Quality detection system based on industrial big data image analysis

Abstract

The invention discloses a quality detection system based on industrial big data image analysis, and belongs to the technical field of industrial quality detection. A key plane area is acquired through an image acquisition module; acquiring a transverse plain-scan image and a longitudinal plain-scan image corresponding to the key plane area through a set plain-scan model; performing image registration on the transverse plain-scan image and the longitudinal plain-scan image to obtain a standard plain-scan image; performing multi-item defect identification on the standard plain scanning image through a set defect type identification model to obtain a plurality of item defect images; carrying out defect analysis on the plurality of defect images by utilizing an industrial defect analysis module to obtain key defect images corresponding to the standard plain scanning images; performing quality evaluation on the key defect image to obtain a quality abnormal value; finally, available industrial objects and to-be-recycled industrial objects are obtained according to the quality abnormal values. And the accuracy of quality detection is effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial quality inspection, and specifically to a quality inspection system based on industrial big data image analysis. Background Art

[0002] Industry is currently the largest downstream application field of the machine vision industry in China. The sales volume in the industrial field accounts for 81.2%. The industrial industry includes industries such as electronic manufacturing, display panels, automobiles, printing, semiconductors, food and beverage packaging, etc.; therefore, it is necessary to ensure the quality of industrial products sold. Currently, the quality inspection of industrial products is usually carried out through manual inspection or roughly through image inspection. The current image inspection methods include obtaining images through traditional shooting, traditional image processing, threshold analysis detection, etc.; therefore, there are image quality problems in obtaining images through traditional shooting. For example, in a high-speed production line, if ordinary cameras are used to shoot industrial images, it may cause image blurring or loss of details, thereby reducing the accuracy of industrial image quality inspection; again, by setting one or more thresholds, the pixel points in the image are divided into different categories, so as to directly separate the target object from the background to obtain defect features; this will lead to unclear contours, and then lead to loss of details, and then lead to a decrease in the accuracy of industrial image quality inspection; therefore, in order to solve the above technical problems, the present invention provides a quality inspection system based on industrial big data image analysis. Summary of the Invention

[0003] In order to solve the above technical problems, the present invention provides a quality inspection system based on industrial big data image analysis; The object of the present invention can be achieved by the following technical solutions: A quality inspection system based on industrial big data image analysis, the system includes an image acquisition module, an image processing module, an industrial defect analysis module, a quality evaluation module, and a classification module; The image acquisition module is provided with a flat scanning model for obtaining a three-dimensional industrial object, marking key areas of the three-dimensional industrial object to obtain key plane areas; collecting horizontal flat scanning images and vertical flat scanning images corresponding to the key plane areas according to the flat scanning model; The image processing module is provided with an image registration unit and a multiple recognition unit; the image registration unit is used to register the horizontal flat scanning image and the vertical flat scanning image to obtain a standard flat scanning image; the multiple recognition unit is provided with a defect type recognition model for performing multiple defect recognitions on the standard flat scanning image according to the defect type recognition model to obtain multiple defect images; The industrial defect analysis module is used to perform defect analysis on the multiple defect images to obtain key defect images corresponding to the standard flat scanning image; The quality assessment module is used to perform quality assessment on the key defect images to obtain quality outliers; The classification module is used to obtain available industrial objects and industrial objects to be recycled according to the quality outliers.

[0004] Furthermore, the setting process of the flat scan model includes: The flat scan model is provided with an automatic recognition unit, a linear flat scan camera, and a detection channel; the automatic recognition unit is used to recognize the key plane area corresponding to the industrial object to be detected; the linear flat scan camera includes a horizontal linear flat scan camera and a vertical linear flat scan camera, which are respectively used to scan the key plane area to obtain the horizontal flat scan image and the vertical flat scan image corresponding to the key plane area; The detection channel is provided with a transmission interval time and a transmission speed; the transmission spacing is obtained according to the transmission interval time and the transmission speed, and detection areas are set at the positions of several transmission spacings in the detection channel, which are used to collect the industrial object to be detected according to the transmission interval time; the linear flat scan camera is arranged directly above the detection area.

