Surface defect quality inspection platform and method based on image recognition algorithm

Abstract

The present application discloses a surface defect quality inspection platform and method based on image recognition algorithm, the platform includes: a background center, a cloud center, an Internet terminal and a production workshop device; wherein the production workshop device is used to collect images of objects to be detected, input the images of the objects to be detected into a preset defect detection model, obtain the detected defect type information, defect location information and defect size information, and send the detected object image information, defect type information, defect location information and defect size information to the cloud center; the cloud center is used to send the received data to the background center; the background center is used to count the received data and generate an object defect detection report; the Internet terminal is used to query the object defect detection report. According to the quality inspection platform, the defect type, defect location and defect size information in the production process can be automatically and accurately identified, saving a lot of manual management costs for enterprises.

Description

Technical Field

[0001] The present invention relates to the technical field of factory quality inspection, and in particular to a surface defect quality inspection platform and method based on an image recognition algorithm. Background Art

[0002] In the production process of cloth, furniture, etc., various defects will occur due to various influences. In order to ensure production quality, defect identification has become an important part of production and quality management in the textile industry and other fields.

[0003] At present, this kind of high-intensity and high-repetitive work can basically only be completed by manual inspection. Due to the strictness of quality inspection requirements, the progress of manual quality inspection is generally slow, and the inspection process is prone to missed inspections or wrong inspections. Therefore, it is difficult for the pass rate of manual inspection to reach 100%. It can be seen that supporting quality inspection work requires a lot of manpower and management costs, as well as a lot of time costs. The final inspection results are difficult to guarantee "zero defects", which brings greater pressure to enterprises. Summary of the invention

[0004] The embodiments of the present application provide a surface defect quality inspection platform and method based on an image recognition algorithm. In order to have a basic understanding of some aspects of the disclosed embodiments, a simple summary is given below. This summary is not a general review, nor is it intended to identify key / important components or describe the scope of protection of these embodiments. Its only purpose is to present some concepts in a simple form as a preface to the detailed description that follows.

[0005] In a first aspect, an embodiment of the present application provides a surface defect quality inspection platform based on an image recognition algorithm, comprising:

[0006] Backstage center, cloud center, Internet terminal and production workshop equipment;

[0007] Among them, the production workshop equipment is used to collect images of objects to be detected, input the images of the objects to be detected into a preset defect detection model, obtain the detected defect type information, defect location information and defect size information, and send the detected object image information, defect type information, defect location information and defect size information to the cloud center;

[0008] The cloud center is used to send the received object image information, defect type information, defect location information and defect size information to the backend center;

[0009] The backend center is used to collect statistics on object image information, defect type information, defect location information, and defect size information to generate an object defect detection report;

[0010] The Internet terminal is used to query the object defect detection report.

[0011] In one embodiment, the production workshop equipment includes:

[0012] A camera, used to capture images of objects to be detected;

[0013] A quality inspection processor, used to input the image of the object to be inspected into a preset defect detection model to obtain information on the type, location and size of the detected defects;

[0014] An integrated switch, used for interconnecting with the cloud center, and sending the object image, defect type information, defect location information, and defect size information to the cloud center;

[0015] Fill light, used to provide light source for the camera;

[0016] The sound and light alarm is used to send out sound and light alarm information after detecting defect information.

[0017] In one embodiment, the quality inspection processor is specifically used to perform format conversion on the image of the object to be inspected to obtain a converted object image;

[0018] The converted object image is evenly divided to obtain multiple image blocks with equal length and width;

[0019] Calculate the characteristic values ​​of the pixels of each image block, and determine different types of defect points according to the calculated characteristic values;

[0020] Different types of defect points are classified and labeled, and the position information and size information of each defect point are determined according to the coordinate information and size information of the labeling box.

[0021] In one embodiment, the production workshop equipment further includes:

[0022] The quality inspection duty management platform is connected to the cloud center and is used to view and count the data stored in the cloud center; it is also used to supervise the equipment in the production workshop.

[0023] In one embodiment, the production workshop equipment further includes:

[0024] The quality inspection operation platform is used to display the object image and defect image in real time; it is also used to remotely set the parameters of the camera to collect images; it is also used to control the rotation of the motor; it is also used to control the fill light and the sound and light alarm.

[0025] In one embodiment, the backend center includes:

[0026] The intelligent server group is used to score each defect point according to the preset scoring rules and defect type information and defect size information to obtain the defect score;

[0027] Used to count the score information, type information, location information, quantity information and size information of defects in objects, and determine the quality level of objects;

[0028] Used to generate object detection reports based on the object's image information, production batch information, and quality grade information.

[0029] In one embodiment, the backend center further includes:

[0030] A storage server group, used to store defect type information, defect size information, defect location information, defect quantity information, defect score information, object image information, defect image information, object quality grade information, production batch information, and storage record information;

[0031] It is also used to perform data fusion based on the stored data and generate structured data reports.

[0032] In one embodiment, an Internet terminal comprises:

[0033] Quality inspection and supervision mobile phone terminals and quality inspection and supervision computer terminals are used to query object defect detection reports and data reports, are also used for emergency warnings, and are also used to supervise production processes.

[0034] In a second aspect, an embodiment of the present application provides a surface defect quality inspection method based on an image recognition algorithm, the method comprising:

[0035] Acquire an image of an object to be detected;

[0036] Input the image of the object to be detected into a preset defect detection model to obtain the detected defect type information, defect location information and defect size information;

[0037] The defect type information, defect location information and defect size information are counted to generate a defect detection report.

[0038] In one embodiment, the image of the object to be detected is input into a preset defect detection model to obtain the detected defect type information, defect location information and defect size information, including:

[0039] The defect detection model performs format conversion on the image of the object to be detected to obtain a converted object image;

[0040] The converted object image is evenly divided to obtain multiple image blocks with equal length and width;

[0041] Calculate the characteristic values ​​of the pixels of each image block, and determine different types of defect points according to the calculated characteristic values;

[0042] Different types of defect points are classified and labeled, and the position information and size information of each defect point are determined according to the coordinate information and size information of the labeling box.

[0043] The technical solution provided by the embodiments of the present application may have the following beneficial effects:

[0044] According to the defect detection method provided in the embodiment of the present application, the surface defect type, defect location, defect size and other data of objects such as cloth and furniture can be automatically identified and processed, and the data in the quality inspection process can be automatically classified and counted to generate an inspection report. The method has high accuracy, real-time performance and stability, greatly improves production efficiency and the degree of automation of production, effectively reduces the labor cost of production, and improves production quality.

[0045] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0047] Figure 1 is a flow chart of a surface defect quality inspection method based on an image recognition algorithm according to an exemplary embodiment;

[0048] Figure 2 is a schematic diagram of a surface defect quality inspection method based on an image recognition algorithm according to an exemplary embodiment;

[0049] Figure 3 is a schematic diagram showing a data storage method according to an exemplary embodiment;

[0050] Figure 4 is a structural schematic diagram of a surface defect quality inspection platform based on an image recognition algorithm according to an exemplary embodiment;

[0051] Figure 5 is a structural schematic diagram of a surface defect quality inspection platform based on an image recognition algorithm according to an exemplary embodiment;

[0052] Figure 6 is a schematic diagram of a fabric quality inspection platform according to an exemplary embodiment;

[0053] Figure 7 is a schematic diagram of a fabric quality inspection platform according to an exemplary embodiment;

[0054] Figure 8It is a schematic diagram of a finished product quality inspection platform according to an exemplary embodiment. DETAILED DESCRIPTION

[0055] The following description and the drawings sufficiently illustrate specific embodiments of the invention to enable those skilled in the art to practice them.

[0056] It should be clear that the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] In an exemplary scenario, a visual recognition system is used to inspect the quality of cloth. The main difficulties are as follows: (1) Accuracy. There are many types of cloth defects and impurities, with different shapes and sizes, which are difficult to determine in advance. However, it is necessary to be able to accurately detect and mark all defects. (2) Real-time performance. When the machine is running at high speed, the visual recognition algorithm must be able to process fast enough. (3) Statistical analysis. Mark the defect location of the product and perform statistical analysis on the defect location, quantity and other information of the product to provide a more efficient method for production management. (4) Stability. It is required to be able to work continuously. It must also be able to work normally or ensure that data is not lost in more extreme scenarios. (5) Convenient operation. The functional operation interface is simple and easy to use.

[0058] The present application solves the above-mentioned difficult problems accordingly. Through the original defect recognition processing algorithm, it does not rely on any third-party algorithm library, and uses C / C++ to build the algorithm from the bottom up. Most third-party image algorithm libraries are not open source, and the code encapsulation level is deep, resulting in low code execution processing efficiency. It is not practical in application scenarios with very strict real-time requirements, and the development language of some algorithm libraries is not completely unified, and the compatibility of software or systems is poor. Since the visual recognition processing algorithm of this application is constructed from the bottom up in C / C++ language, the code encapsulation level is low. It has the characteristics of stronger compatibility, lower redundancy, and faster processing speed, and can meet the real-time requirements of quality inspection.

[0059] The defect recognition algorithm of the present application can extract various features from cloth images, screen defects, impurities and other features, perform intelligent statistical analysis, achieve high-precision defect recognition and marking, and learn and record defect types that are not recorded in the algorithm library, gradually improving the accuracy of detection. In the process of defect detection, the system background performs intelligent statistical analysis on the process and result data of identification and detection, integrates and processes big data, and provides data reports for various types of production management. The operation terminal interface is customized according to functional requirements, with a simple interface and convenient operation. It basically only includes the operations of setting cloth information, sampling, starting, and stopping. It is simple and easy to understand, and the operation is stable.

[0060] The surface defect quality inspection method based on the image recognition algorithm provided by the embodiment of the present application will be described in detail below with reference to the accompanying drawings. Figure 1 As shown, the method comprises the following steps:

[0061] S101 obtains an image of an object to be detected.

[0062] In one possible implementation, defects in cloth production can be detected to obtain cloth images. Defects in aluminum and furniture building materials production can also be detected to obtain aluminum images, furniture images, etc.

[0063] The present application embodiment is illustrated with a cloth image, for example, an industrial high-speed camera is used to capture a stable and clear image of a cloth moving rapidly on a test table. The industrial high-speed camera used in the present application embodiment can have a maximum frame rate of 90 frames, which can fully capture the cloth image accurately, in real time and stably.

[0064] S102: Inputting the image of the object to be detected into a preset defect detection model to obtain the detected defect type information, defect location information and defect size information.

[0065] In a possible implementation, a cloth image to be detected is input into a preset defect detection model, and the defect detection model performs format conversion on the cloth image to be detected to obtain a converted cloth image; the converted cloth image is evenly divided to obtain a plurality of image blocks with equal length and width; the characteristic values ​​of the pixel points of each image block are calculated, and different types of defect points are determined according to the calculated characteristic values; different types of defect points are classified and labeled, and the position information and size information of each defect point are determined according to the coordinate information and size information of the labeled box.

