A method and system for quality inspection of cable production

Abstract

The present invention discloses a method and system for detecting the production quality of cables, including: acquiring the surface image of the cable insulation layer with qualified production quality and generating an image fingerprint, which is stored in the qualified cable database; after the current cable completes insulation injection molding, three camera devices surrounding the cable take images of the cable insulation layer once every s seconds; the images of the cable insulation layer at three different angles taken are respectively subjected to image compression and grayscale processing, and then discrete cosine Fourier transform is performed. After generating a transform coefficient matrix, a sub-matrix is taken as the low-frequency domain matrix; a binary sequence is generated through the low-frequency domain matrix and converted into a hexadecimal string; the image fingerprint is obtained from the qualified cable database, and the Hamming distance is calculated with the hexadecimal string to determine whether there are defects on the surface of the insulation layer. The present invention realizes automatic quality monitoring during the cable production process, achieves high detection accuracy while ensuring the detection speed, and has a simple calculation process.

Description

Technical Field

[0001] The present invention relates to the field of cable production quality inspection, and specifically relates to a cable production quality inspection method and system. Background Art

[0002] With the development of society, cables have become indispensable industrial products. The social demand is increasing continuously, and the number of manufacturers and production scale are also expanding continuously. The characteristic of a cable is that it can conduct electricity inside but is insulated outside. It is composed of a structure similar to a rope, manifested as one, several or several groups of wires, and a highly insulating covering material is wrapped around the periphery of these wires to protect the wires and provide electrical isolation. In the production process of cables, it is particularly important to detect the production quality of the wire insulation layer, which is related to the overall production quality of the cable and the problem of electrical safety.

[0003] However, most of the existing cable production quality inspections are carried out by manual sampling, and automatic quality monitoring during the production process has not been realized. On the other hand, most of the inspection tasks in industrial production have relatively high requirements for speed, with the precondition of not reducing production efficiency. However, the existing machine learning models and deep learning models are usually large in scale, slow in detection speed and difficult to train. How to improve the detection speed while ensuring the detection accuracy is an urgent problem to be solved. Summary of the Invention

[0004] Aiming at the above technical deficiencies, the purpose of the present invention is to provide a cable production quality inspection method and system, aiming to solve the problems that most of the existing cable production quality inspections are carried out by manual sampling, and automatic quality monitoring during the production process has not been realized. In addition, the existing machine learning models and deep learning models are usually large in scale, slow in detection speed and difficult to train, and it is difficult to improve the detection speed while ensuring the detection accuracy.

[0005] In view of the above problems, the present application provides a cable production quality inspection method and system.

[0006] In the first aspect disclosed in the present application, a cable production quality inspection method is provided, and the method includes the following steps:

[0007] Step 1: Obtain the surface image of the cable insulation layer with qualified production quality, generate the image fingerprint of the surface image of the cable insulation layer with qualified production quality, and store it in the qualified cable database;

[0008] Step 2: After the current cable completes insulation injection molding, the image of the cable insulation layer is taken once every s seconds by three camera devices surrounding the cable. Among them, the three camera devices surrounding the cable are separated at equal intervals of 120°, and s is less than the ratio of the distance that the cable moves per second on the production line to the width of the shooting range of the camera device;

[0009] Step 3: Compress and grayscale the cable insulation layer images taken by three camera devices surrounding the cable at three different angles respectively, to generate three compressed grayscale images with a resolution of 32×32 pixels;

[0010] Step 4: Perform discrete cosine Fourier transform on the three compressed grayscale images with a resolution of 32×32 pixels respectively, to generate three transformation coefficient matrices with dimensions of 32×32. Take sub-matrices with dimensions of 8×8 from the upper left corners of the three transformation coefficient matrices with dimensions of 32×32 respectively, to generate three low-frequency domain matrices;

[0011] Step 5: For each low-frequency domain matrix, calculate the corresponding matrix average value, and compare each element of the low-frequency domain matrix with the corresponding matrix average value. If it is greater than or equal to the low-frequency domain matrix average value, record it as 1; if it is less than the low-frequency domain matrix average value, record it as 0, to generate three 64-bit binary sequences. Then convert the three 64-bit binary sequences into hexadecimal respectively, to generate three strings with a length of 16;

[0012] Step 6: Obtain the image fingerprint of the current cable from the qualified cable database, calculate the Hamming distance with the three strings with a length of 16 in sequence and sum them up. If the sum is greater than the determination threshold, it is determined that there is a suspected surface defect in the insulation layer.