[0005] Furthermore, the process of the image acquisition module acquiring the horizontal flat scan image and the vertical flat scan image corresponding to the key plane area according to the flat scan model includes: Obtain the three-dimensional industrial object image corresponding to the industrial object to be detected through three-dimensional modeling; perform key area marking on the plane area of the three-dimensional industrial object image to obtain the key plane area, and send it to the automatic recognition unit; The industrial object to be detected is sequentially transmitted to the detection area through the detection channel according to the transmission interval time and the transmission speed; the automatic recognition unit sequentially recognizes the key plane areas of the industrial object to be detected corresponding to the detection area, and sequentially flips the key plane areas to directly above, and then sequentially performs flat scans through the linear flat scan camera to obtain the corresponding horizontal flat scan image and vertical flat scan image.

[0006] Furthermore, the process of the image registration unit obtaining the standard flat scan image includes: Obtain the center point, corner points, and edge points corresponding to the horizontal flat scan image and the vertical flat scan image; construct a unit two-dimensional coordinate system, translate the center points of the horizontal flat scan image and the vertical flat scan image corresponding to the key plane area to the origin of the unit two-dimensional coordinate system, and then obtain the coordinate points corresponding to the corner points and edge points, and then completely overlap the coordinate points corresponding to the horizontal flat scan image and the vertical flat scan image to obtain the standard flat scan image corresponding to the key plane area; and mark the corresponding code, denoted as i-k; where i represents the number of the industrial object to be detected; k represents the number of the key plane area; and both i and k take positive integer values.

[0007] Furthermore, the process of the multiple recognition unit setting the defect type recognition model includes: Collect the defect types corresponding to different industrial object materials and the defect characteristics corresponding to the defect types based on big data; map the industrial object materials to the defect types and map the defect types to the defect characteristics to obtain the industrial object defect type chain; Collect a defect type sample image and several defect characteristic sample images corresponding to the industrial object defect type chain; extract the light parameter sets corresponding to each defect characteristic sample image, and integrate and generate a defect characteristic light parameter sample library; perform light processing on the defect type sample image through the defect characteristic light parameter sample library to obtain the corresponding light sample image, and obtain the defect characteristic regions corresponding to the defect characteristics in each light sample image, and then obtain the clarity of the defect characteristics; Mark the light parameters of the light sample image corresponding to the highest clarity as the optimal light parameter set corresponding to the industrial object defect type chain, and map it with the industrial object defect type chain to generate a defect type recognition chain; then connect all the defect type recognition chains to construct a defect type recognition model.

[0008] Further, the process of obtaining multiple defect images includes: Send the industrial object material corresponding to the standard plain scan image of the industrial object to be detected to the defect type recognition model; schedule the defect types corresponding to the industrial object material based on the defect type recognition model, and then schedule the optimal light parameter set corresponding to the defect type recognition chain according to the defect type, and then perform light processing on the standard plain scan image according to the optimal light parameter set to obtain the corresponding defect image, and integrate the defect images corresponding to the standard plain scan image to obtain multiple defect images.

[0009] Further, the process by which the industrial defect analysis module obtains the key defect image corresponding to the standard plain scan image includes: Judge whether the corresponding defect characteristics appear in each defect image of the multiple defect images; if they appear, mark the corresponding defect image as a defect characteristic image; otherwise, do nothing; Extract and integrate the defect characteristics corresponding to the defect characteristic images by translation at the same position to obtain the key defect image corresponding to the key plane region.