[0066] Among them, calculating the characteristic values ​​of the pixels of each image block, and determining different types of defect points according to the calculated characteristic values, includes: calculating the characteristic values ​​of the target type of all pixels in each image block, including calculating the red characteristic value, green characteristic value, blue characteristic value, yellow characteristic value, cyan characteristic value, purple characteristic value, white characteristic value, black characteristic value, maximum characteristic value, minimum characteristic value, and average characteristic value of all pixels in each image block. Different types of defects can be determined according to the calculation of different types of characteristic values.

[0067] Furthermore, a detection value of the characteristic value of the target type is calculated based on the characteristic value of the target type; a statistical mean value of the characteristic value of the target type is calculated based on the detection value of the characteristic value of the target type; suspected defect points of the target type are determined based on the statistical mean value of the characteristic value of the target type; when the number of suspected defect points of the target type in the target area of ​​the image block is greater than a preset threshold, the target area is determined to be a defect point of the target type.

[0068] In an exemplary embodiment, the specific steps of the defect detection algorithm are as follows:

[0069] (1) Image format conversion;

[0070] By using the QImage method of the development tool Qt software, the JPG image is converted into RGB image data and saved in container A, where the image width is nW and the image height is nH. Each image pixel is 3 bytes, so the value range of container A is 3*nW*nH.

[0071] (2) Feature value extraction;

[0072] Step 1: Divide the RGB image frame data evenly, divide the image width into nmw=256 parts evenly, and divide the image height into nmh=256 parts evenly, thereby obtaining nmh*nmw image blocks after division and saving them in container B.

[0073] The width of an image block is: w = nW / nmh.

[0074] The height of an image block is: h = nH / nmw.

[0075] Step 2: Extract a single image block from container B and save it to container C. All the pixels of a single image block are equivalent to a table with h rows and w columns. The starting pixel is (a1, b1), where a equals w and b equals h, as follows:

[0076] C 11 ,C 12 ,C 13 ,...,C 1w

[0077] C 21 ,C 22 ,C 23 ,…,C 2w

[0078] C 31 ,C 32 ,C 33 ,…,C 3w

[0079] …

[0080] C h1 ,C h2 ,C h3 ,…,C hw

[0081] Step 3: Extract pixels from container C for feature extraction. The maximum, minimum, and average values ​​of the eight colors and pixels of red, green, blue, yellow, cyan, purple, white, and black are used to extract 11 feature values, as follows:

[0082] a) Red eigenvalue: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula:

[0083] R nH*nW (X)>G nH*nW (X) and R nH*nW (X)>B nH*nW (X)

[0084] Then calculate the red feature value of the pixel point, and the calculation formula is as follows:

[0085] BTR nH*nW (X)=((R nH*nW (X)-G nH*nW (X))+(R nH*nW (X)-G nH*nW (X)))

[0086] Otherwise BTR nH*nW (X) = 0,

[0087] The red feature values ​​of all pixels in the image block are counted. The statistical formula is as follows:

[0088]

[0089] b) Green feature value: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula:

[0090] G nH*nW (X)>R nH*nW (X) and G nH*nW (X)>B nH*nW (X)

[0091] Then calculate the green feature value of the pixel point, and the calculation formula is as follows:

[0092] BTG nH*nW (X)=((G nH*nW (X)-R nH*nW (X))+(G nH*nW (X)-B nH*nW (X)))

[0093] Otherwise BTG nH*nW (X) = 0,

[0094] Count the green feature values ​​of all pixels in the image block. The statistical formula is as follows:

[0095]

[0096] c) Blue eigenvalue: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula:

[0097] B nH*nW (X)>R nH*nW (X) and B nH*nW (X)>G nH*nW (X)

[0098] Then calculate the blue feature value of the pixel point, and the calculation formula is as follows:

[0099] BTB nH*nW (X)=((B nH*nW (X)-R nH*nW (X))+(B nH*nW (X)-G nH*nW (X)))

[0100] Otherwise BTB nH*nW (X) = 0,

[0101] Count the blue feature values ​​of all pixels in the image block. The statistical formula is as follows:

[0102]

[0103] d) Yellow eigenvalue: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula:

[0104] R nH*nW (X)>B nH*nW (X) and G nH*nW (X)>B nH*nW (X)

[0105] Then calculate the yellow feature value of the pixel point, and the calculation formula is as follows:

[0106] BTH nH*nW (X)=((R nH*nW (X)-B nH*nW (X))+(G nH*nW (X)-B nH*nW (X)))

[0107] Otherwise BTH nH*nW (X) = 0,

[0108] The yellow feature values ​​of all pixels in the image block are counted. The statistical formula is as follows:

[0109]

[0110] e) Cyan feature value: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula:

[0111] R nH*nW (X)>G nH*nW (X) and B nH*nW (X)>G nH*nW (X)

[0112] Then calculate the cyan feature value of the pixel point, and the calculation formula is as follows:

[0113] BTT nH*nW (X)=((R nH*nW (X)-G nH*nW (X))+(B nH*nW (X)-G nH*nW (X)))

[0114] Otherwise BTT nH*nW (X) = 0,

[0115] Count the cyan feature values ​​of all pixels in the image block. The statistical formula is as follows:

[0116]

[0117] f) Purple eigenvalue: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula:

[0118] G nH*nW (X)>R nH*nW (X) and B nH*nW (X)>R nH*nW (X)

[0119] Then calculate the purple eigenvalue of the pixel point, and the calculation formula is as follows:

[0120] BTZ nH*nW (X)=((G nH * nW(X)-R nH*nW (X))+(B nH*nW (X)-R nH*nW (X)))

[0121] Otherwise BTZ nH*nW (X) = 0,

[0122] Count the purple eigenvalues ​​of all pixels in the image block. The statistical formula is as follows:

[0123]

[0124] g) White feature value: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula:

[0125] G nH*nW (X) = R nH*nW (X) and B nH*nW (X) = R nH*nW (X)

[0126] The white feature values ​​of all pixels in the image block are counted. The statistical formula is as follows:

[0127]

[0128] h) Black feature value: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula:

[0129] G nH*nW (X) = R nH*nW (X) and B nH*nW (X) = R nH*nW (X)

[0130] The black feature values ​​of all pixels in the image block are counted. The statistical formula is as follows:

[0131]

[0132] i) Minimum eigenvalue: Count the eigenvalues ​​of all pixels in the image block. The statistical formula is as follows:

[0133]

[0134] j) Maximum eigenvalue: Count the eigenvalues ​​of all pixels in the image block. The statistical formula is as follows:

[0135] AT1 nmh*nmw (a1, b1)

[0136] k) Average eigenvalue: Calculate the average of the first 10 eigenvalues ​​in the image block. The statistical formula is as follows:

[0137]

[0138] (3) The difference values ​​construct 11 new eigenvalues;

[0139] According to the eigenvalues ​​calculated in step (2), 11 new eigenvalues ​​are reconstructed, and the eigenvalues ​​of red, green, blue, yellow, cyan, purple, white, black, minimum, maximum, and average are reconstructed as follows:

[0140] a) Red difference eigenvalue: Reconstruct the red eigenvalue, and the calculation formula is as follows:

[0141]

[0142] b) Green difference eigenvalue: Reconstruct the green eigenvalue, and the calculation formula is as follows:

[0143]

[0144] c) Blue difference eigenvalue: Reconstruct the blue eigenvalue, and the calculation formula is as follows:

[0145]

[0146] d) Yellow difference eigenvalue: Reconstruct the yellow eigenvalue, and the calculation formula is as follows:

[0147]

[0148] e) Cyan difference feature value: Reconstruct the cyan feature value, and the calculation formula is as follows:

[0149]

[0150] f) Purple difference eigenvalue: Reconstruct the purple eigenvalue, and the calculation formula is as follows:

[0151]

[0152] g) White difference eigenvalue: Reconstruct the white eigenvalue, and the calculation formula is as follows:

[0153]

[0154] h) Black difference eigenvalue: Reconstruct the black eigenvalue, and the calculation formula is as follows:

[0155]

[0156] i) Minimum difference eigenvalue: Reconstruct the minimum eigenvalue, and the calculation formula is as follows:

[0157]

[0158] j) Maximum difference eigenvalue: Reconstruct the maximum eigenvalue, and the calculation formula is as follows:

[0159]

[0160] k) Average difference eigenvalue: Calculate the average value of the first 10 difference eigenvalues ​​in the image block. The statistical formula is as follows:

[0161]

[0162] (4) Find the left and right edge values;

[0163] A: If the background is white, find the edge value of each row based on the maximum difference feature value;

[0164] B: If the background is black, find the edge value of each row based on the minimum difference feature value;

[0165] The formula for finding the edge value is as follows:

[0166] a) Find the eigenvalue PB1 nmh*nmw (i, j) average value PBT = ATA, let the edge threshold be: a = PBT / 2

[0167] b) Find the left edge, if there are 5 BT1 in a row nmh*nmw (i,j)>a,BL nmh (i) = j

[0168] c) Find the right edge. There are 5 consecutive BT1s starting from the right. nmh*nmw (i,j)>a,BR nmh (i)>nmw-j

[0169] d) Clear all features outside the left and right edges:

[0170]

[0171]

[0172] (5) Find the maximum value of each eigenvalue and save each maximum value during sampling;

[0173] a) The maximum value of the red eigenvalue is calculated as follows:

[0174] Assume the maximum value MaxBTR=0,

[0175] Find the maximum value

[0176] Save the maximum value of this feature during sampling: MaxBTR0 = MaxBTR

[0177] b) The maximum value of the green eigenvalue is calculated as follows:

[0178] Assume the maximum value MaxBTG=0,

[0179] Find the maximum value

[0180] Save the maximum value of this feature during sampling: MaxBTG0 = MaxBTG

[0181] c) The maximum value of the blue eigenvalue is calculated as follows:

[0182] Set the maximum value MaxBTB=0,

[0183] Find the maximum value

[0184] The maximum value of this feature is saved during sampling: MaxBTB0 = MaxBTB

[0185] d) Maximum value of yellow eigenvalue, calculated by the following formula:

[0186] Assume the maximum value MaxBTH=0,

[0187] Find the maximum value

[0188] Save the maximum value of this feature during sampling: MaxBTH0 = MaxBTH

[0189] e) The maximum cyan feature value is calculated as follows:

[0190] Assume the maximum value MaxBTT=0,

[0191] Find the maximum value

[0192] Save the maximum value of this feature during sampling: MaxBTT0 = MaxBTT

[0193] f) The maximum value of the purple eigenvalue is calculated as follows:

[0194] Assume the maximum value MaxBTZ = 0,

[0195] Find the maximum value

[0196] Save the maximum value of this feature during sampling: MaxBTZ0 = MaxBTZ

[0197] g) The maximum value of white eigenvalue is calculated as follows:

[0198] Assume the maximum value MaxBT1R = 0,

[0199] Find the maximum value

[0200] The maximum value of this feature is saved during sampling: MaxBT1R0 = MaxBT1R

[0201] h) Maximum black eigenvalue, calculated as follows:

[0202] Assume the maximum value MaxBT1G = 0,

[0203] Find the maximum value

[0204] The maximum value of this feature is saved during sampling: MaxBT1G0 = MaxBT1G

[0205] i) Minimum eigenvalue and maximum eigenvalue, the calculation formula is as follows:

[0206] Assume the maximum value MaxBT1B = 0,

[0207] Find the maximum value

[0208] The maximum value of this feature is saved during sampling: MaxBT1B0 = MaxBT1B

[0209] j) Maximum value of the eigenvalue, the calculation formula is as follows:

[0210] Assume the maximum value MaxBT1 = 0,

[0211] Find the maximum value

[0212] The maximum value of this feature is saved during sampling: MaxBT10 = MaxBT1

[0213] k) The maximum value of the average eigenvalue is calculated as follows:

[0214] Assume the maximum value MaxBTA=0,

[0215] Find the maximum value

[0216] The maximum value of this feature is saved during sampling: MaxBTA0 = MaxBTA

[0217] l) The maximum value of the red difference eigenvalue is calculated as follows:

[0218] Assume the maximum value MaxBTRD=0,

[0219] Find the maximum value

[0220]

[0221] Save the maximum value of this feature during sampling: MaxBTRD0 = MaxBTRD

[0222] m) The maximum value of the green difference eigenvalue is calculated as follows:

[0223] Assume the maximum value MaxBTGD = 0,

[0224] Find the maximum value

[0225]

[0226] The maximum value of this feature is saved during sampling: MaxBTGD0 = MaxBTGD

[0227] n) The maximum value of the blue difference eigenvalue is calculated as follows:

[0228] Assume the maximum value MaxBTBD=0,

[0229] Find the maximum value

[0230]

[0231] The maximum value of this feature is saved during sampling: MaxBTBD0 = MaxBTBD

[0232] o) Maximum value of yellow difference characteristic value, calculated by the following formula:

[0233] Assume the maximum value MaxBTHD=0,

[0234] Find the maximum value

[0235]

[0236] The maximum value of this feature is saved during sampling: MaxBTHD0 = MaxBTHD

[0237] p) The maximum value of the cyan difference characteristic value is calculated as follows:

[0238] Assume the maximum value MaxBTTD=0,

[0239] Find the maximum value

[0240]

[0241] The maximum value of this feature is saved during sampling: MaxBTTD0 = MaxBTTD

[0242] q) The maximum value of the purple difference eigenvalue is calculated as follows:

[0243] Assume the maximum value MaxBTZD=0,

[0244] Find the maximum value

[0245]

[0246] The maximum value of this feature is saved during sampling: MaxBTZD0 = MaxBTZD

[0247] r) The maximum value of the white difference characteristic value is calculated as follows:

[0248] Assume the maximum value MaxBT1RD=0,

[0249] Find the maximum value

[0250]

[0251] The maximum value of this feature is saved during sampling: MaxBT1RD0 = MaxBT1RD

[0252] s) The maximum value of the black difference characteristic value is calculated as follows:

[0253] Assume the maximum value MaxBT1GD = 0,

[0254] Find the maximum value

[0255]

[0256] The maximum value of this feature is saved during sampling: MaxBT1GD0 = MaxBT1GD

[0257] t) The minimum difference eigenvalue maximum value is calculated as follows:

[0258] Assume the maximum value MaxBT1BD=0,

[0259] Find the maximum value

[0260]

[0261] The maximum value of this feature is saved during sampling: MaxBT1BD0 = MaxBT1BD

[0262] u) Maximum value of the maximum difference eigenvalue, calculated as follows:

[0263] Assume the maximum value MaxBT1D = 0,

[0264] Find the maximum value

[0265]

[0266] The maximum value of this feature is saved during sampling: MaxBT1D0 = MaxBT1D

[0267] v) Maximum value of the mean difference eigenvalue, calculated as follows:

[0268] Assume the maximum value MaxBTAD=0,

[0269] Find the maximum value

[0270]

[0271] The maximum value of this feature is saved during sampling: MaxBTAD0 = MaxBTAD

[0272] (6) During quality inspection, the detection values ​​of 11 feature values ​​in the n*m ​​image block are obtained according to the following method;

[0273] a) Calculate the red feature detection value DTR nmh*nmw (i,j), when RTR(i,j)>MaxBTR0,

[0274]

[0275] Otherwise, DTR nmh*nmw (i,j)=0

[0276] b) Calculate the red difference feature detection value DTRD nmh*nmw (i,j), when RTRd(i,j)>MaxBTR0,

[0277]

[0278] Otherwise, DTRD nmh*nmw (i, j) = 0

[0279] c) Calculate the green feature detection value DTG nmh*nmw (i,j), when RTG(i,j)>MaxBTG0,

[0280]

[0281] Otherwise, BTG nmh*nmw (i,j)=0

[0282] d) Calculate the green difference feature detection value DTGD nmh*nmw (i,j), when RTGd(i,j)>MaxBTG0,

[0283]

[0284] Otherwise, DTGD nmh*nmw (i,j)=0

[0285] e) Calculate the blue feature detection value DTB nmh*nmw (i,j), when RTB(i,j)>MaxBTB0,

[0286]

[0287] Otherwise, DTB nmh*nmw (i,j)=0

[0288] f) Calculate the blue difference feature detection value DTBD nmh*nmw (i,j), when RTBd(i,j)>MaxBTB0,

[0289]

[0290] Otherwise, DTBD nmh*nmw (i,j)=0

[0291] g) Calculate the yellow feature detection value DTH nmh*nmw (,j), when RTH(i,j)>MaxBTH0,

[0292]

[0293] Otherwise, DTH nmh*nmw (i,j)=0

[0294] h) Calculate the yellow difference characteristic detection value DTHD nmh*nmw (,j), when RTHd(i,j)>MaxBTH0,

[0295]

[0296] Otherwise, DTHD nmh*nmw (i,j)=0

[0297] i) Calculate the cyan feature detection value DTT nmh*nmw (,j), when RTT(i,j)>MaxBTT0,

[0298]

[0299] Otherwise, DTT nmh*nmw (i,j)=0

[0300] j) Calculate the cyan difference feature detection value DTTD nmh*nmw (i,j), when RTTd(i,j)>MaxBTT0,

[0301]

[0302] Otherwise, DTTD nmh*nmw(i, j) = 0

[0303] k) Find the purple feature detection value DTZ nmh*nmw (i, j), when RTZ(i, j)>MaxBTZ0,

[0304]

[0305] Otherwise, DTZ nmh*nmw (i, j) = 0

[0306] l) Calculate the purple difference feature detection value DTZD nmh*nmw (i, j), when RTZd(i, j)>MaxBTZ0,

[0307]

[0308] Otherwise, DTZD nmh*nmw (i, j) = 0

[0309] m) Calculate the white feature detection value DT1R nmh*nmw (i, j), when RT1R(i, j)>MaxBT1R0,

[0310]

[0311] Otherwise, DT1R nmh*nmw (i, j) = 0

[0312] n) Calculate the white difference feature detection value DT1RD nmh*nmw (i, j), when RT1Rd(i, j)>MaxBT1R0,

[0313]

[0314] Otherwise, DT1RD nmh*nmw (i, j) = 0

[0315] o) Calculate the black feature detection value DT1G nmh*nmw (i, j), when RT1G(i, j)>MaxBT1G0,

[0316]

[0317] Otherwise, DT1G nmh*nmw (i, j) = 0

[0318] p) Calculate the black difference feature detection value DT1GD nmh*nmw (i, j), when RT1Gd(i, j)>MaxBT1G0,

[0319]

[0320] Otherwise, DT1GD nmh*nmw (i, j) = 0

[0321] q) Find the minimum feature detection value DT1B nmh*nmw (i, j), when RTB(i, j)>MaxBTB0,

[0322]

[0323] Otherwise, DT1B nmh*nmw (i, j) = 0

[0324] r) Find the minimum difference feature detection value DT1BD nmh*nmw (i, j), when RTBd(i, j)>MaxBTB0,

[0325]

[0326] Otherwise, DT1BD nmh*nmw (i, j) = 0

[0327] s) Find the maximum feature detection value DT1 nmh*nmw (i, j), when RT1(i, j)>MaxBT10,

[0328]

[0329] Otherwise, DT1 nmh*nmw (i, j) = 0

[0330] t) Find the maximum difference feature detection value DT1D nmh*nmw (i, j), when RT1d(i, j)>MaxBT10,

[0331]

[0332] Otherwise, DT1D nmh*nmw (i, j) = 0

[0333] u) Calculate the average characteristic detection value DTA nmh*nmw (i, j), when RTA(i, j)>MaxBTA0,

[0334]

[0335] Otherwise, DTA nmh*nmw (i, j) = 0

[0336] v) Calculate the average difference feature detection value DTAD nmh*nmw (i, j), when RTAd(i, j)>MaxBTA0,

[0337]

[0338] Otherwise, DTAD nmh*nmw (i, j) = 0

[0339] (7) Calculate the statistical averages of 22 feature detection values ​​according to 11 partition statistical methods;

[0340] ①4*4 partition

[0341] a) Statistical average value of red feature detection value PTR nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0342]

[0343] b) Statistical average value of red difference feature detection value PTRd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0344]

[0345] c) Statistical average value of green feature detection value PTG nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0346]

[0347] d) Statistical average value of green difference feature detection value PTGd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0348]

[0349] e) Statistical average of blue feature detection values ​​PTB nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0350]

[0351] f) Statistical average value of blue difference feature detection value PTBd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0352]

[0353] g) Statistical average value of yellow characteristic detection value PTH nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0354]

[0355] h) Statistical mean value of yellow difference characteristic detection value PTHd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0356]

[0357] i) Statistical average value PTT of cyan feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0358]

[0359] j) Statistical average value PTTd of cyan difference feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0360]

[0361] k) Statistical average of purple feature detection values ​​PTZ nmh*nmw (i, j),

[0362] Threshold value det = 50, count value detn = 0,

[0363]

[0364] 1) Statistical mean value of purple difference feature detection value PTZd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0365]

[0366] m) Statistical average value of white feature detection value PT1R nmh*nmw (i, j),

[0367] Threshold value det = 50, count value detn = 0,

[0368]

[0369] n) Statistical average value of white difference characteristic detection value PT1Rd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0370]

[0371] o) Statistical average of black feature detection values ​​PT1G nmh*nmw (i, j),

[0372] Threshold value det = 50, count value detn = 0,

[0373]

[0374] p) Statistical average value of black difference characteristic detection value PT1Gd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0375]

[0376] q) Statistical average of the minimum feature detection value PT1B nmh*nmw (i, j),

[0377] Threshold value det = 50, count value detn = 0,

[0378]

[0379] r) Statistical average value of minimum difference feature detection value PT1Bd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0380]

[0381] s) Statistical mean value PT1 of the maximum feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0382]

[0383] t) Statistical mean value of the maximum difference feature detection value PT1d nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0384]

[0385] u) Statistical average value of the characteristic detection value (PTA) nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0386]