[0013] Preferably, the specific steps of step 1 are as follows:

[0014] Step 1.1: Obtain the surface image of the cable insulation layer with qualified production quality, perform image compression and grayscale processing, and generate a compressed grayscale standard image with a resolution of 32×32 pixels for the surface image of the cable insulation layer with qualified production quality;

[0015] Step 1.2: Perform discrete cosine Fourier transform on the compressed grayscale standard image generated in step 1.1, generate a standard transformation coefficient matrix with dimensions of 32×32 for the surface image of the cable insulation layer with qualified production quality, and take a sub-matrix with dimensions of 8×8 from the upper left corner of the standard transformation coefficient matrix, to generate a standard low-frequency domain matrix;

[0016] Step 1.3: Calculate the matrix average value of the standard low-frequency domain matrix, and compare each element of the standard low-frequency domain matrix with this matrix average value. If it is greater than or equal to this matrix average value, record it as 1; if it is less than this matrix average value, record it as 0, to generate a 64-bit binary standard sequence. Then convert the 64-bit binary standard sequence into hexadecimal, to generate a standard string with a length of 16 as the image fingerprint of the surface image of the cable insulation layer with qualified production quality;

[0017] Step 1.4: Using the production batch of the currently produced qualified cables as the primary key, store the image fingerprints of the surfaces of the insulating layers of the produced qualified cables in the form of strings in the qualified cable database.

[0018] Preferably, the specific steps of step 3 are as follows:

[0019] Step 3.1: Perform average pooling on the images of the cable insulating layers taken from three different angles by three camera devices surrounding the cable respectively to generate three compressed images with a resolution of 32×32 pixels.

[0020] Step 3.2: Perform grayscale processing on the three compressed images with a resolution of 32×32 pixels respectively by the weighted average method to generate three compressed grayscale images with a resolution of 32×32 pixels.

[0021] Preferably, the specific steps of step 4 are as follows:

[0022] Step 4.1: Convert the three compressed grayscale images with a resolution of 32×32 pixels into matrix form, denoted as P x , where x = 1, 2, 3, is used to represent the labels of the three compressed grayscale images with a resolution of 32×32 pixels, and the dimension of P x is 32×32;

[0023] Step 4.2: Calculate and generate three transformation coefficient matrices with a dimension of 32×32 by the method of D x =AP x A T , where D x is the transformation coefficient matrix with a dimension of 32×32, matrix A is the transformation matrix with a dimension of 32×32, , and the value of N is the dimension value of matrix A and matrix P x , which is 32 here, and i and j are the row index and column index of matrix A respectively;

[0024] Step 4.3: Take out sub-matrices with a dimension of 8×8 from the upper left corners of the three transformation coefficient matrices with a dimension of 32×32 respectively to generate three low-frequency domain matrices.

[0025] Preferably, the specific steps of step 6 are as follows:

[0026] Step 6.1: Obtain the image fingerprint of the current cable from the qualified cable database;

[0027] Step 6.2: If the length of the image fingerprint is greater than 16, intercept and retain the first 16 bits of the image fingerprint to make the length of the image fingerprint equal to 16. If the length of the image fingerprint is less than 16, supplement random characters at the end of the image fingerprint to make the length of the image fingerprint equal to 16;

[0028] Step 6.3: Compare the image fingerprint with three strings each of length 16 character by character, count the number of different characters at the corresponding positions, generate three Hamming distance values, and sum up the three Hamming distance values to generate the total Hamming distance;

[0029] Step 6.4: If the total Hamming distance is greater than the determination threshold, it is determined that the current cable is suspected of having a surface defect in the insulating layer. Among them, the determination threshold is set to 6, and the surface defects of the insulating layer include surface unevenness, cracks or bubbles.