[0010] Further, the process by which the quality evaluation module obtains the quality outlier value includes: Obtain the total number of defect types of the key defect image corresponding to the industrial object to be detected according to the defect characteristics; and obtain the areas of each standard plain scan image of the industrial object to be detected, denoted as S i-k; Set the unit defect border for covering the defect features of the key defect image with the minimum number of unit defect borders to obtain the number of unit defect borders corresponding to the defect features; obtain the defect feature area corresponding to each defect feature through the number of unit defect borders, and then obtain the sum of the defect feature areas corresponding to the defect types and mark it as the total defect type area corresponding to the defect type; further obtain the quality anomaly value corresponding to the industrial object to be detected according to the total number of defect types and the total defect type area. That is, the specific formula is: ; M i and Q i respectively represent the quality anomaly value and the total number of defect types corresponding to the i-th industrial object to be detected; S L represents the total defect type area corresponding to the defect type of the L-th category; S i-k represents the area of the standard plain scan image corresponding to the encoding i - k; a and b respectively represent the weight coefficients corresponding to the total number of defect types and the total defect type area, that is, a + b = 1; W L represents the quality anomaly weight coefficient corresponding to the defect type of the L-th category.

[0011] Furthermore, the process by which the classification module obtains the available industrial object and the industrial object to be recycled includes: Set the threshold range (0, M1) of the available quality anomaly value corresponding to the industrial object to be detected; If 0 < M i < M1, then mark the corresponding industrial object to be detected as an available industrial object; If M i ≥ M1, then mark the corresponding industrial object to be detected as an industrial object to be recycled.

[0012] Compared with the prior art, the beneficial effects of the present invention are: 1. The present invention obtains the key plane area through the image acquisition module; acquires the horizontal plain scan image and the vertical plain scan image corresponding to the key plane area through the set plain scan model; performs image registration on the horizontal plain scan image and the vertical plain scan image to obtain the standard plain scan image; the plain scan model can improve the shooting clarity of the industrial object, thereby improving the accuracy of quality detection; 2. The present invention performs multiple defect identifications on the standard plain scan image through the set defect type identification model to obtain multiple defect images; then uses the industrial defect analysis module to perform defect analysis on the multiple defect images to obtain the key defect image corresponding to the standard plain scan image; effectively improves the contour clarity of the defect features, thereby improving the detection ability of the detected defect features, and further improving the accuracy of quality detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the accompanying drawings required for use in the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other accompanying drawings can also be obtained based on these drawings.

[0014] Figure 1 This is the system schematic diagram of the present invention. Detailed implementation manners

[0015] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0016] As Figure 1 shown, a quality inspection system based on industrial big data image analysis, the system includes an image acquisition module, an image processing module, an industrial defect analysis module, a quality evaluation module, and a classification module; The image acquisition module is provided with a flat scan model for obtaining a three-dimensional industrial object, marking key areas of the three-dimensional industrial object to obtain key plane areas; collecting a horizontal flat scan image and a vertical flat scan image corresponding to the key plane areas according to the flat scan model; The image processing module is provided with an image registration unit and multiple recognition units; the image registration unit is used to register the horizontal flat scan image and the vertical flat scan image to obtain a standard flat scan image; the multiple recognition units are provided with a defect type recognition model for performing multiple defect recognitions on the standard flat scan image according to the defect type recognition model to obtain multiple defect images; The industrial defect analysis module is used to analyze the multiple defect images to obtain key defect images corresponding to the standard flat scan image; The quality evaluation module is used to evaluate the quality of the key defect images to obtain quality abnormal values; The classification module is used to obtain available industrial objects and industrial objects to be recycled according to the quality abnormal values.