[0387] v) Statistical mean of the mean difference feature detection values ​​PTAd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0388]

[0389] ②8*2 partitions

[0390] a) Statistical average value of red feature detection value PTR nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0391]

[0392] b) Statistical average value of red difference feature detection value PTRd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0393]

[0394] c) Statistical average value of green feature detection value PTG nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0395]

[0396] d) Statistical average value of green difference feature detection value PTGd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0397]

[0398] e) Statistical average of blue feature detection values ​​PTB nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0399]

[0400] f) Statistical average value of blue difference feature detection value PTBd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0401]

[0402] g) Statistical average value of yellow characteristic detection value PTH nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0403]

[0404] h) Statistical mean value of yellow difference characteristic detection value PTHd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0405]

[0406] i) Statistical average value PTT of cyan feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0407]

[0408] j) Statistical average value PTTd of cyan difference feature detection valuenmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0409]

[0410] k) Statistical average of purple feature detection values ​​PTZ nmh*nmw (i, j), Threshold der = 50, count value detn = 0,

[0411]

[0412] 1) Statistical mean value of purple difference feature detection value PTZd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0413]

[0414] m) Statistical average value of white feature detection value PT1R nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0415]

[0416] n) Statistical average value of white difference characteristic detection value PT1Rd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0417]

[0418] o) Statistical average of black feature detection values ​​PT1G nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0419]

[0420] p) Statistical average value of black difference characteristic detection value PT1Gd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0421]

[0422] q) Statistical average of the minimum feature detection value PT1B nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0423]

[0424] r) Statistical average value of minimum difference feature detection value PT1Bd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0425]

[0426] s) Statistical mean value PT1 of the maximum feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0427]

[0428] t) Statistical mean value PTld of the maximum difference feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0429]

[0430] u) Statistical average value of the characteristic detection value (PTA) nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0431]

[0432] v) Statistical mean of the mean difference feature detection values ​​PTAd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0433]

[0434] ③2*8 partitions;

[0435] a) Statistical average value of red feature detection value PTR nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0436]

[0437] b) Statistical average value of red difference feature detection value PTRd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0438]

[0439] c) Statistical average value of green feature detection value PTG nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0440]

[0441] d) Statistical average value of green difference feature detection value PTGd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0442]

[0443] e) Statistical average of blue feature detection values ​​PTB nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0444]

[0445] f) Statistical average value of blue difference feature detection value PTBd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0446]

[0447] g) Statistical average value of yellow characteristic detection value PTH nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0448]

[0449] h) Statistical mean value of yellow difference characteristic detection value PTHd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0450]

[0451] i) Statistical average value PTT of cyan feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0452]

[0453] j) Statistical average value PTTd of cyan difference feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0454]

[0455] k) Statistical average of purple feature detection values ​​PTZ nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0456]

[0457] 1) Statistical mean value of purple difference feature detection value PTZd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0458]

[0459] m) Statistical average value of white feature detection value PT1R nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0460]

[0461] n) Statistical average value of white difference characteristic detection value PT1Rd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0462]

[0463] o) Statistical average of black feature detection values ​​PT1G nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0464]

[0465] p) Statistical average value of black difference characteristic detection value PT1Gd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0466]

[0467] q) Statistical average of the minimum feature detection value PT1B nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0468]

[0469] r) Statistical average value of minimum difference feature detection value PT1Bd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0470]

[0471] s) Statistical mean value PT1 of the maximum feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0472]

[0473] t) Statistical mean value of the maximum difference feature detection value PT1d nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0474]

[0475] u) Statistical average value of the characteristic detection value (PTA) nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0476]

[0477] v) Statistical mean of the mean difference feature detection values ​​PTAd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0478]

[0479] ④1*16 partitions;

[0480] a) Statistical average value of red feature detection value PTR nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0481]

[0482] b) Statistical average value of red difference feature detection value PTRd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0483]

[0484] c) Statistical average value of green feature detection value PTG nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0485]

[0486] d) Statistical average value of green difference feature detection value PTGd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0487]

[0488] e) Statistical average of blue feature detection values ​​PTB nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0489]

[0490] f) Statistical average value of blue difference feature detection value PTBd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0491]

[0492] g) Statistical average value of yellow characteristic detection value PTH nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0493]

[0494] h) Statistical mean value of yellow difference characteristic detection value PTHd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0495]

[0496] i) Statistical average value PTT of cyan feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0497]

[0498] j) Statistical average value PTTd of cyan difference feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0499]

[0500] k) Statistical average of purple feature detection values ​​PTZ nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0501]

[0502] 1) Statistical mean value of purple difference feature detection value PTZd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0503]

[0504] m) Statistical average value of white feature detection value PT1R nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0505]

[0506] n) Statistical average value of white difference characteristic detection value PT1Rd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0507]

[0508] o) Statistical average of black feature detection values ​​PT1G nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0509]

[0510] p) Statistical average value of black difference characteristic detection value PT1Gd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0511]

[0512] q) Statistical average of the minimum feature detection value PT1B nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0513]

[0514] r) Statistical average value of minimum difference feature detection value PT1Bd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0515]

[0516] s) Statistical mean value PT1 of the maximum feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0517]

[0518] t) Statistical mean value of the maximum difference feature detection value PT1d nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0519]

[0520] u) Statistical average value of the characteristic detection value (PTA) nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0521]

[0522] v) Statistical mean of the mean difference feature detection values ​​PTAd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0523]

[0524] ⑤16*1 partition;

[0525] a) Statistical average value of red feature detection value PTR nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0526]

[0527] b) Statistical average value of red difference feature detection value PTRd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0528]

[0529] c) Statistical average value of green feature detection value PTG nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0530]

[0531] d) Statistical average value of green difference feature detection value PTGd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0532]

[0533] e) Statistical average of blue feature detection values ​​PTB nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0534]

[0535] f) Statistical average value of blue difference feature detection value PTBd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0536]

[0537] g) Statistical average value of yellow characteristic detection value PTH nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0538]

[0539] h) Statistical mean value of yellow difference characteristic detection value PTHd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0540]

[0541] i) Statistical average value PTT of cyan feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0542]

[0543] j) Statistical average value PTTd of cyan difference feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0544]

[0545] k) Statistical average of purple feature detection values ​​PTZ nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0546]

[0547] 1) Statistical mean value of purple difference feature detection value PTZd nmh*nmw (i, j), Threshold der = 50, count value detn = 0,

[0548]

[0549] m) Statistical average value of white feature detection value PT1R nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0550]

[0551] n) Statistical average value of white difference characteristic detection value PT1Rd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0552]

[0553] o) Statistical average of black feature detection values ​​PT1G nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0554]

[0555] p) Statistical average value of black difference characteristic detection value PT1Gd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0556]

[0557] q) Statistical average of the minimum feature detection value PT1Bnmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0558]

[0559] r) Statistical average value of minimum difference feature detection value PT1Bd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0560]

[0561] s) Statistical mean value PT1 of the maximum feature detection value nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0562]

[0563] t) Statistical mean value of the maximum difference feature detection value PT1d nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0564]

[0565] u) Statistical average value of the characteristic detection value (PTA) nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0566]

[0567] v) Statistical mean of the mean difference feature detection values ​​PTAd nmh*nmw (i, j), Threshold value det = 50, count value detn = 0,

[0568]

[0569] (8) Find suspected defects based on the statistical average of each characteristic detection value;

[0570] Assume that the judgment threshold value detm=1, and judge the suspected defects according to different partition conditions as follows:

[0571] ①4*4 partitions;

[0572] a) If the statistical average of the red feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTR nmh*nmw (i, j) plus 1

[0573] b) If the statistical average of the red difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTRd nmh*nmw (i, j) plus 1

[0574] c) If the statistical average of the green feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTG nmh*nmw (i, j) plus 1

[0575] d) If the statistical average of the green difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTGd nmh*nmw (i, j) plus 1

[0576] e) If the statistical average of the blue feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTB. nmh*nmw (i, j) plus 1

[0577] f) If the statistical average of the blue difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTBd nmh*nmw (i, j) plus 1

[0578] g) If the statistical average of the yellow feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTH nmh*nmw (i, j) plus 1

[0579] h) If the statistical average of the yellow difference characteristic detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTHd nmh*nmw (i, j) plus 1

[0580] i) If the statistical mean of the cyan feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTT nmh*nmw (i, j) plus 1

[0581] j) If the statistical average of the cyan difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTTd nmh*nmw (i, j) plus 1

[0582] k) If the statistical average of the purple feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTZ nmh*nmw (i, j) plus 1

[0583] l) If the statistical average of the purple difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTZd nmh*nmw (i, j) plus 1

[0584] m) If the statistical average of the white feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1R nmh*nmw (i, j) plus 1

[0585] n) If the statistical average of the white difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Rd nmh*nmw (i, j) plus 1

[0586] o) If the statistical average of the black feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1G nmh*nmw (i, j) plus 1

[0587] p) If the statistical average of the black difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Gd nmh*nmw (i, j) plus 1

[0588] q) If the statistical mean of the minimum feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1B nmh*nmw (i, j) plus 1

[0589] r) If the statistical mean of the minimum difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Bd nmh*nmw (i, j) plus 1

[0590] s) If the statistical mean of the maximum feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1 nmh*nmw (i, j) plus 1

[0591] t) If the statistical mean of the maximum difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1d nmh*nmw (i, j) plus 1

[0592] u) If the statistical average value of the average color feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTA nmh*nmw (i, j) plus 1

[0593] v) If the statistical mean of the mean difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTAd. nmh*nmw (i, j) plus 1

[0594] ②8*2 partitions;

[0595] a) If the statistical average of the red feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTR nmh*nmw (i, j) plus 1

[0596] b) If the statistical average of the red difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTRd nmh*nmw (i, j) plus 1

[0597] c) If the statistical average of the green feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTG nmh*nmw (i, j) plus 1

[0598] d) If the statistical average of the green difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTGd nmh*nmw (i, j) plus 1

[0599] e) If the statistical average of the blue feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTB. nmh*nmw (i, j) plus 1

[0600] f) If the statistical average of the blue difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTBd nmh*nmw (i, j) plus 1

[0601] g) If the statistical average of the yellow feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTH nmh*nmw (i, j) plus 1

[0602] h) If the statistical average of the yellow difference characteristic detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTHd nmh*nmw (i, j) plus 1

[0603] i) If the statistical mean of the cyan feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTT nmh*nmw (i, j) plus 1

[0604] j) If the statistical average of the cyan difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTTd nmh*nmw (i, j) plus 1

[0605] k) If the statistical average of the purple feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTZ nmh*nmw (i, j) plus 1

[0606] l) If the statistical average of the purple difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTZd nmh*nmw (i, j) plus 1

[0607] m) If the statistical average of the white feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1R nmh*nmw (i, j) plus 1

[0608] n) If the statistical average of the white difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Rd nmh*nmw (i, j) plus 1

[0609] o) If the statistical average of the black feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1G nmh*nmw (i, j) plus 1