[0030] In the second aspect disclosed in the present application, a cable production quality detection system is provided. The system is used to implement the above-mentioned cable production quality detection method. The system includes:

[0031] A database module, which is used to obtain the surface images of the cable insulating layers with qualified production quality, generate the image fingerprints of the surface images of the cable insulating layers with qualified production quality, and store them in the qualified cable database;

[0032] A camera module, which is used to take images of the cable insulating layer once every s seconds through three camera devices surrounding the cable after the current type of cable is completed with insulating injection molding. Among them, the three camera devices surrounding the cable are separated at equal intervals of 120°, and s is less than the ratio of the distance that the cable moves per second on the production line to the width of the shooting range of the camera device;

[0033] A preprocessing module, which is used to respectively compress and grayscale the three cable insulating layer images taken by the three camera devices surrounding the cable to generate three compressed grayscale images with a resolution of 32×32 pixels;

[0034] A transformation module, which is used to respectively perform discrete cosine Fourier transformation on the three compressed grayscale images with a resolution of 32×32 pixels to generate three transformation coefficient matrices with a dimension of 32×32, and respectively extract sub-matrices with a dimension of 8×8 from the upper left corners of the three transformation coefficient matrices with a dimension of 32×32 to generate three low-frequency domain matrices;

[0035] A hashing module, which is used to calculate the corresponding matrix average value for each low-frequency domain matrix, compare each element of the low-frequency domain matrix with the corresponding matrix average value, record it as 1 if it is greater than or equal to the low-frequency domain matrix average value, and record it as 0 if it is less than the low-frequency domain matrix average value, generate three 64-bit binary sequences, and then respectively convert the three 64-bit binary sequences into hexadecimal to generate three strings with a length of 16;

[0036] A detection module, which is used to obtain the image fingerprint of the current type of cable from a qualified cable database, calculate the Hamming distance with three strings of length 16 in sequence and sum them. If the sum is greater than the determination threshold, it is determined that there is a suspected surface defect in the insulating layer.

[0037] The beneficial effects of the present invention are as follows:

[0038] (1) It realizes the automated quality monitoring of the insulating layer during the cable production process, and solves the problems of low efficiency and easy mistakes in the existing manual sampling inspection for cable production quality detection.

[0039] (2) It does not use existing large-scale machine learning models or deep learning models, does not require a large amount of data sets for training, and achieves high detection accuracy while ensuring the detection speed, without reducing the production efficiency.

[0040] (3) The calculation process is simple, and the detection task can be completed only with a few matrix transformations and string comparisons, without consuming too much computing resources. Description of the Drawings

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0042] Figure 1 It is an overall flowchart of a cable production quality detection method.

[0043] Figure 2 It is an overall structure diagram of a cable production quality detection system. Detailed Embodiments

[0044] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0045] As Figure 1 shown, the embodiments of the present application provide a cable production quality detection method, and the method includes the following steps:

[0046] Step 1: Obtain the surface image of the cable insulation layer with qualified production quality, generate the image fingerprint of the surface image of the cable insulation layer with qualified production quality, and store it in the qualified cable database;

[0047] Step 2: After the current cable completes insulation injection molding, use three camera devices surrounding the cable to take images of the cable insulation layer once every s seconds. Among them, the three camera devices surrounding the cable are separated at equal intervals of 120°, and s is less than the ratio of the distance the cable moves per second on the production line to the width of the shooting range of the camera device;

[0048] Step 3: Respectively perform image compression and grayscale processing on the three cable insulation layer images taken from different angles by the three camera devices surrounding the cable to generate three compressed grayscale images with a resolution of 32×32 pixels;