[0017] It should be further noted that the setting process of the flat scan model includes: The flat scan model is provided with an automatic recognition unit, a linear flat scan camera, and a detection channel. The automatic recognition unit is used to recognize the key plane areas corresponding to the industrial objects to be detected; the linear flat scan camera includes a horizontal linear flat scan camera and a vertical linear flat scan camera, which are respectively used to scan the key plane areas to obtain a horizontal flat scan image and a vertical flat scan image corresponding to the key plane areas; The detection channel is set with a transmission interval time and a transmission speed; the transmission distance is obtained according to the transmission interval time and the transmission speed, and detection areas are set at positions of several transmission distances in the detection channel for collecting the industrial object to be detected according to the transmission interval time; the linear sweeping camera is arranged directly above the detection area; In the above embodiment, it should be further noted that the transmission interval time is used to represent the time corresponding to all plane areas of the industrial object to be detected by the detection area; It should be further noted that the process of the image acquisition module obtaining the three-dimensional industrial object image and marking the key area of the three-dimensional industrial object image to obtain the key plane area includes: Obtaining the three-dimensional industrial object image corresponding to the industrial object to be detected through three-dimensional modeling; marking the key area of the plane area of the three-dimensional industrial object image to obtain the key plane area, and sending it to the automatic recognition unit; In the above embodiment, it should be further noted that the key marking can mark the key plane area as a color area, for example, red, yellow, blue, etc.; the number of corresponding key plane areas includes but is not limited to one or more; It should be further noted that the process of collecting the horizontal sweeping image and the vertical sweeping image corresponding to the key plane area according to the sweeping model includes: The industrial object to be detected is sequentially transmitted to the detection area through the detection channel according to the transmission interval time and the transmission speed; the automatic recognition unit sequentially recognizes the key plane areas of the industrial object to be detected corresponding to the detection area, and sequentially flips the key plane areas to directly above, and then sequentially performs sweeping through the linear sweeping camera to obtain the corresponding horizontal sweeping image and vertical sweeping image; In the above embodiment, it should be further noted that the transmission interval time is set as the time for detecting all plane areas of the industrial object to be detected. The sweeping time of the linear sweeping camera for one plane is about 10 - 30 s. In the case of a complex structure of the industrial object to be detected, the transmission interval time is too long. Here, the maximum value of the key plane area corresponding to the industrial object to be detected with a complex structure can be obtained and marked as the transmission interval time; using the key plane area can not only avoid the detection of unnecessary areas, but also save the time for detecting all of the industrial object to be detected with a complex structure; further, the transmission order and the detection order can be better controlled; after all key plane areas of an industrial object to be detected are completely detected, wait for the next industrial object to be detected to reach the detection area for detection.

[0018] It should be further noted that the process of the image registration unit performing image registration processing on the horizontal sweeping image and the vertical sweeping image to obtain the standard sweeping image includes: Obtain the center points, corner points, and edge points corresponding to the horizontal flat-swept image and the vertical flat-swept image; In the above embodiment, it should be further noted that the corner points are used to represent the intersection points of the edges in the image (such as the corners of an object); the edge points are used to represent the boundary points of the object contour; Construct a unit two-dimensional coordinate system, which is a two-dimensional coordinate system with the origin at (0, 0) and the unit length of the horizontal and vertical coordinates being 1; Translate the center points of the horizontal flat-swept image and the vertical flat-swept image corresponding to the key plane area to the origin of the unit two-dimensional coordinate system, and then obtain the coordinate points corresponding to the corner points and edge points. Then, completely overlap the coordinate points corresponding to the horizontal flat-swept image and the vertical flat-swept image to obtain the standard flat-swept image corresponding to the key plane area; In the above embodiment, it should be further noted that the horizontal flat-swept image and the vertical flat-swept image obtained by horizontal and vertical flat-sweeping with a linear flat-sweeping camera are translated to the unit two-dimensional coordinate system without changing their positions to obtain the coordinate points of the corner points and edge points. Then, completely overlap the corresponding coordinate points to obtain the standard flat-swept image; in this process, not only the incompleteness of the overlap caused by directly overlapping the intersection points and edge points is avoided, but also the incompleteness of the obtained flat-swept image caused by only horizontal or vertical flat-sweeping is avoided; thus, the integrity of the flat-swept image is improved and the loss of details is prevented; further, it may also include multi-angle direction flat-sweeping, etc.; the coordinate points of the corner points and edge points are obtained from the obtained flat-swept image through the unit two-dimensional coordinate system, and then completely overlapped to obtain the standard flat-swept image.

[0019] Number the industrial objects to be detected in the transmission order, denoted as i, where i = 1, 2, 3,..., j, and j takes positive integers; and number the key plane areas in the flat-sweeping order, denoted as k, where k = 1, 2, 3,..., g, and g takes positive integers; then obtain the codes corresponding to each standard flat-swept image, denoted as i - k.