[0610] p) If the statistical average of the black difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Gd nmh*nmw (i, j) plus 1

[0611] q) If the statistical mean of the minimum feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1B nmh*nmw (i, j) plus 1

[0612] r) If the statistical mean of the minimum difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Bd nmh*nmw (i, j) plus 1

[0613] s) If the statistical mean of the maximum feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1 nmh*nmw (i, j) plus 1

[0614] t) If the statistical mean of the maximum difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1d nmh*nmw (i, j) plus 1

[0615] u) If the statistical average value of the average color feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTA nmh*nmw (i, j) plus 1

[0616] v) If the statistical mean of the mean difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTAd. nmh*nmw (i, j) plus 1

[0617] ③2*8 partitions;

[0618] a) If the statistical average of the red feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTR nmh*nmw (i, j) plus 1

[0619] b) If the statistical average of the red difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTRd nmh*nmw (i, j) plus 1

[0620] c) If the statistical average of the green feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTG nmh*nmw (i, j) plus 1

[0621] d) If the statistical average of the green difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTGd nmh*nmw (i, j) plus 1

[0622] e) If the statistical average of the blue feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTB. nmh*nmw (i, j) plus 1

[0623] f) If the statistical average of the blue difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTBdnmh*nmw (i, j) plus 1

[0624] g) If the statistical average of the yellow feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTH nmh*nmw (i, j) plus 1

[0625] h) If the statistical average of the yellow difference characteristic detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTHd nmh*nmw (i, j) plus 1

[0626] i) If the statistical mean of the cyan feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTT nmh*nmw (i, j) plus 1

[0627] j) If the statistical average of the cyan difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTTd nmh*nmw (i, j) plus 1

[0628] k) If the statistical average of the purple feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTZ nmh*nmw (i, j) plus 1

[0629] l) If the statistical average of the purple difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTZd nmh*nmw (i, j) plus 1

[0630] m) If the statistical average of the white feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1R nmh*nmw (i, j) plus 1

[0631] n) If the statistical average of the white difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Rd nmh*nmw (i, j) plus 1

[0632] o) If the statistical average of the black feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1G nmh*nmw (i, j) plus 1

[0633] p) If the statistical average of the black difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Gd nmh*nmw(i, j) plus 1

[0634] q) If the statistical mean of the minimum feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1B nmh*nmw (i, j) plus 1

[0635] r) If the statistical mean of the minimum difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Bd nmh*nmw (i, j) plus 1

[0636] s) If the statistical mean of the maximum feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1 nmh*nmw (i, j) plus 1

[0637] t) If the statistical mean of the maximum difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1d nmh*nmw (i, j) plus 1

[0638] u) If the statistical average value of the average color feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTA nmh*nmw (i, j) plus 1

[0639] v) If the statistical mean of the mean difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTAd. nmh*nmw (i, j) plus 1

[0640] ④1*16 partitions;

[0641] a) If the statistical average of the red feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTR nmh*nmw (i, j) plus 1

[0642] b) If the statistical average of the red difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTRd nmh*nmw (i, j) plus 1

[0643] c) If the statistical average of the green feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTG nmh*nmw (i, j) plus 1

[0644] d) If the statistical average of the green difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTGd nmh*nmw (i, j) plus 1

[0645] e) If the statistical average of the blue feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTB. nmh*nmw (i, j) plus 1

[0646] f) If the statistical average of the blue difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTBd nmh*nmw (i, j) plus 1

[0647] g) If the statistical average of the yellow feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTH nmh*nmw (i, j) plus 1

[0648] h) If the statistical average of the yellow difference characteristic detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTHd nmh*nmw (i, j) plus 1

[0649] i) If the statistical mean of the cyan feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTT nmh*nmw (i, j) plus 1

[0650] j) If the statistical average of the cyan difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTTd nmh*nmw (i, j) plus 1

[0651] k) If the statistical average of the purple feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTZ nmh*nmw (i, j) plus 1

[0652] l) If the statistical average of the purple difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTZd nmh*nmw (i, j) plus 1

[0653] m) If the statistical average of the white feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1R nmh*nmw (i, j) plus 1

[0654] n) If the statistical average of the white difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Rd nmh*nmw (i, j) plus 1

[0655] o) If the statistical average of the black feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1G nmh*nmw (i, j) plus 1

[0656] p) If the statistical average of the black difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Gd nmh*nmw (i, j) plus 1

[0657] q) If the statistical mean of the minimum feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1B nmh*nmw (i, j) plus 1

[0658] r) If the statistical mean of the minimum difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Bd nmh*nmw (i, j) plus 1

[0659] s) If the statistical mean of the maximum feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1 nmh*nmw (i, j) plus 1

[0660] t) If the statistical mean of the maximum difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1d nmh*nmw (i, j) plus 1

[0661] u) If the statistical average value of the average color feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTA nmh*nmw (i, j) plus 1

[0662] v) If the statistical mean of the mean difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTAd. nmh*nmw (i, j) plus 1

[0663] ⑤16*1 partition

[0664] a) If the statistical average of the red feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTR nmh*nmw (i, j) plus 1

[0665] b) If the statistical average of the red difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTRd nmh*nmw (i, j) plus 1

[0666] c) If the statistical average of the green feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTG nmh*nmw (i, j) plus 1

[0667] d) If the statistical average of the green difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTGd nmh*nmw (i, j) plus 1

[0668] e) If the statistical average of the blue feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTB. nmh*nmw (i, j) plus 1

[0669] f) If the statistical average of the blue difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTBd nmh*nmw (i, j) plus 1

[0670] g) If the statistical average of the yellow feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTH nmh*nmw (i, j) plus 1

[0671] h) If the statistical average of the yellow difference characteristic detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTHd nmh*nmw (i, j) plus 1

[0672] i) If the statistical mean of the cyan feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTT nmh*nmw (i, j) plus 1

[0673] j) If the statistical average of the cyan difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTTd nmh*nmw (i, j) plus 1

[0674] k) If the statistical average of the purple feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTZ nmh*nmw (i, j) plus 1

[0675] l) If the statistical average of the purple difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTZd nmh*nmw (i, j) plus 1

[0676] m) If the statistical average of the white feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1R nmh*nmw (i, j) plus 1

[0677] n) If the statistical average of the white difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Rd nmh*nmw (i, j) plus 1

[0678] o) If the statistical average of the black feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1G nmh*nmw (i, j) plus 1

[0679] p) If the statistical average of the black difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Gd nmh*nmw (i, j) plus 1

[0680] q) If the statistical mean of the minimum feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1B nmh*nmw (i, j) plus 1

[0681] r) If the statistical mean of the minimum difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1Bd nmh*nmw (i, j) plus 1

[0682] s) If the statistical mean of the maximum feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1 nmh*nmw (i, j) plus 1

[0683] t) If the statistical mean of the maximum difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TT1d nmh*nmw (i, j) plus 1

[0684] u) If the statistical average value of the average color feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTA nmh*nmw (i, j) plus 1

[0685] v) If the statistical mean of the mean difference feature detection value The block is recorded as a suspected defect, and the characteristic value of the suspected defect is TTAd. nmh*nmw (i, j) plus 1

[0686] (9) Count the number of suspected defects in each block according to the 16*16 block structure. If the number is greater than the threshold value detnm≥3, the block is considered a defect.

[0687] a) Red defect characteristics TTTBR 16*16 (i, j),

[0688] )

[0689] b) Red difference defect feature TTTBRd 16*16 (i, j),

[0690]

[0691] c) Green defect feature TTTBG 16*16 (i,j),

[0692]

[0693] d) Green difference defect characteristics TTTBGd 16*16 (i,j),

[0694]

[0695] e) Blue defect feature TTTBB 16*16 (i,j),

[0696]

[0697] f) Blue difference defect feature TTTBBd 16*16 ( i ,j),

[0698]

[0699] g) Yellow defect characteristics TTTBH 16*16 (i,j),

[0700]

[0701] h) Yellow difference defect characteristics TTTBHd 16*16 (i,j),

[0702] i) Cyan defect feature TTTBT 16*16 ( i ,j),

[0703]

[0704] j) Cyan difference defect feature TTTBTd 16*16 (i,j),

[0705]

[0706] k) Purple defect feature TTTBZ 16*16 (i,j),

[0707]

[0708] l) Purple difference defect feature TTTBZd 16*16 (i,j),

[0709]

[0710] m) White defect feature TTTB1R 16*16 (i,j),

[0711]

[0712] n) White difference defect feature TTTB1Rd 16*16 (i,j), TTTB1Rd 16*16 (i, j)

[0713] o) Black defect characteristics TTTB1G 16*16 (i,j),

[0714]

[0715] p) Black difference defect characteristics TTTB1Gd 16*16 (i, j), TTTTB1G 16*16 (i, j)

[0716] q) Minimum defect feature TTTB1B 16*16 (i,j),

[0717]

[0718] r) Minimum difference defect feature TTTB1Bd 16*16 (i, j), TTTB1Bd 16*16 (i, j)

[0719] s) Maximum defect feature TTTB1 16*16 (i,j),

[0720]

[0721] t) Maximum value difference defect characteristic TTTB1d 16*16 (i,j),

[0722]

[0723] u) Average value defect characteristics TTTBA 16*16 (i,j),

[0724]

[0725] v) Average value difference defect characteristic TTTBAd 16*16 (i,j),

[0726]

[0727] (10) Mark the defects and calculate the number of marked defects;

[0728] a) Red defects, set the threshold detnn = 1, the number of defects numberr = 0

[0729] Then draw the box in red, numberr++

[0730] b) Red difference defects, set the threshold detnn = 1, the number of defects numberrd = 0

[0731] Then draw the box in red, numberrd++

[0732] c) Green defects, set the threshold detnn = 1, the number of defects numberg = 0

[0733] Then draw the box in red, numberg++

[0734] d) Green difference defects, set the threshold detnn = 1, the number of defects numbergd = 0

[0735] Then draw the box in red, numbergd++

[0736] e) Blue defects, set the threshold detnn = 1, the number of defects numberb = 0

[0737] Then draw the box in red, numberb++

[0738] f) Blue difference defects, set the threshold detnn = 1, the number of defects numberbd = 0

[0739] Then draw the box in red, numberbd++

[0740] g) Yellow defects, set the threshold detnn = 1, the number of defects numberh = 0

[0741] Then draw the box in red, numberh++

[0742] h) Yellow difference defects, set the threshold detnn = 1, the number of defects numberhd = 0

[0743] Then draw the box in red, numberhd++

[0744] i) Cyan defects, set threshold detnn = 1, number of defects numbert = 0

[0745] Then draw the box in red, numbert++

[0746] j) Cyan difference defects, set threshold detnn = 1, number of defects numbertd = 0

[0747] Then draw the box in red, numbertd++

[0748] k) Purple defect, set the threshold detnn = 1, the number of defects numberz = 0