[0049] Step 4: Respectively perform discrete cosine Fourier transform on the three compressed grayscale images with a resolution of 32×32 pixels to generate three transformation coefficient matrices with dimensions of 32×32. Take sub-matrices with dimensions of 8×8 from the upper left corners of the three transformation coefficient matrices with dimensions of 32×32 to generate three low-frequency domain matrices;

[0050] Step 5: For each low-frequency domain matrix, calculate the corresponding matrix average value, and compare each element of the low-frequency domain matrix with the corresponding matrix average value. If it is greater than or equal to the low-frequency domain matrix average value, record it as 1; if it is less than the low-frequency domain matrix average value, record it as 0 to generate three 64-bit binary sequences. Then, respectively convert the three 64-bit binary sequences into hexadecimal to generate three strings with a length of 16;

[0051] Step 6: Obtain the image fingerprint of the current cable from the qualified cable database, calculate the Hamming distance with the three strings with a length of 16 in sequence and sum them. If the sum is greater than the determination threshold, it is determined that there is a suspected surface defect in the insulation layer.

[0052] Furthermore, Step 1 specifically includes the following steps:

[0053] Step 1.1: Obtain the surface image of the cable insulation layer with qualified production quality, perform image compression and grayscale processing, and generate a compressed grayscale standard image with a resolution of 32×32 pixels for the surface image of the cable insulation layer with qualified production quality;

[0054] Step 1.2: Perform discrete cosine Fourier transform on the compressed grayscale standard image generated in Step 1.1 to generate a standard transformation coefficient matrix with dimensions of 32×32 for the surface image of the cable insulation layer with qualified production quality. Take a sub-matrix with dimensions of 8×8 from the upper left corner of the standard transformation coefficient matrix to generate a standard low-frequency domain matrix;

[0055] Step 1.3: Calculate the matrix average value of the standard low-frequency domain matrix, and compare each element of the standard low-frequency domain matrix with this matrix average value. If it is greater than or equal to the matrix average value, it is recorded as 1; if it is less than the matrix average value, it is recorded as 0, generating a 64-bit binary standard sequence. Then, convert the 64-bit binary standard sequence into hexadecimal to generate a standard string with a length of 16 as the image fingerprint of the surface image of the cable insulation layer with qualified production quality.

[0056] Step 1.4: Using the production batch of the cable with qualified production quality as the primary key, store the image fingerprint of the surface image of the cable insulation layer with qualified production quality in the qualified cable database in the form of a string.

[0057] Specifically, the image fingerprint can characterize the features of the surface image of the cable insulation layer with qualified production quality, and project the image information into the semantic space.

[0058] Furthermore, Step 3 specifically includes the following steps:

[0059] Step 3.1: Perform average pooling on the cable insulation layer images taken from three different angles by three camera devices surrounding the cable respectively, generating three compressed images with a resolution of 32×32 pixels.

[0060] Step 3.2: Perform grayscale processing on the three compressed images with a resolution of 32×32 pixels respectively by the weighted average method, generating three compressed grayscale images with a resolution of 32×32 pixels.

[0061] Specifically, the steps of performing image compression and grayscale processing on the image can simplify unnecessary calculations and improve the speed and computational speed of the method.

[0062] Furthermore, Step 4 specifically includes the following steps:

[0063] Step 4.1: Convert the three compressed grayscale images with a resolution of 32×32 pixels into matrix form, denoted as P x , where x = 1, 2, 3, is used to represent the labels of the three compressed grayscale images with a resolution of 32×32 pixels, and the dimension of P x is 32×32;

[0064] Step 4.2: Calculate and generate three transformation coefficient matrices with a dimension of 32×32 in the way of D x =AP x A T , where D x is the transformation coefficient matrix with a dimension of 32×32, matrix A is the transformation matrix with a dimension of 32×32, , and the value of N is the matrix A and matrix P xThe dimensional value here is 32, where i and j are the row index and column index of matrix A respectively;

[0065] Step 4.3: Respectively extract sub-matrices with dimensions of 8×8 from the upper left corners of the three transformation coefficient matrices with dimensions of 32×32 to generate three low-frequency domain matrices.