[0020] It should be further noted that the process of setting the defect type recognition model in the multiple recognition unit includes: Based on big data, collect the defect types corresponding to different industrial object materials and the defect characteristics corresponding to the defect types; map the industrial object materials to the defect types, and map the defect types to the defect characteristics to obtain the industrial object defect type chain; Collect a sample image of a defect type corresponding to the industrial object defect type chain and several sample images of defect features; extract the illumination parameter sets corresponding to each sample image of the defect feature, and integrate them to generate a defect feature illumination parameter sample library; perform illumination processing on the sample image of the defect type through the defect feature illumination parameter sample library to obtain the corresponding illumination sample image, and obtain the defect feature regions corresponding to the defect features in each illumination sample image, and further obtain the clarity of the defect features. In the above embodiment, it should be further noted that the curve feature region is extracted from the defect feature region through the binary image, and then the average extraction amplitude corresponding to the feature region is obtained through the gradient algorithm as the clarity of the defect feature corresponding to the defect feature region; this is not specifically described in this embodiment; further, the illumination parameter set includes but is not limited to illumination intensity, angle, wavelength, polarization direction, spectral range, illumination method, etc. Mark the illumination parameters of the illumination sample image corresponding to the highest clarity as the optimal illumination parameter set corresponding to the industrial object defect type chain, and map it with the industrial object defect type chain to generate a defect type recognition chain; then connect all the defect type recognition chains to construct a defect type recognition model.

[0021] In the above embodiment, it should be further noted that the defect types include but are not limited to scratches, cracks, depressions, protrusions, dirt, oil stains, rust, material impurities, etc.; for example, the defect feature corresponding to a scratch is a line.

[0022] It should be further noted that the process of obtaining multiple defect images by performing multiple defect recognitions on the standard flat scan image according to the defect type recognition model includes: Obtain the industrial object material of the industrial object to be detected corresponding to the standard flat scan image and send it to the defect type recognition model; based on the defect type recognition model, schedule the defect types corresponding to the industrial object material, and then schedule the optimal illumination parameter set corresponding to the defect type recognition chain according to the defect type, and then perform illumination processing on the standard flat scan image according to the optimal illumination parameter set to obtain the corresponding defect image, and integrate the defect images corresponding to the standard flat scan image to obtain multiple defect images.

[0023] It should be further noted that the process of the industrial defect analysis module analyzing the defect types of multiple images to obtain the key defect image corresponding to the standard flat scan image includes: Judge whether the defect features corresponding to each defect image of the multiple defect images appear; if they appear, mark the corresponding defect image as a defect feature image; otherwise, do not perform any processing. Translate and extract the defect features corresponding to the defect feature images at the same position and integrate them to obtain the key defect image corresponding to the key plane region.

[0024] It should be further noted that the process of the quality assessment module obtaining the quality outlier by performing quality assessment on the key defect images includes: Obtaining the total number of defect types of the key defect images corresponding to the industrial object to be detected according to the defect features; and obtaining the areas of the respective standard flat scan images of the industrial object to be detected, denoted as S i-k ; Setting a unit defect border for covering the defect features of the key defect images with the fewest unit defect borders to obtain the number of unit defect borders corresponding to the defect features; obtaining the defect feature areas corresponding to the respective defect features through the number of unit defect borders, and further obtaining the sum of the defect feature areas corresponding to the defect types, denoted as the total defect type area corresponding to the defect types; and further obtaining the quality outlier corresponding to the industrial object to be detected according to the total number of defect types and the total defect type area; That is, the specific formula is: ; where M i and Q i respectively represent the quality outlier corresponding to the i-th industrial object to be detected and the total number of defect types; S L represents the total defect type area corresponding to the defect type of the L-th category; S i-k represents the area of the standard flat scan image encoded as i-k; a and b respectively represent the weight coefficients corresponding to the total number of defect types and the total defect type area, that is, a + b = 1; W L represents the quality outlier weight coefficient corresponding to the defect type of the L-th category.

[0025] In the above embodiment, it should be further noted that the unit defect border represents a square of m×m; m of the unit defect border can be set according to the size of the industrial object to be detected; the defect feature area of the defect feature can be accurately obtained by using the area of the unit defect border; this process reflects the diversity of defect types; the more defect types, the more complex the quality risk, and the accuracy of the quality outlier is improved.