[0749] Then draw the box in red, numberz++

[0750] l) Purple difference defect, set threshold detnn = 1, number of defects numberzd = 0

[0751] Then draw the box in red, numberz++

[0752] m) White defects, set the threshold detnn = 1, the number of defects number1r = 0

[0753] Then draw the box in red, number1r++

[0754] n) White difference defects, set the threshold detnn = 1, the number of defects number1rd = 0

[0755] Then draw the box in red, number1rd++

[0756] o) Black defects, set the threshold value detnn = 1, the number of defects number1g = 0

[0757] Then draw the box in red, number1g++

[0758] p) Black difference defects, set the threshold detnn = 1, the number of defects number1gd = 0

[0759] Then draw the box in red, number1gd++

[0760] q) Minimum defect, set threshold detnn = 1, number of defects number1b = 0

[0761] Then draw the box in red, number1b++

[0762] r) Minimum difference defect, set threshold detnn = 1, number of defects number1bd = 0

[0763] Then draw the box in red, number1Bd++

[0764] s) Maximum defect, set threshold detnn = 1, number of defects number1 = 0

[0765] Then draw the box in red, number1++

[0766] t) Maximum value difference defect, set threshold detnn = 1, number of defects number1d = 0

[0767] Then draw the box in red, number1d++

[0768] u) Average value defect, set threshold detnn = 1, number of defects numbera = 0

[0769] Then draw the box in red, numbera++

[0770] v) Average value difference defect, set threshold detnn = 1, number of defects numberad = 0

[0771] Then draw the box in red, numberad++

[0772] Through the above algorithm, defects marked with different colors can be obtained. Different colors correspond to different defect types. For example, the defect type corresponding to red is yarn leakage, and the defect type corresponding to green is hole, etc., which can be specifically corresponded according to the specific characteristic value calculation method. The location information and size information of the defect can also be obtained according to the size and coordinates of the marked box, and the image information of the defect can also be obtained according to the marked defect.

[0773] Optionally, according to the needs of the production process, the cloth can be identified by algorithmic defects in the four stages of weaving, semi-finished products, finished products, and finished product warehousing, thereby improving the accuracy of defect detection.

[0774] Optionally, when a defect whose type cannot be determined appears during the system quality inspection process, the system automatically enlarges the defect location in the image and uploads it to the management platform. The on-duty personnel review the defect image and describe the defect type.

[0775] According to the defect detection algorithm provided in the embodiment of the present application, the obtained image data can be calculated at a sufficiently fast speed to extract the fabric feature information in the frame image. The algorithm adopted is to analyze and count millions of pixels, and then identify the fabric defects, find and locate the defects, and record the type information, size information, position information and image information of the defects, etc., thereby solving the problem of accurate and rapid identification of fabric defects in the prior art.

[0776] S103 collects statistics on the defect type information, defect location information and defect size information to generate a defect detection report.

[0777] In one embodiment, after obtaining the defect detection information, the defect information of the cloth can also be counted.

[0778] Specifically, each defect point is scored according to the preset scoring rules and defect type information and defect size information to obtain the defect score. The defect scoring standards commonly used in Western Europe and the United States are mainly "ten-point scoring method" and "four-point scoring method". Both of these refer to the maximum score for a single defect being ten or four points. Each defect can be scored according to the type of defect and the size of the defect. The defect scoring types can be divided into leaking yarn, tight yarn, rotten hole, connecting line, missing needle, cut, tooth edge, etc.

[0779] Furthermore, the score information, type information, location information, quantity information and size information of the defects in the cloth are counted, and the quality grade of the cloth is determined. By comprehensively scoring all defects in the entire roll of cloth, the quality grade of the roll of cloth is automatically analyzed and statistically analyzed according to the preset cloth quality standards. A cloth inspection report is generated based on data such as cloth image information, production batch information, defect image information and quality grade information.

[0780] In an optional embodiment, it also includes recording cloth information, collecting all the cloth surface information of each roll of cloth according to the rotation speed and a certain image size, and uploading and storing the cloth surface image, real-time statistical data of quality inspection, batch data and other information to the private cloud center storage server, so as to facilitate the later viewing and statistics of the cloth surface information of the whole roll of cloth.

[0781] In an optional embodiment, the in-and-out information of the cloth can also be recorded. Before the cloth enters the warehouse, the relevant cloth information of the storage server of the platform background center is automatically retrieved for the last defect analysis to improve the accuracy of defect detection. The cloth information of each roll entering and leaving the warehouse can also be provided. The information mainly includes the picture information of the whole roll of cloth, defect picture information, defect type and size statistics, defect scoring information, cloth scoring information, cloth inspection report and other information. The above information can be viewed according to the information of different materials such as semi-finished products and finished products entering the warehouse.

[0782] In an optional embodiment, customers can also trace product quality. Customers can query the product batch, product name, product code, production plant, production batch, production date, customs declaration, departure information and graphic information records of product entry through product traceability code or QR code. Product quality traceability reflects the degree of informatization of enterprise management level. In addition to archiving, tracking, feedback, query and management of the entire life cycle of the product, product quality traceability also greatly facilitates product quality tracking and after-sales service. Once a quality dispute or problem occurs, effective evidence can be provided or the cause of the quality problem can be quickly found, which is helpful to improve the effect of international brands or solve quality problems.

[0783] In order to facilitate understanding of the real-time detection method for surface defects provided in the embodiment of the present application, the following is a Figure 2 For explanation. Figure 2 As shown, the method includes the following steps.

[0784] First, obtain the cloth surface image. The industrial high-speed camera can collect stable and clear images from the cloth moving rapidly on the inspection table. For each roll of cloth, according to the rotation speed and a certain image size, collect all the cloth surface information of the cloth, and record the collected cloth surface information.

[0785] Furthermore, the acquired fabric image is input into the defect detection algorithm, and the defects in the fabric image can be identified by the intelligent defect detection algorithm to obtain the type information, size information and location information of the defects. For unknown types of defects, the unknown types of defects can be sent to the management platform for manual review. The identified defect information is stored in the storage server.

[0786] Furthermore, each defect may be scored based on the type, size and other information of the defect using the "10-point scoring method" or "4-point scoring method" commonly used in Western Europe and the United States.

[0787] Furthermore, the cloth can also be scored. By comprehensively scoring all defects in the entire roll of cloth, the quality grade of the roll of cloth can be automatically analyzed and statistically analyzed according to the preset cloth quality standards and a cloth inspection report can be generated.

[0788] Furthermore, the embodiment of the present application is a multi-layer detection. According to the needs of the production process, multi-layer algorithm recognition can be performed on the cloth in four links: weaving process, semi-finished products, finished products, and finished product warehousing, thereby improving the accuracy of defect detection.

[0789] Furthermore, it also includes recording the inventory information of cloth, and customers can trace product quality by querying QR codes.

[0790] The defect recognition algorithm provided in the embodiment of the present application can not only detect cloth defects, but also detect defects in objects such as aluminum materials, furniture and building materials. It only needs to adjust the corresponding defect type, and the embodiment of the present application will not be described in detail.

[0791] The present application also provides a surface defect quality inspection platform based on an image recognition algorithm, such as Figure 4 As shown in the figure, the platform includes: backend center, cloud center, Internet terminal and production workshop equipment; the backend center and production workshop equipment are connected to the cloud center, and the Internet terminal is connected to the backend center. The data interaction between various devices in the platform is realized through network transmission. The data transmission adopts TCP / IP network protocol, which can ensure the stability of data transmission at the bottom layer of the protocol.

[0792] Among them, the production workshop equipment is used to collect images of objects to be inspected, input the images of objects to be inspected into a preset defect detection model, obtain the detected defect type information, defect location information and defect size information, and send the detected object image information, defect type information, defect location information and defect size information to the cloud center.

[0793] Specifically, the production workshop equipment includes: quality inspection duty management platform, quality inspection equipment and integrated switches.

[0794] Among them, the quality inspection duty management platform is connected to the cloud center to view and count the data stored in the cloud center; it is also used to supervise the equipment in the production workshop. It mainly accesses the cloud center through the network, and can supervise all the quality inspection equipment in the workshop through the platform operation interface, remotely monitor the various processes of quality inspection in real time, and also view, count, and export the data information of the cloud center, providing corresponding data information support for production management.

[0795] The quality inspection equipment belongs to a set of testing equipment, including a quality inspection operation platform, a quality inspection processor, an access switch, an image recognition camera, a quality inspection controller, a fill light, and an audible and visual alarm.

[0796] The operation process of quality inspection equipment is as follows Figure 5 As shown in the figure, the production equipment in the production workshop acquires images by installing an image recognition camera, and transmits the image information of the camera to the quality inspection processor through the USB protocol. In order to meet the requirements of fast and stable transmission of image information, the USB interface protocol adopts the USB 3.0 protocol, and the transmission rate can reach 5.0Gbps, that is, the rate reaches 500MB / S. According to the type of production machine and the size of the produced cloth, a production equipment can be equipped with multiple image recognition cameras. In order to achieve fast and stable recognition and acquisition of images, each image recognition camera is equipped with an independent quality inspection processor.

[0797] The quality inspection processor performs defect recognition processing on the collected fabric information. Fabric defect recognition is detected by the built-in defect detection algorithm. The defect detection algorithm has very high requirements for light. By installing a fill light device just above the fabric to be inspected, the light is evenly compensated. The power on and off of the fill light is controlled by the quality inspection operation platform. In order to speed up the inspection of fabrics, the inspection area of ​​the fabric is divided into several small inspection surfaces, and each small inspection surface uses a quality inspection processor to run the defect recognition algorithm program. Using multiple quality inspection processors to inspect fabric defects is faster and has lower hardware costs than using a high-performance computer independently. Multiple quality inspection processors are connected to the network through an access switch.

[0798] The main functions of the quality inspection operation platform are four aspects. The first aspect is the display of real-time images, sample sampling of cloth, and display of defect images. The second aspect is to remotely set the parameters of the camera image acquisition, cloth batch information, etc. through the quality inspection operation platform. The third aspect is to control the rotation of the motor through the quality inspection controller, thereby controlling the forward, backward, and stop control of the cloth inspection machine. The fourth aspect is to control the sound and light alarm and fill light through the display interface operation of the quality inspection operation platform and the matching quality inspection controller. When the quality inspection operation platform finds defects, the quality inspection controller will promptly stop the transmission motor and issue a sound and light alarm, waiting for the operator to confirm the defects and start the motor. Each set of quality inspection operation platform includes a quality inspection operation platform, a quality inspection controller, a sound and light alarm, and a motor control interface.

[0799] The access switch provides an interface for the network connection between the quality inspection operation platform and the quality inspection processor, unified access to the cloud center, and provides stable data information transmission functions for various devices. The access switch adopts the storage and forwarding data method. When the access switch and the cloud center network fail, the quality inspection equipment can work normally and automatically upload data to the cloud center after the network failure is restored.

[0800] The integrated switch is a general-purpose Gigabit / 10 Gigabit network switch that connects all the testing equipment in the workshop with the private cloud center equipment to forward and transmit data.

