142] In the set of motion sample images, there are a walking image subset, a running image subset, a jumping image subset, and a spinning image subset; according to the recognition of each image subset, walking difference data, running difference data, jumping difference data, and spinning difference data are obtained; and then, based on the comprehensive recognition of the walking difference data, running difference data, jumping difference data, and spinning difference data, motion difference data is obtained.

[0143] Among them, comprehensive recognition means setting weights according to different motion types and performing weighted processing on the difference data; the weights of different motion types can be set according to the motion time distribution of the first user during the motion test.

[0144] The dyeing grade evaluation module constructs a dyeing uniformity recognition model to recognize the initial difference data, stretching difference data, and motion difference data of the same functional area, and obtain the color anomaly coefficient; then, through the recognition of the color distribution anomaly coefficient, the sock dyeing uniformity coefficient is obtained for grade evaluation.

[0145] The dyeing uniformity recognition model includes a first difference recognition layer, a second difference recognition layer, a third difference recognition layer, and a dyeing uniformity recognition layer;

[0146] The first difference recognition layer is used to recognize the initial difference data of the functional area to obtain the initial difference area and the initial difference coefficient of the functional area; according to the ratio of the area of the initial difference area to the area of the corresponding functional area, the initial difference coefficient is weighted to obtain the initial color anomaly coefficient of the functional area.

[0147] The second difference recognition layer is used to recognize the stretching difference data of the functional area to obtain the stretching difference area and the stretching difference coefficient of the functional area; according to the ratio of the area of the stretching difference area to the area of the corresponding functional area, the stretching difference coefficient is weighted to obtain the stretching color anomaly coefficient of the functional area.

[0148] The third difference recognition layer is used to recognize the motion difference data of the functional area to obtain the motion difference area and the motion difference coefficient of the functional area; according to the ratio of the area of the motion difference area to the area of the corresponding functional area, the motion difference coefficient is weighted to obtain the motion color anomaly coefficient of the functional area.

[0149] The dyeing uniformity recognition layer performs weighted recognition based on the initial color anomaly coefficient, the stretching color anomaly coefficient, and the motion color anomaly coefficient of each functional area to obtain the sock dyeing uniformity coefficient.

[0150] In order to verify the recognition accuracy of the dyeing uniformity recognition model, a verification experiment is carried out on it. In the verification experiment, Model One, Model Two, Model Three, and Model Four are constructed.

[0151] Among them, Model One is the dyeing uniformity recognition model, which considers the initial state, the stretching state, and the motion state, and performs recognition through the initial difference data, the stretching difference data, and the motion difference data; Model Two is based on Model One, without considering the motion state, and performs recognition through the initial difference data and the stretching difference data; Model Three is based on Model One, without considering the stretching state, and performs recognition through the initial difference data and the motion difference data; Model Four is based on Model One, without considering the stretching state and the motion state, and only performs recognition through the initial difference data.

[0152] The same verification dataset is used to recognize Model One, Model Two, Model Three, and Model Four to obtain their recognition accuracies, and the recognition accuracy is obtained based on the data difference between the sock dyeing uniformity coefficient recognized by the model and the dyeing uniformity label; the results of the verification experiment are shown in Table 2.

[0153] Table 2 Verification data table of the dyeing uniformity recognition model

[0154]

[0155] From the data in Table 2, it can be seen that Model One has the highest recognition accuracy.

[0156] After obtaining the sock dyeing uniformity coefficient of the sock product A sample, the sock dyeing uniformity coefficient is rated according to the set threshold to obtain the dyeing uniformity level data of the same batch of products.

[0157] The present invention obtains an initial sock sample by random sampling; divides and photographs the initial sock sample to obtain an initial sample image set; constructs a color difference recognition model to recognize the initial sample image set to obtain initial difference data; conducts a stretching test on the initial sock sample to obtain a stretched sock sample and a stretched sample image set; recognizes the stretched sample image set to obtain stretched difference data; conducts a motion test on the stretched sock sample to obtain a motion sample image set; recognizes the motion sample image set to obtain motion difference data; constructs a dyeing uniformity recognition model to recognize the initial difference data, the stretched difference data, and the motion difference data to obtain the sock dyeing uniformity coefficient; and realizes an accurate rating of the sock dyeing uniformity.