[0026] It should be further noted that the process of the classification module classifying the industrial object to be detected according to the quality outlier to obtain the available industrial object and the industrial object to be recycled includes: Setting the available quality outlier threshold range (0, M1) corresponding to the industrial object to be detected; If 0 < M i < M1, then mark the corresponding industrial object to be detected as an available industrial object; If M i ≥ M1, then mark the corresponding industrial object to be detected as an industrial object to be recycled.

[0027] In the above embodiments, it should be further noted that the quality outlier cannot be equal to 0 because the total number and total area of the defect types included in the key defect image are not 0.

[0028] The features and exemplary embodiments of various aspects of the present application will be described in detail below. To make the objectives, technical solutions, and advantages of the present application clearer, the present application will be further described in detail below in combination with the accompanying drawings and specific embodiments; it should be understood that the specific embodiments described herein are only intended to explain the present application and not to limit the present application; for those skilled in the art, the present application can be implemented without some of these specific details; the above description of the embodiments is only to provide a better understanding of the present application by showing examples of the present application.

[0029] The above embodiments are only used to illustrate rather than limit the technical method of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A quality inspection system based on industrial big data image analysis, characterized in that: The system includes an image acquisition module, an image processing module, an industrial defect analysis module, a quality assessment module and a classification module; The image acquisition module is provided with a flat scan model, which is used to acquire a three-dimensional industrial object, mark key areas of the three-dimensional industrial object to acquire key plane areas; and acquire a horizontal flat scan image and a vertical flat scan image corresponding to the key plane area according to the flat scan model; The image processing module is provided with an image registration unit and a multi-item recognition unit; the image registration unit is used to perform image registration on the horizontal plain scan image and the vertical plain scan image to obtain a standard plain scan image; the multi-item recognition unit is provided with a defect type recognition model, which is used to perform multi-item defect recognition on the standard plain scan image according to the defect type recognition model to obtain a multi-item defect image; The industrial defect analysis module is used to perform defect analysis on multiple defect images to obtain key defect images corresponding to the standard plain scan images; The quality assessment module is used to perform quality assessment on key defect images to obtain quality abnormality values; The classification module is used to obtain usable industrial objects and industrial objects to be recycled according to quality abnormality values.

2. A quality inspection system based on industrial big data image analysis according to claim 1, characterized in that: The setting process of the flat scan model includes: The flat-scan model is provided with an automatic recognition unit, a linear flat-scan camera and a detection channel; the automatic recognition unit is used to recognize the key plane area corresponding to the industrial object to be detected; the linear flat-scan camera includes a transverse linear flat-scan camera and a longitudinal linear flat-scan camera, which are respectively used to scan the key plane area to obtain a transverse flat-scan image and a longitudinal flat-scan image corresponding to the key plane area; The detection channel is provided with a transmission interval time and a transmission speed; the transmission spacing is obtained according to the transmission interval time and the transmission speed, and a detection area is set at the position of several transmission spacings in the detection channel for collecting industrial objects to be detected according to the transmission interval time; the linear flat scan camera is set directly above the detection area.

3. A quality inspection system based on industrial big data image analysis according to claim 2, characterized in that: The process of the image acquisition module acquiring the horizontal plain scan image and the vertical plain scan image corresponding to the key plane area according to the plain scan model includes: Acquire a three-dimensional industrial object image corresponding to the industrial object to be detected through three-dimensional modeling; perform key area marking on the plane area of ​​the three-dimensional industrial object image to obtain the key plane area, and send it to the automatic recognition unit; The industrial objects to be inspected are transmitted to the inspection area through the inspection channel in sequence according to the transmission interval time and transmission speed; the automatic recognition unit identifies the key plane areas corresponding to the industrial objects to be inspected in the inspection area in sequence, and flips the key plane areas to the top in sequence, and then uses the linear flat-scan camera to perform flat-scanning in sequence to obtain the corresponding horizontal flat-scanning images and vertical flat-scanning images.