[0066] Specifically, the discrete cosine Fourier transform converts the image from the spatial domain at the pixel level to the frequency domain, obtaining a transformation coefficient matrix with dimensions of 32×32 to capture the low-frequency information of the image. After the discrete cosine Fourier transform, the frequency characteristics of the image are concentrated in the upper left corner of the transformation coefficient matrix. Usually, the sub-matrix with dimensions of 8×8 in the upper left corner of the transformation coefficient matrix is retained because the sub-matrix with dimensions of 8×8 in the upper left corner of the transformation coefficient matrix presents the lowest frequency in the picture.

[0067] Furthermore, step 6 specifically includes the following steps:

[0068] Step 6.1: Obtain the image fingerprint of the current cable from the qualified cable database;

[0069] Step 6.2: If the length of the image fingerprint is greater than 16, intercept and retain the first 16 bits of the image fingerprint to make the length of the image fingerprint equal to 16. If the length of the image fingerprint is less than 16, supplement random characters at the end of the image fingerprint to make the length of the image fingerprint equal to 16;

[0070] Step 6.3: Compare the image fingerprint with the three strings with a length of 16 one by one character by character, generate three Hamming distance values, and sum the three Hamming distance values to generate the total Hamming distance;

[0071] Step 6.4: If the total Hamming distance is greater than the determination threshold, it is determined that the current cable is suspected of having surface defects in the insulating layer, where the determination threshold is set to 6, and the surface defects in the insulating layer include surface unevenness, cracks or bubbles.

[0072] Specifically, by comparing using the Hamming distance in the above-mentioned manner, the similarity of the images can be evaluated. The smaller the Hamming distance, the more similar the images are.

[0073] In summary, a cable production quality detection method provided by an embodiment of the present application has the following technical effects:

[0074] (1) Realize the automated quality monitoring of the insulating layer during the cable production process, and solve the problems of low efficiency and easy mistakes in the existing manual sampling inspection for cable production quality detection.

[0075] (2) It does not use existing large-scale machine learning models or deep learning models, does not require a large amount of data sets for training, achieves high detection accuracy while ensuring the detection speed, and does not reduce production efficiency.

[0076] (3) The calculation process is simple. Only a few matrix transformations and string comparisons are required to complete the detection task, and it does not consume excessive computing resources.

[0077] Based on the same inventive concept as a cable production quality detection method in the foregoing embodiment, as Figure 2 shown, the present application provides a cable production quality detection system, and the system includes:

[0078] A database module, which is used to obtain the surface images of the cable insulation layers with qualified production quality, generate the image fingerprints of the surface images of the cable insulation layers with qualified production quality, and store them in the qualified cable database;

[0079] A camera module, which is used to take images of the cable insulation layer once every s seconds through three camera devices surrounding the cable after the current type of cable is completed with insulation injection molding. Among them, the three camera devices surrounding the cable are separated at equal intervals of 120°, and s is less than the ratio of the distance that the cable moves per second on the production line to the width of the shooting range of the camera device;

[0080] A preprocessing module, which is used to respectively compress and grayscale the three cable insulation layer images taken by the three camera devices surrounding the cable to generate three compressed grayscale images with a resolution of 32×32 pixels;

[0081] A transformation module, which is used to respectively perform discrete cosine Fourier transforms on the three compressed grayscale images with a resolution of 32×32 pixels to generate three transformation coefficient matrices with dimensions of 32×32, and respectively take out sub-matrices with dimensions of 8×8 from the upper left corners of the three transformation coefficient matrices with dimensions of 32×32 to generate three low-frequency domain matrices;

[0082] A hash module, which is used to calculate the corresponding matrix average value for each low-frequency domain matrix, compare each element of the low-frequency domain matrix with the corresponding matrix average value, record it as 1 if it is greater than or equal to the low-frequency domain matrix average value, record it as 0 if it is less than the low-frequency domain matrix average value, generate three 64-bit binary sequences, and then respectively convert the three 64-bit binary sequences into hexadecimal to generate three strings with a length of 16;

[0083] A detection module, which is used to obtain the image fingerprint of the current type of cable from a qualified cable database, calculate the Hamming distance with three strings of length 16 in sequence and sum them up. If the sum is greater than the determination threshold, it is determined that there is a suspected surface defect in the insulating layer.