[0158] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for determining uniformity of sock dyeing, characterized in that: include: S10. Collect data on sock products to obtain knitting process data and standard color distribution data; Through random sampling, an initial sample of socks is obtained; S20. Divide the initial sock sample into multiple functional areas according to the knitting process data, including a sole area, a toe area, an instep area, and a sock opening area; photograph the functional areas to obtain an initial sample image set; S30. Constructing a color difference recognition model to recognize the initial sample image set, segmenting and detecting the image of the functional area through the standard color distribution data, and obtaining the initial difference data; The color difference recognition model includes a sock data acquisition layer, a sock image segmentation layer and a sock color recognition layer; The socks data acquisition layer is used to acquire standard color distribution data and a sample image set of socks, wherein the sample image set includes a sole area image, a toe area image, an instep area image, and a sock opening area image; The sock image segmentation layer segments the functional region images in the sample image set according to the standard color distribution data, and divides the sub-regions in the region image that belong to the same color distribution into one category, thereby obtaining a region image subset; The sock color recognition layer is used to recognize the color distribution data in the regional image subset to obtain the regional subset color data; obtain the difference region according to the regional subset color data and the standard color distribution data; obtain the difference coefficient according to the data difference between the color distribution data of the difference region and the corresponding standard color distribution data; and output the difference region and the difference coefficient as the difference data; S40. Determine stretch data according to the size data of the initial sock sample, perform a stretch test on the initial sock sample according to the stretch data, and obtain a stretched sock sample; photograph the functional area of ​​the stretched sock sample to obtain a stretched sample image set; S50. Recognize the stretched sample image set through the color difference recognition model to obtain the stretched difference data; S60. Screening users according to the size data of the initial sock sample to determine the first user; performing a sports test on the first user and the stretch sock sample to obtain a sports sock sample; photographing the functional area of ​​the sports sock sample to obtain a sports sample image set; S70. Recognize the motion sample image set through the color difference recognition model to obtain motion difference data; S80. Constructing a dyeing uniformity recognition model, identifying the initial difference data, stretching difference data and motion difference data of the same functional area, and obtaining a color anomaly coefficient; Then the uniformity coefficient of sock dyeing is obtained by identifying the abnormal coefficient of color distribution, and the grade is evaluated; The dyeing uniformity recognition model includes a first difference recognition layer, a second difference recognition layer, a third difference recognition layer and a dyeing uniformity recognition layer; The first difference recognition layer is used to recognize the initial difference data of the functional area to obtain the initial difference area and the initial difference coefficient of the functional area; the initial difference coefficient is weighted according to the ratio of the area of ​​the initial difference area to the area of ​​the corresponding functional area to obtain the initial color abnormality coefficient of the functional area; The second difference recognition layer is used to recognize the stretch difference data of the functional area to obtain the stretch difference area and the stretch difference coefficient of the functional area; according to the ratio of the area of ​​the stretch difference area to the area of ​​the corresponding functional area, the stretch difference coefficient is weighted to obtain the stretch color abnormality coefficient of the functional area; The third difference recognition layer is used to recognize the motion difference data of the functional area to obtain the motion difference area and the motion difference coefficient of the functional area; according to the ratio of the area of ​​the motion difference area to the area of ​​the corresponding functional area, the motion difference coefficient is weighted to obtain the motion color abnormality coefficient of the functional area; The dyeing uniformity identification layer performs weighted identification according to the initial color abnormality coefficient, the stretching color abnormality coefficient and the motion color abnormality coefficient of each functional area to obtain the socks dyeing uniformity coefficient.

2. A method for determining uniformity of sock dyeing according to claim 1, characterized in that: The knitting process data includes three-dimensional shape data and textile structure data of the socks; the initial sock sample is divided into a sole area, a toe area, an instep area and a sock opening area according to the knitting process data; The initial sample image set includes a first sole area image, a first toe area image, a first instep area image and a first sock opening area image; the stretching sample image set includes a second sole area image, a second toe area image, a second instep area image and a second sock opening area image; the motion sample image set includes a third sole area image, a third toe area image, a third instep area image and a third sock opening area image.

3. A method for determining uniformity of sock dyeing according to claim 1, characterized in that: The process of the tensile test is: Obtaining size data of the initial sock sample; Determining the user's foot shape data according to the size data, and using the foot shape data as stretching data; Determine foot mold parameters according to the stretching data, and obtain an experimental foot mold according to the foot mold parameters; wear the initial sock sample on the experimental foot mold to obtain a stretched sock sample; The stretched socks sample is photographed by a photographing device to obtain a stretched sample image set.

4. A method for determining uniformity of sock dyeing according to claim 1, characterized in that: The process of the exercise test is as follows: Determine the user's foot shape data according to the size data; identify the user according to the foot shape data, and take the user who meets the foot shape data as the first user; The process of the first user wearing the stretch socks sample to exercise is identified, and the socks sample in exercise is used as a sports socks sample; the exercise type includes walking, running, jumping and spinning; Photographing the functional areas of the sports socks samples to obtain a sports sample image set; The motion sample image set includes a walking image subset, a running image subset, a jumping image subset and a rotating image subset; based on the identification of each image subset, walking difference data, running difference data, jumping difference data and rotation difference data are obtained; and then based on the comprehensive identification of the walking difference data, running difference data, jumping difference data and rotation difference data, motion difference data is obtained.

