2.5 D imaging and AI combined safety buckle full-appearance visual detection method and system

Abstract

The invention innovatively provides a 2.5 D imaging and AI combined safety buckle full-appearance visual inspection method, which comprises the following steps: an image analysis step: acquiring a safety buckle surface image shot by a visual camera based on a 2.5 D lighting technology as a reference image; an anomaly obtaining step, which is used for obtaining an edge line and an abnormal area in the reference image through semantic segmentation, and obtaining a pre-detection result of the abnormal area through a security buckle detection model; a lamplight shooting control step: adjusting the angle and parameters of irradiation equipment, shooting the surface of the safety buckle through a path planning strategy and adjusting a lighting angle and a shooting angle, and obtaining safety buckle images at different irradiation angles and lighting angles as to-be-analyzed images, the irradiation equipment including a lighting device and a shooting device; and an abnormity detection step: analyzing the abnormal area on the surface of the safety buckle through an abnormal information analysis strategy according to the to-be-analyzed image in different light irradiation directions to obtain a detection result.

Description

Technical Field

[0001] The present invention relates to the technical field of visual inspection, and more specifically to a safety buckle full-appearance visual inspection system combining 2.5D imaging and AI. Background Art

[0002] The safety buckle is a relatively important safety protection device, and its quality is directly related to the safety of use. With the development of machine vision technology, camera-based automatic detection systems are gradually applied to the field of industrial inspection. In the production process of the safety buckle, the safety buckle is formed by splicing at least two sets of panels, and an angle is set between the two sets of panels. In the process of using 2.5D lighting to inspect the surface of the safety buckle, when one set of panels is inspected, there may be a blind spot in the lighting for the other set of panels, which affects the inspection effect of the safety buckle. Summary of the invention

[0003] To achieve the above object, the present invention provides the following technical solutions: A safety buckle full appearance visual inspection method combining 2.5D imaging and AI. An image analysis step is used to obtain a safety buckle surface image captured by a visual camera based on a 2.5D lighting technology as a reference image, and obtain a reference surface on the safety buckle according to the reference image; An abnormality acquisition step, the abnormality acquisition step is used to obtain edge lines and abnormal areas in the reference image through semantic segmentation, and obtain pre-detection results of the abnormal area through a safety buckle detection model, the pre-detection results including the position, size and type of the abnormal area; The lighting shooting control step is used to obtain the pre-detection result, and is provided with an illumination database. According to the pre-detection result, the corresponding illumination sequence is retrieved from the illumination database, and the illumination sequence is preset with an illumination instruction. According to the illumination instruction, the angle and parameters of the illumination device are adjusted, and the lighting angle and the shooting angle are adjusted through the path planning strategy to shoot the surface of the safety buckle, and the safety buckle images under different illumination angles and lighting angles are obtained as the images to be analyzed, and the illumination device includes a lighting device and a shooting device; In the anomaly detection step, the image to be analyzed under different lighting directions is analyzed through an abnormal information analysis strategy to obtain the abnormal area on the surface of the safety buckle and obtain the detection result.

[0004] Preferably, the abnormal information analysis strategy includes: The position determination step is used to convert the coordinates in the image to be analyzed into relative position coordinates, and to partition the reference surface, each partition corresponding to the image to be analyzed under different illumination angles, and obtain the abnormal area, and calculate the coordinates of its center point to determine the partition to which the point belongs; The size calculation step is to calculate the size of the abnormal area by scaling the relative sizes of the safety buckle reference image and the image to be analyzed; In the type identification step, a safety buckle defect type library is established, and the features of the abnormal area are matched with the features in the type library to obtain the defect type.

[0005] Preferably, a depth analysis step is also included, in which the partition to which the abnormal area belongs is used as the detection image, the abnormal area in the detection image is obtained, and a number of edge points are drawn to obtain an edge trajectory line, and any point in the edge trajectory line and the surrounding area is selected as an abnormal item point, and the point outside the edge trajectory line is selected as a non-defect item point, and the depth of the abnormal item point and the non-defect item point is compared to obtain the depth value of the abnormal item point.

[0006] Preferably, when the depth value is compared with a preset value, if the depth value is less than the preset value, it is a scratch, and when the depth value is greater than the preset value, it is output as a pit.

[0007] Preferably, a path planning strategy is also included, at least two groups of reference planes are arranged on the safety buckle, the two groups of reference planes are arranged continuously, the angle and parameters of the irradiation equipment are obtained, the angle of the irradiation equipment is defined as the detection angle, the edge line of the safety buckle and the reference plane are determined, and the lighting angle is adjusted so that the light shoots the reference plane, the inclined shooting plane and the edge line of the safety buckle in sequence.