4. A quality inspection system based on industrial big data image analysis according to claim 3, characterized in that: The process of the image registration unit acquiring the standard plain scan image includes: Obtain the center points, corner points and edge points corresponding to the horizontal and vertical flat scan images; construct a unit two-dimensional coordinate system, translate the center points of the horizontal and vertical flat scan images corresponding to the key plane area to the origin of the unit two-dimensional coordinate system, and then obtain the coordinate points corresponding to the corner points and edge points, and then completely overlap the coordinate points corresponding to the horizontal and vertical flat scan images to obtain the standard flat scan image corresponding to the key plane area.

5. The quality inspection system based on industrial big data image analysis according to claim 1 is characterized in that: The process of setting the defect type recognition model by the multiple recognition units includes: Based on big data, the defect types corresponding to different industrial object materials and the defect features corresponding to the defect types are collected; the industrial object materials are mapped with the defect types, and the defect types are mapped with the defect features to obtain the industrial object defect type chain; Collect a defect type sample image and several defect feature sample images corresponding to the defect type chain of the industrial object; extract the illumination parameter set corresponding to each defect feature sample image, and integrate them to generate a defect feature illumination parameter sample library; perform illumination processing on the defect type sample image through the defect feature illumination parameter sample library to obtain the corresponding illumination sample image, and obtain the defect feature area corresponding to the defect feature in each illumination sample image, and then obtain the clarity of the defect feature; The illumination parameters of the highest definition corresponding illumination sample image are marked as the optimal illumination parameter set corresponding to the industrial object defect type chain, and mapped with the industrial object defect type chain to generate a defect type recognition chain; then all defect type recognition chains are connected to construct a defect type recognition model.

6. A quality inspection system based on industrial big data image analysis according to claim 5, characterized in that: The process of acquiring multiple defect images includes: The industrial object material of the industrial object to be inspected corresponding to the standard flat scan image is obtained and sent to the defect type recognition model; the defect type corresponding to the industrial object material is scheduled based on the defect type recognition model, and then the optimal lighting parameter set corresponding to the defect type recognition chain is scheduled according to the defect type, and then the standard flat scan image is illuminated according to the optimal lighting parameter set to obtain the corresponding defect image, and the defect images corresponding to the standard flat scan image are integrated to obtain multiple defect images.

7. A quality inspection system based on industrial big data image analysis according to claim 6, characterized in that: The process of the industrial defect analysis module acquiring the key defect image corresponding to the standard plain scan image includes: Determine whether each defect image of the plurality of defect images has a corresponding defect feature; if so, mark the corresponding defect image as a defect feature image; otherwise, do not perform any processing; The defect features corresponding to the defect feature image are extracted by translation at the same position and integrated to obtain the key defect image corresponding to the key plane area.

8. The quality inspection system based on industrial big data image analysis according to claim 7 is characterized in that: The process of obtaining the quality abnormal value by the quality assessment module includes: The total number of defect types corresponding to the key defect image of the industrial object to be inspected is obtained according to the defect characteristics; and the area of ​​each standard flat scan image of the industrial object to be inspected is obtained; a unit defect border is set to cover the defect characteristics of the key defect image with a minimum unit defect border to obtain the number of unit defect borders corresponding to the defect characteristics; the defect feature area corresponding to each defect feature is obtained through the number of unit defect borders, and then the sum of the defect feature areas corresponding to the defect type is obtained and marked as the total area of ​​the defect type corresponding to the defect type; and then the quality anomaly value corresponding to the industrial object to be inspected is obtained according to the total number of defect types and the total area of ​​defect types.

9. A quality inspection system based on industrial big data image analysis according to claim 8, characterized in that: The process of the classification module acquiring usable industrial objects and industrial objects to be recycled includes: Set the available quality anomaly threshold range (0, M1) corresponding to the industrial object to be detected; If the quality anomaly value is greater than 0 and less than M1, the corresponding industrial object to be detected is marked as an available industrial object; If the quality abnormality value is greater than or equal to M1, the corresponding industrial object to be detected is marked as an industrial object to be recycled.

Images