[0084] Through the foregoing detailed description of a cable production quality detection method in this specification, those skilled in the art can clearly know a cable production quality detection system in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple. For related parts, refer to the description in the method part.

[0085] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0086] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A cable production quality detection method, characterized in that: The method comprises the following steps: Step 1: Obtain a surface image of the insulation layer of a cable with qualified production quality, generate an image fingerprint of the surface image of the insulation layer of a cable with qualified production quality, and store it in a qualified cable database; Step 2: After the insulation injection molding of the current cable is completed, an image of the cable insulation layer is captured every s seconds by three cameras surrounding the cable, wherein the three cameras surrounding the cable are equally spaced 120° apart, and s is less than the ratio of the cable moving distance per second on the production line to the width of the camera shooting range; Step 3: The cable insulation layer images at three different angles taken by three cameras surrounding the cable are compressed and grayscaled to generate three compressed grayscale images with a resolution of 32×32 pixels; Step 4: Perform discrete cosine Fourier transform on the three compressed grayscale images with a resolution of 32×32 pixels, generate three transformation coefficient matrices with a dimension of 32×32, and take out sub-matrices with a dimension of 8×8 from the upper left corners of the three transformation coefficient matrices with a dimension of 32×32, to generate three low-frequency domain matrices; Step 5: For each low-frequency domain matrix, calculate the corresponding matrix average value, and compare each element of the low-frequency domain matrix with the corresponding matrix average value. If it is greater than or equal to the low-frequency domain matrix average value, it is recorded as 1, and if it is less than the low-frequency domain matrix average value, it is recorded as 0. Generate three 64-bit binary sequences, and then convert the three 64-bit binary sequences into hexadecimal respectively to generate three strings of length 16; Step 6: Obtain the image fingerprint of the current cable from the qualified cable database, calculate the Hamming distance with three strings of length 16 in turn and sum them up. If the sum is greater than the judgment threshold, it is judged that there is a suspected surface defect of the insulation layer.

2. A cable production quality detection method according to claim 1, characterized in that: The step 1 specifically comprises the following steps: Step 1.1: Obtain a surface image of a cable insulation layer with qualified production quality, perform image compression and grayscale processing, and generate a compressed grayscale standard image with a resolution of 32×32 pixels for the surface image of the cable insulation layer with qualified production quality; Step 1.2: Perform discrete cosine Fourier transform on the compressed grayscale standard image generated in step 1.1 to generate a standard transformation coefficient matrix with a dimension of 32×32 for producing a cable insulation layer surface image with qualified production quality, and take out a sub-matrix with a dimension of 8×8 from the upper left corner of the standard transformation coefficient matrix to generate a standard low-frequency domain matrix; Step 1.3: Calculate the matrix average value of the standard low-frequency domain matrix, and compare each element of the standard low-frequency domain matrix with the matrix average value. If it is greater than or equal to the matrix average value, it is recorded as 1, and if it is less than the matrix average value, it is recorded as 0, to generate a 64-bit binary standard sequence, and then convert the 64-bit binary standard sequence into hexadecimal to generate a standard string of length 16 as the image fingerprint of the surface image of the cable insulation layer of qualified production quality; Step 1.4: Using the production batch of the currently produced cables with qualified quality as the primary key, the image fingerprint of the surface image of the insulation layer of the cables with qualified production quality is stored in the qualified cable database in the form of a character string.