[0008] Preferably, the light trajectory planning step also includes a trajectory priority step. The trajectory priority step determines that one of the groups of reference planes is the first reference plane. During the shooting process, the first reference plane is given priority, and the inclined shooting plane, another reference plane and the safety buckle edge line are photographed in turn to ensure that the last group of shots are on the safety buckle edge line where the first reference plane is located.

[0009] Preferably, a verification step is also included. During the lighting angle adjustment process, a visual camera is used to obtain an image of the safety buckle reference surface as a comparison image, and an image to be analyzed under the same lighting angle is obtained. Both the image to be analyzed and the comparison image are converted into a standard image to determine whether the features in the image to be analyzed and the comparison image are consistent. If not, an abnormality is detected.

[0010] Preferably, a model building step is also included, in which a safety buckle detection model is built using a deep learning algorithm, and iterative training is performed using historical safety buckle abnormal images and historical abnormal information under different lighting conditions of the safety buckle to complete the construction of the safety buckle detection model.

[0011] A safety buckle full appearance visual inspection system combining 2.5D imaging and AI. An image analysis module is used to obtain the safety buckle surface image captured by the visual camera based on the 2.5D lighting technology as a reference image, and obtain the safety buckle upper reference surface according to the reference image; An abnormality acquisition module, wherein the abnormality acquisition step is used to acquire edge lines and abnormal areas in the reference image through semantic segmentation, and to acquire pre-detection results of the abnormal areas through a safety buckle detection model, wherein the pre-detection results include the position, size and type of the abnormal areas; A lighting shooting control module is used to obtain the pre-detection result and is provided with an illumination database. According to the pre-detection result, a corresponding illumination sequence is retrieved from the illumination database. The illumination sequence is preset with an illumination instruction. According to the illumination instruction, the angle and parameters of the illumination device are adjusted. The lighting angle and the shooting angle are adjusted through a path planning strategy to shoot the surface of the safety buckle, and the safety buckle images under different illumination angles and lighting angles are obtained as images to be analyzed. The illumination device includes a lighting device and a shooting device. The anomaly detection module analyzes the abnormal area on the surface of the safety buckle using the abnormal information analysis strategy of the image to be analyzed under different lighting directions to obtain the detection results.

[0012] The beneficial effects of the present invention are as follows: the surface of the safety buckle is photographed by a 2.5D lighting device, so as to obtain a preliminary detection result of the surface of the safety buckle, the preliminary detection result includes the position, size and type of the abnormal area, and an irradiation instruction is generated according to the preliminary detection result. According to the irradiation instruction, the angle and parameters of the irradiation device are adjusted, the surface of the safety buckle is photographed from multiple angles, the abnormal area is further verified, and the defect information of the surface of the safety buckle is output. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is the overall flow chart of the present invention; Figure 2 is the safety buckle processing image of the present invention; Figure 3 It is a processing diagram of the present invention. DETAILED DESCRIPTION

[0014] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, 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.