[0801] The backend center includes the parsing server group, the intelligent server group, the storage server group, and the LAN terminal. The network connection of the backend center is to interconnect the data information of various devices through the core switch to realize data interaction.

[0802] The analysis server group is mainly responsible for the access management of LAN terminals and cloud server devices, the conversion and interactive transmission of Internet and LAN data, the preliminary analysis and screening of data, filtering of invalid or illegal data, and ensuring the security and stability of the background data center.

[0803] The smart server group belongs to the system data processing center, which is mainly responsible for network interactive transmission of data, data integration and statistical analysis, and intelligent processing of certain problems. Through cloud computing technology and unique data processing algorithms, it can quickly integrate and statistically analyze system data to achieve various functions; at the same time, combined with the built-in expert system, in the event of an emergency, it can intelligently match the emergency plan and handle events, discover and solve problems in the first place.

[0804] For example, each defect point is scored according to preset scoring rules and defect type information and defect size information to obtain a defect score; the score information, type information, location information, quantity information and size information of the defects in the object are counted, and the quality grade of the object is determined; an object detection report is generated based on the object's image information, production batch information and quality grade information.

[0805] The storage server group is the data storage center, which is mainly responsible for the rapid storage and rapid query of system data. It uses proprietary coded data warehouse technology to quickly build multi-dimensional storage space, and combines the private cloud operating system data communication protocol to quickly encode, classify and store system data, and accurately query it, providing efficient and stable data storage guarantee for the system.

[0806] For example, it is used to store defect type information, defect size information, defect location information, defect quantity information, defect score information, object image information, defect image information, object quality grade information, production batch information, and warehouse entry and exit record information in objects; it is also used to perform data fusion based on the stored data to generate structured data reports.

[0807] The LAN terminal is the management terminal of the associated platform system, including the platform maintenance management terminal, the quality inspection platform management terminal, etc. The platform maintenance management terminal is mainly responsible for the equipment configuration, equipment supervision (operating status, faults, etc.), remote software upgrade and update, statistical report query and other functions of the entire platform system; the quality inspection platform management terminal is mainly responsible for the management of each production workshop and the supervision of the production process, and the remote management of the workshop quality inspection equipment and other functions.

[0808] The cloud center is used to send the received object image information, defect type information, defect location information and defect size information to the background center.

[0809] The cloud center is the access terminal of the system's Internet center. The cloud center manages the accessed backend analysis servers, various Internet terminals, and quality inspection equipment in various production workshops through Internet network communication technology, and realizes data interactive transmission through the private cloud operating system data communication protocol, solving the problem of rapid access of various functional terminal users and functional data. At the same time, through the rules of system data interaction, the transmission direction, transmission speed, and transmission data volume of the system data are automatically controlled through algorithms, effectively reducing the consumption of public network bandwidth and traffic.

[0810] Among them, Internet terminals mainly provide remote supervision of quality inspection and production processes, including quality inspection and supervision mobile terminals and quality inspection and supervision terminals. The functions of these two types of terminals are the same, the difference is that the quality inspection and supervision mobile terminal is a mobile phone, and the quality inspection and supervision terminal is a PC. The functions provided mainly include workshop production process supervision, emergency warning notification, statistical report query, etc.

[0811] In an optional embodiment, the fabric defect detection at each stage of silk fabric production mainly includes fabric defect detection during the weaving process, fabric defect detection, and finished product defect detection. Therefore, the quality inspection platform of this application mainly includes a weaving quality inspection platform, such as Figure 6 As shown in the figure, it is mainly a defect detection platform in the cloth weaving process. According to the type of cloth product equipment and cloth size, 4 sets of lighting and imaging systems are used as visual systems, 4 sets of cloth image processing systems and 1 set of quality inspection operation platform are used. The cloth transmission system is provided by the motor of the factory's weaving production equipment, and the motor control signal is connected to the motor control interface of the quality inspection controller. The lighting system adopts the spotlight method.

[0812] It can also include a fabric quality inspection platform, such as Figure 7 As shown in the figure, it is mainly a defect detection platform in the process of cloth quality inspection. Adding a quality inspection platform can upgrade the original manual inspection method to an intelligent inspection method. This inspection is mainly for quality inspection after dyeing, printing and other processes. The cloth inspection quality inspection platform uses 4 sets of lighting and imaging systems as visual systems, 4 sets of cloth image processing systems and 1 set of quality inspection operation platform according to the type of cloth product equipment and cloth size. The cloth transmission system is provided by the motor of the factory's weaving production equipment, and the motor control signal is connected to the motor control interface of the quality inspection controller. The lighting system adopts the spotlight method.

[0813] It can also include finished product quality inspection platforms, such as Figure 8 As shown, it is mainly the last process of cloth quality inspection, which mainly inspects the quality of the cloth after cutting. The finished product quality inspection platform adopts a set of lighting and imaging systems as a visual system, a set of cloth image processing system and a set of quality inspection operation platform according to the type of cloth product equipment and cloth size. The cloth transmission system is provided by the motor of the factory's weaving production equipment, and the motor control signal is connected to the motor control interface of the quality inspection controller. The lighting system adopts the light panel method.

[0814] In one embodiment, the core of the private cloud operating system is the data communication protocol of the system. The data interaction between various devices in the system must be carried out strictly in accordance with the specifications. The communication protocol information between the independent physical network devices at each layer is different from each other. There are dynamic information codes generated by corresponding rules between each device. The cross-device cannot parse the information code that is not related to its own device, and the dynamic information code has a certain life cycle. The private cloud operating system will update its own dynamic information code from time to time. In this way, data security and data accuracy are guaranteed at the software level, and the stability and reliability of the system are improved.

[0815] In terms of functional protocols, functions and data are classified, analyzed, and integrated according to the problems that the system needs to solve. Equipment at all levels performs multi-level functional analysis and integration to ensure the stability of system functions.

[0816] The cloud center serves as the interface of the Internet terminal. In the private cloud operating system, the data communication between the terminal and the cloud center is set with a dynamic information code. The two parties interact with the system data through the device number (source device number and cloud server number) and the terminal network three-level dynamic information code. If the information code is correct, the functional protocol is parsed and the data is processed; if the information code is wrong, the data is discarded or isolated, and a record is made for the record.

[0817]

[0818] The cloud center is the interface of the Internet terminal, and the resolution server is the interface device of the background center. In the private cloud operating system, the data communication between the cloud center and the resolution server is set with a dynamic information code, and the two parties interact with the system data through the third-level dynamic information code of the access network. If the information code is correct, the functional protocol is parsed and the data is processed; if the information code is wrong, the data is discarded or isolated, and a record is made for the record.

[0819]

[0820]

[0821] Among them, the parsing server is the interface device of the background center, and the smart server is the system data fusion processing center. In the private cloud operating system, the data communication between the parsing server and the smart server is set with a dynamic information code, and the two parties interact with the system data through the three-level dynamic information code of the smart network. If the information code is correct, the functional protocol is parsed and the data is processed; if the information code is wrong, the data is discarded or isolated, and a record is made for filing.

[0822]

[0823]

[0824] Among them, the smart server serves as the system data fusion processing center, and the storage server serves as the system data storage center. In the private cloud operating system, the data communication between the smart server and the storage server sets a dynamic information code, and the two parties interact with the system data through the three-level dynamic information code of the storage network. If the information code is correct, the functional protocol is parsed and the data is processed; if the information code is wrong, the data is discarded or isolated, and a record is made for filing.

[0825]

[0826]

[0827] In one embodiment, the system uses adaptive virtual network technology to construct virtual channels for data transmission, allocates independent virtual channels according to data classification, and transmits different types of data through different virtual channels. Through the effective combination of different virtual channels, fast and controllable data interaction can be achieved. This not only solves the problem of data sticking, but also solves the problem of secure data transmission.

[0828] The construction method and working mode of the adaptive virtual channel are as follows:

[0829] (1) After the equipment is registered and uploaded, the system automatically allocates three independent virtual channels: data, image, and voice.

[0830] (2) Real-time monitoring of virtual channels (loop detection) to automatically adjust data transmission flow according to the amount of data in different time periods, fully utilize system resources, and ensure stable data transmission.

[0831] (3) Analyze valid data. Analyze the data of different virtual channels according to data classification, extract valid data, and remove invalid data.

[0832] (4) Forward valid data: forward valid data through the corresponding virtual channel, discard or record invalid data, reduce the amount of transmitted data, and save system resources.

[0833] (5) Store valid data. Quickly classify and store data obtained from different virtual channels. Only store valid data to save storage space.

[0834] (6) Integrate valid data to obtain functional data.

[0835] (7) Send functional data.

[0836] (8) Display functional data.

[0837] According to the number of system devices or users, virtual channels can be flexibly allocated and used to avoid the contradiction between resource waste and resource shortage; at the same time, the data transmission flow can be adaptively adjusted to fully utilize system resources and ensure stable data transmission.

[0838] According to the surface defect platform based on the image recognition algorithm provided by the embodiment of the present application, visual recognition, cloud computing, and big data technologies are implanted in the mechanical automation factory to upgrade it to an intelligent factory. The platform can effectively reduce the labor cost of production, reduce damage to the naked eye, improve production quality, and improve production efficiency. Because manual inspection is slow, it consumes a lot of manpower, and due to subjective factors, quality inspection lacks consistency and accuracy. In order to simulate the manual inspection process, machine vision technology is used to replace artificial vision to collect images of objects. Visual recognition algorithms and cloud computing technology can be used to detect and locate surface defects of objects. Big data technology is used to record and analyze each defect, score and rate objects, track in and out of the warehouse, and trace product quality.

[0839] It should be noted that the surface defect quality inspection platform based on image recognition algorithm provided in the above embodiment only uses the division of the above functional modules as an example when executing the surface defect quality inspection method based on image recognition algorithm. In actual application, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the surface defect quality inspection platform based on image recognition algorithm provided in the above embodiment and the surface defect quality inspection method based on image recognition algorithm belong to the same concept.

[0840] The above embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present invention. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention patent shall be subject to the attached claims.