3. A cable production quality detection method according to claim 1, characterized in that: The step 3 specifically comprises the following steps: Step 3.1: Perform average pooling on the cable insulation layer images at three different angles taken by three cameras surrounding the cable to generate three compressed images with a resolution of 32×32 pixels; Step 3.2: The three compressed images with a resolution of 32×32 pixels are grayed out by weighted averaging method to generate three compressed gray images with a resolution of 32×32 pixels.

4. A cable production quality detection method according to claim 1, characterized in that: The step 4 specifically comprises the following steps: Step 4.1: Convert the three compressed grayscale images with a resolution of 32×32 pixels into a matrix, denoted as P x , x=1,2,3, used to represent the labels of three compressed grayscale images with a resolution of 32×32 pixels, P x The dimension is 32×32; Step 4.2: Through D x =AP x A T The three transform coefficient matrices with dimensions of 32×32 are calculated and generated in the following way, where D x is a transformation coefficient matrix with a dimension of 32×32, matrix A is a transformation matrix with a dimension of 32×32, , the value of N is the matrix A and the matrix P x The dimension value of is 32 here, i and j are the row and column labels of matrix A respectively; Step 4.3: Take out sub-matrices with a dimension of 8×8 from the upper left corners of the three transform coefficient matrices with a dimension of 32×32 respectively, and generate three low-frequency domain matrices.

5. A cable production quality detection method according to claim 1, characterized in that: The step 6 specifically comprises the following steps: Step 6.1: Obtain the image fingerprint of the current cable from the qualified cable database; Step 6.2: If the length of the image fingerprint is greater than 16, the first 16 bits of the image fingerprint are retained to make the length of the image fingerprint equal to 16. If the length of the image fingerprint is less than 16, random characters are added to the end of the image fingerprint to make the length of the image fingerprint equal to 16. Step 6.3: Compare the image fingerprint with the three strings of length 16 one by one to find the number of characters that are different at the corresponding positions, generate three Hamming distance values, and sum the three Hamming distance values ​​to generate the total Hamming distance; Step 6.4: If the total Hamming distance is greater than the determination threshold, it is determined that the current cable is suspected of having insulation surface defects, wherein the determination threshold is set to 6, and insulation surface defects include surface bumps, cracks or bubbles.

6. A cable production quality detection system, the system comprising: A database module, the database module is used to obtain the surface image of the insulation layer of the cable with qualified production quality, generate the image fingerprint of the surface image of the insulation layer of the cable with qualified production quality, and store it in the qualified cable database; A camera module, wherein after the insulation injection molding of the current type of cable is completed, the camera module takes an image of the cable insulation layer once every s seconds through three camera devices surrounding the cable, wherein the three camera devices surrounding the cable are equally spaced 120° apart, and s is less than the ratio of the cable moving distance per second on the production line to the width of the camera shooting range; A preprocessing module, the preprocessing module is used to compress and grayscale the images of the cable insulation layer at three different angles taken by three cameras surrounding the cable, and generate three compressed grayscale images with a resolution of 32×32 pixels; A transformation module, wherein the transformation module is used to perform discrete cosine Fourier transformation on three compressed grayscale images with a resolution of 32×32 pixels, respectively, to generate three transformation coefficient matrices with a dimension of 32×32, and to extract sub-matrices with a dimension of 8×8 from the upper left corners of the three transformation coefficient matrices with a dimension of 32×32, respectively, to generate three low-frequency domain matrices; A hash module is used to calculate the corresponding matrix average value for each low-frequency domain matrix, and compare each element of the low-frequency domain matrix with the corresponding matrix average value. If it is greater than or equal to the low-frequency domain matrix average value, it is recorded as 1, and if it is less than the low-frequency domain matrix average value, it is recorded as 0, generating three 64-bit binary sequences, and then converting the three 64-bit binary sequences into hexadecimal respectively to generate three strings of length 16; The detection module is used to obtain the image fingerprint of the current type of cable from the qualified cable database, calculate the Hamming distance with three strings of length 16 in turn and sum them up. If the sum is greater than the judgment threshold, it is judged as a suspected surface defect of the insulation layer.

Images