5. A socks dyeing uniformity determination system, characterized in that: include: The sample sampling module collects sock product data and obtains knitting process data and standard color distribution data; Through random sampling, an initial sample of socks is obtained; The initial image acquisition module divides the initial sock sample into multiple functional areas according to the knitting process data, including the sole area, the toe area, the instep area and the sock opening area; and photographs the functional areas to obtain an initial sample image set; The initial image recognition module builds a color difference recognition model to recognize the initial sample image set, and segments and detects the image of the functional area through standard color distribution data to obtain initial difference data; The color difference recognition model includes a sock data acquisition layer, a sock image segmentation layer and a sock color recognition layer; The socks data acquisition layer is used to acquire standard color distribution data and a sample image set of socks, wherein the sample image set includes a sole area image, a toe area image, an instep area image, and a sock opening area image; The sock image segmentation layer segments the functional region images in the sample image set according to the standard color distribution data, and divides the sub-regions in the region image that belong to the same color distribution into one category, thereby obtaining a region image subset; The sock color recognition layer is used to recognize the color distribution data in the regional image subset to obtain the regional subset color data; obtain the difference region according to the regional subset color data and the standard color distribution data; obtain the difference coefficient according to the data difference between the color distribution data of the difference region and the corresponding standard color distribution data; and output the difference region and the difference coefficient as the difference data; A stretching image acquisition module determines stretching data according to the size data of the initial sock sample, performs a stretching test on the initial sock sample according to the stretching data, and obtains a stretching sock sample; photographs the functional area of ​​the stretching sock sample to obtain a stretching sample image set; A stretched image recognition module, which recognizes the stretched sample image set through a color difference recognition model to obtain stretched difference data; The motion image acquisition module screens users according to the size data of the initial sock sample to determine the first user; performs a motion test on the first user and the stretched sock sample to obtain a motion sock sample; and photographs the functional area of ​​the motion sock sample to obtain a motion sample image set; The motion image recognition module uses a color difference recognition model to identify the motion sample image set to obtain motion difference data; The dyeing grade assessment module builds a dyeing uniformity recognition model to identify the initial difference data, stretching difference data and movement difference data of the same functional area to obtain the color abnormality coefficient; then the color distribution abnormality coefficient is used to obtain the socks dyeing uniformity coefficient for grade assessment; The dyeing uniformity recognition model includes a first difference recognition layer, a second difference recognition layer, a third difference recognition layer and a dyeing uniformity recognition layer; The first difference recognition layer is used to recognize the initial difference data of the functional area to obtain the initial difference area and the initial difference coefficient of the functional area; the initial difference coefficient is weighted according to the ratio of the area of ​​the initial difference area to the area of ​​the corresponding functional area to obtain the initial color abnormality coefficient of the functional area; The second difference recognition layer is used to recognize the stretch difference data of the functional area to obtain the stretch difference area and the stretch difference coefficient of the functional area; according to the ratio of the area of ​​the stretch difference area to the area of ​​the corresponding functional area, the stretch difference coefficient is weighted to obtain the stretch color abnormality coefficient of the functional area; The third difference recognition layer is used to recognize the motion difference data of the functional area to obtain the motion difference area and the motion difference coefficient of the functional area; according to the ratio of the area of ​​the motion difference area to the area of ​​the corresponding functional area, the motion difference coefficient is weighted to obtain the motion color abnormality coefficient of the functional area; The dyeing uniformity identification layer performs weighted identification according to the initial color abnormality coefficient, the stretching color abnormality coefficient and the motion color abnormality coefficient of each functional area to obtain the socks dyeing uniformity coefficient.

6. A socks dyeing uniformity determination system according to claim 5, characterized in that: The knitting process data includes three-dimensional shape data and textile structure data of the socks; the initial sock sample is divided into a sole area, a toe area, an instep area and a sock opening area according to the knitting process data; The initial sample image set includes a first sole area image, a first toe area image, a first instep area image and a first sock opening area image; the stretching sample image set includes a second sole area image, a second toe area image, a second instep area image and a second sock opening area image; the motion sample image set includes a third sole area image, a third toe area image, a third instep area image and a third sock opening area image.

7. A storage medium, characterized in that: The storage medium stores a computer program, which, when executed by a processor, implements a method for determining the uniformity of sock dyeing as claimed in any one of claims 1 to 4.

Images