Claims

1. A surface defect quality inspection platform based on image recognition algorithm, characterized in that: The platform includes: Backstage center, cloud center, Internet terminal and production workshop equipment; The production workshop equipment is used to collect images of objects to be detected, input the images of the objects to be detected into a preset defect detection model, obtain the detected defect type information, defect location information and defect size information, and send the detected object image information, defect type information, defect location information and defect size information to the cloud center; The production workshop equipment includes: a camera for collecting images of objects to be detected; A quality inspection processor, used to input the image of the object to be inspected into a preset defect detection model to obtain information on the type, location and size of the detected defects; The method includes converting the format of the object image to be detected into an RGB image to obtain a converted object image; evenly dividing the converted object image to obtain a plurality of image blocks with equal length and width; calculating the characteristic values ​​of the pixels of each image block, and determining different types of defects according to the calculated characteristic values; including: calculating the characteristic values ​​of the target type of all pixels in each image block, including calculating the red characteristic value, green characteristic value, blue characteristic value, yellow characteristic value, cyan characteristic value, purple characteristic value, white characteristic value, black characteristic value, maximum characteristic value, minimum characteristic value, and average characteristic value of all pixels in each image block; Red eigenvalue: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies: R>G, and R>B; Then calculate the red feature value of the pixel point, and the calculation formula is as follows: BTR nH*nW (X)=((R nH*nW (X)-G nH*nW (X))+(R nH*nW (X)-B nH*nW (X))) Otherwise BTR nH*nW (X) = 0, the red feature values ​​of all pixels in the image block are counted, and the statistical formula is as follows: Green eigenvalue: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula: G>R, and G>B; Then calculate the green feature value of the pixel point, and the calculation formula is as follows: BTG nH*nW (X)=((G nH*nW (X)-R nH*nW (X))+(G nH*nW (X)-B nH*nW (X))) Otherwise BTG nH*nW (X) = 0, the green feature values ​​of all pixels in the image block are counted, and the statistical formula is as follows: Blue eigenvalue: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula: B>R, and B>G; Then calculate the blue feature value of the pixel point, and the calculation formula is as follows: BTB nH*nW (X)=((B nH*nW (X)-R nH*nW (X))+(B nH*nW (X)-G nH*nW (X))) Otherwise BTB nH*nW (X) = 0, the blue feature values ​​of all pixels in the image block are counted, and the statistical formula is as follows: Yellow eigenvalue: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula: R>B and G>B; Then calculate the yellow feature value of the pixel point, and the calculation formula is as follows: BTH nH*nW (X)=((R nH*nW (X)-B nH*nW (X))+(G nH*nW (X)-B nH*nW (X))) Otherwise BTH nH*nW (X) = 0, the yellow feature values ​​of all pixels in the image block are counted, and the statistical formula is as follows: Cyan feature value: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula: R>G and B>G; Then calculate the cyan feature value of the pixel point, and the calculation formula is as follows: BTT nH*nW (X)=((R nH*nW (X)-G nH*nW (X))+(B nH*nW (X)-G nH*nW (X))) Otherwise BTT nH*nW (X) = 0, the cyan feature values ​​of all pixels in the image block are counted, and the statistical formula is as follows: Purple eigenvalue: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula: G>R and B>R; Then calculate the purple eigenvalue of the pixel point, and the calculation formula is as follows: BTZ nH*nW (X)=((G nH*nW (X)-R nH*nW (X))+(B nH*nW (X)-R nH*nW (X))) Otherwise BTZ nH*nW (X) = 0, the purple feature values ​​of all pixels in the image block are counted, and the statistical formula is as follows: White feature value: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula: G = R and B = R; The white feature values ​​of all pixels in the image block are counted. The statistical formula is as follows: Black feature value: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula: G = R and B = R; The black feature values ​​of all pixels in the image block are counted. The statistical formula is as follows: Minimum eigenvalue: Count the eigenvalues ​​of all pixels in the image block. The statistical formula is as follows: Maximum eigenvalue: Count the eigenvalues ​​of all pixels in the image block. The statistical formula is as follows: Among them, the image width is nW, the image height is nH, the image width is divided into nmw = 256 parts on average, and the image height is divided into nmh = 256 parts on average. After segmentation, nmh*nmw image blocks can be obtained. The width of an image block is w, the height of an image block is h, and the starting pixel point is (a1, b1), a is equal to w, and b is equal to h; Average eigenvalue: Calculate the average value of the red eigenvalue, green eigenvalue, blue eigenvalue, yellow eigenvalue, cyan eigenvalue, purple eigenvalue, white eigenvalue, black eigenvalue, maximum eigenvalue, and minimum eigenvalue in the image block; Determine different types of defects according to the calculation of different types of characteristic values; including: calculate the detection value of the characteristic value of the target type according to the characteristic value of the target type; calculate the statistical average value of the characteristic value of the target type according to the detection value of the characteristic value of the target type; determine the suspected defects of the target type according to the statistical average value of the characteristic value of the target type; when the number of suspected defects of the target type in the target area of ​​the image block is greater than a preset threshold, determine the target area as a defect of the target type; classify and annotate different types of defects, and determine the position information and size information of each defect according to the coordinate information and size information of the annotation box; An integrated switch, used for interconnecting with the cloud center, and sending the object image, defect type information, defect location information, and defect size information to the cloud center; a fill light, used for providing light source for the camera; an audible and visual alarm, used for issuing audible and visual alarm information after detecting defect information; The cloud center is used to send the received object image information, defect type information, defect location information and defect size information to the backend center; The backend center is used to collect statistics on the object image information, defect type information, defect location information and defect size information to generate an object defect detection report; The Internet terminal is used to query the object defect detection report.

2. The platform according to claim 1, characterized in that The production workshop equipment also includes: The quality inspection duty management platform is connected to the cloud center and is used to view and count the data stored in the cloud center; it is also used to supervise the equipment in the production workshop.

3. The platform according to claim 1, characterized in that: The production workshop equipment also includes: The quality inspection operation platform is used to display the object image and defect image in real time; it is also used to remotely set the parameters of the camera to collect images; it is also used to control the rotation of the motor; it is also used to control the fill light and the sound and light alarm.

4. The platform according to claim 1, characterized in that: The backend center includes: The intelligent server group is used to score each defect according to the preset scoring rules and the defect type information and defect size information to obtain a defect score; Used to count the score information, type information, location information, quantity information and size information of defects in objects, and determine the quality level of objects; Used to generate object detection reports based on the object's image information, production batch information, and quality grade information.

5. The platform according to claim 1, characterized in that: The backend center also includes: A storage server group, used to store defect type information, defect size information, defect location information, defect quantity information, defect score information, object image information, defect image information, object quality grade information, production batch information, and storage record information; It is also used to perform data fusion based on the stored data and generate structured data reports.

6. The platform according to claim 1, characterized in that: The Internet terminal comprises: Quality inspection and supervision mobile phone terminals and quality inspection and supervision computer terminals are used to query object defect detection reports and data reports, are also used for emergency warnings, and are also used to supervise production processes.

7. A surface defect quality inspection method based on image recognition algorithm, characterized in that: The method comprises: Acquire an image of an object to be detected; Input the object image to be detected into a preset defect detection model to obtain the detected defect type information, defect location information and defect size information; including: the defect detection model performs format conversion on the object image to be detected, converts it into an RGB image, and obtains a converted object image; averagely divides the converted object image to obtain a plurality of image blocks with equal length and width; calculates the characteristic values ​​of the pixels of each image block, and determines different types of defects according to the calculated characteristic values; including: calculating the characteristic values ​​of the target type of all pixels in each image block, including calculating the red characteristic value, green characteristic value, blue characteristic value, yellow characteristic value, cyan characteristic value, purple characteristic value, white characteristic value, black characteristic value, maximum characteristic value, minimum characteristic value, and average characteristic value of all pixels in each image block; Red eigenvalue: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies: R>G, and R>B; Then calculate the red feature value of the pixel point, and the calculation formula is as follows: BTR nH*nW (X)=((R nH*nW (X)-G nH*nW (X))+(R nH*nW (X)-B nH*nW (X))) Otherwise BTR nH*nW (X) = 0, the red feature values ​​of all pixels in the image block are counted, and the statistical formula is as follows: Green eigenvalue: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula: G>R, and G>B; then calculate the green eigenvalue of the pixel point, and the calculation formula is as follows: BTG nH*nW (X)=((G nH*nW (X)-R nH*nW (X))+(G nH*nW (X)-B nH*nW (X))) Otherwise BTG nH*nW (X) = 0, the green feature values ​​of all pixels in the image block are counted, and the statistical formula is as follows: Blue eigenvalue: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula: B>R, and B>G; then calculate the blue eigenvalue of the pixel point, and the calculation formula is as follows: BTB nH*nW (X)=((B nH*nW (X)-R nH*nW (X))+(B nH*nW (X)-G nH*nW (X))) Otherwise BTB nH*nW (X) = 0, the blue feature values ​​of all pixels in the image block are counted, and the statistical formula is as follows: Yellow eigenvalue: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula: R>B and G>B; then calculate the yellow eigenvalue of the pixel point, and the calculation formula is as follows: BTH nH*nW (X)=((R nH*nW (X)-B nH*nW (X))+(G nH*nW (X)-B nH*nW (X))) Otherwise BTH nH*nW (X) = 0, the yellow feature values ​​of all pixels in the image block are counted, and the statistical formula is as follows: Cyan feature value: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula: R>G and B>G; then calculate the cyan feature value of the pixel point, and the calculation formula is as follows: BTT nH*nW (X)=((R nH*nW (X)-G nH*nW (X))+(B nH*nW (X)-G nH*nW (X))) Otherwise BTT nH*nW (X) = 0, the cyan feature values ​​of all pixels in the image block are counted, and the statistical formula is as follows: Purple eigenvalue: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula: G>R and B>R; Then calculate the purple eigenvalue of the pixel point, and the calculation formula is as follows: BTZ nH*nW (X)=((G nH*nW (X)-R nH*nW (X))+(B nH*nW (X)-R nH*nW (X))) Otherwise BTZ nH*nW (X) = 0, the purple feature values ​​of all pixels in the image block are counted, and the statistical formula is as follows: White feature value: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula: G = R and B = R; The white feature values ​​of all pixels in the image block are counted. The statistical formula is as follows: Black feature value: If the relationship between the R, G, and B values ​​of the pixel point X(i, j) satisfies the following formula: G = R and B = R; The black feature values ​​of all pixels in the image block are counted. The statistical formula is as follows: Minimum eigenvalue: Count the eigenvalues ​​of all pixels in the image block. The statistical formula is as follows: Maximum eigenvalue: Count the eigenvalues ​​of all pixels in the image block. The statistical formula is as follows: Among them, the image width is nW, the image height is nH, the image width is divided into nmw = 256 parts on average, and the image height is divided into nmh = 256 parts on average. After segmentation, nmh*nmw image blocks can be obtained. The width of an image block is w, the height of an image block is h, and the starting pixel point is (a1, b1), a is equal to w, and b is equal to h; Average eigenvalue: Calculate the average value of the red eigenvalue, green eigenvalue, blue eigenvalue, yellow eigenvalue, cyan eigenvalue, purple eigenvalue, white eigenvalue, black eigenvalue, maximum eigenvalue, and minimum eigenvalue in the image block; Determine different types of defects according to the calculation of different types of characteristic values; including: calculating the detection value of the characteristic value of the target type according to the characteristic value of the target type; calculating the statistical average value of the characteristic value of the target type according to the detection value of the characteristic value of the target type; determining the suspected defects of the target type according to the statistical average value of the characteristic value of the target type; when the number of suspected defects of the target type in the target area of ​​the image block is greater than a preset threshold, determine the target area as a defect of the target type; Different types of defects are classified and marked, and the location information and size information of each defect are determined according to the coordinate information and size information of the marking box; The defect type information, defect location information and defect size information are counted to generate a defect detection report.

Images