[0015] It should be noted that when a component is referred to as being "fixed to" another component, it may be directly on the other component or there may also be a component centered. When a component is considered to be "connected to" another component, it may be directly connected to the other component or there may also be a component centered. When a component is considered to be "set on" another component, it may be directly set on the other component or there may also be a component centered. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0016] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0017] The embodiments of the present invention are further described in detail below with reference to the accompanying drawings: like Figure 1-Figure 3 As shown, the present invention provides a safety buckle full appearance visual inspection method combining 2.5D imaging and AI, including The image analysis step is used to obtain the safety buckle surface image captured by the visual camera based on the 2.5D lighting technology as a reference image, and obtain the reference surface on the safety buckle according to the reference image; the 2.5D lighting technology is used to ensure that the safety buckle surface is evenly and fully illuminated, thereby capturing high-quality images and increasing image contrast, so that tiny defects on the safety buckle surface are more prominent and obvious in the image, thereby significantly improving detection efficiency; The abnormality acquisition step is used to obtain the edge line and abnormal area in the reference image through semantic segmentation, and obtain the pre-detection result of the abnormal area through the safety buckle detection model, and the pre-detection result includes the position, size and type of the abnormal area; the safety buckle detection model is built based on the AI ​​algorithm, and the semantic segmentation algorithm uses the deep defect model to perform pixel-level segmentation on the reference image, distinguish the background, the safety buckle edge line and the abnormal area, and extract multi-scale features through the full convolution network to accurately locate the position and range of the abnormal area, divide the reference image into different areas, including the background, edge line and abnormal area, further extract the precise position and range of the abnormal area from the semantic segmentation result, and use the safety buckle detection model to perform a deeper detection on the abnormal area to obtain the pre-detection result, including the position (center point coordinates), size (area, perimeter) and type (such as scratches, pits, stains, etc.) of the abnormal area; more advanced semantic segmentation algorithms can be used, such as U-Net++. Based on U-Net, it enhances the information flow between features of different scales through dense connections, and can more accurately segment the abnormal area of ​​the safety buckle, especially for defective areas with blurred boundaries, the segmentation effect will be better; The lighting shooting control step is used to obtain the pre-detection result, and an illumination database is set. According to the pre-detection result, the corresponding illumination sequence is retrieved from the illumination database. The illumination sequence is preset with an illumination instruction. According to the illumination instruction, the angle and parameters of the illumination device are adjusted, and the lighting angle and the shooting angle are adjusted through the path planning strategy to shoot the surface of the safety buckle, and the safety buckle images under different illumination angles and lighting angles are obtained as the images to be analyzed. The illumination device includes a lighting device and a shooting device. Through the abnormal step, the pre-detection result of the safety buckle surface is obtained, and multiple illumination sequences are included in the illumination database. Each irradiation sequence is designed for different types of abnormal area detection requirements. The irradiation sequence has preset irradiation instructions, including the angle and parameters (brightness, color temperature) of the irradiation device and the parameters (exposure time, ISO) of the shooting device. According to the pre-detection results, the appropriate irradiation sequence is retrieved from the irradiation database. The type, position and size of the abnormal feature area in the pre-detection results are analyzed. According to the analysis results, the matching irradiation sequence is searched in the irradiation database. If there are multiple abnormal areas, the appropriate irradiation sequence is selected according to the size of the abnormal features. According to the irradiation instructions in the retrieved irradiation sequence, the angle and parameters of the irradiation device and the shooting device are adjusted. The irradiation equipment (such as lighting) is adjusted. The device) is used to adjust the angle and parameters to ensure that the light can be irradiated to the surface of the safety buckle according to the preset irradiation sequence, and the parameters of the shooting device are adjusted, such as exposure time, ISO, etc., to ensure that a clear and accurate image can be captured; the lighting angle and shooting angle are adjusted through the path planning strategy to obtain the safety buckle images under different irradiation angles and lighting angles, and the moving paths of the lighting device and the shooting device are planned according to the instructions in the irradiation sequence and the position of the abnormal area in the pre-detection result, and the path planning is performed to adjust the lighting angle and shooting angle to obtain the required image, and the safety buckle images are taken at different irradiation angles and lighting angles as the images to be analyzed; The path analysis strategy includes the reference plane determination step and the light trajectory planning step. The reference plane determination step is as follows: at least two groups of reference planes are set on the safety buckle, and the two groups of reference planes are set continuously. One of the reference planes is determined as the first reference plane, and the angle and parameters of the irradiation device are obtained, and the angle of the irradiation device is defined as the detection angle; the edge line and reference plane of the safety buckle are determined, and the light angle is adjusted so that the light can sequentially shoot the reference plane and the edge line of the safety buckle, and the light shooting angle where the edge line of the safety buckle is located is the detection angle; according to the reference plane and the inclined shooting plane, the moving path of the light is planned, and the specific positions of the edge line and the reference plane of the safety buckle are authorized to be clarified. By adjusting the angle of the light, the light can sequentially shoot the reference plane, the inclined shooting plane and the edge line of the safety buckle. When the light shoots the edge line of the safety buckle, the light shooting angle where it is located is the previously defined detection angle. This step ensures that the shooting angle is consistent with the preset detection angle, thereby improving the accuracy and reliability of the measurement; it ensures that all the reference planes of the safety buckle are illuminated at the same detection angle around the safety buckle, so as to facilitate the detection of the surface image of the safety buckle and improve the accuracy of the judgment.

[0018] The lighting trajectory planning step also includes a trajectory priority step. In the trajectory priority step, one of the reference planes is determined to be the first reference plane. During the shooting process, the first reference plane is given priority, and the inclined shooting plane, the other reference plane and the edge line of the safety buckle are photographed in turn. It is ensured that the last group of shots is on the edge line of the safety buckle where the first reference plane is located, and that the lighting shooting angle returns to the edge line where the first reference plane is located, so as to facilitate the correction of the lighting and re-shooting.

[0019] In the anomaly detection step, the image to be analyzed under different lighting directions is analyzed through an abnormal information analysis strategy to obtain the abnormal area on the surface of the safety buckle and obtain the detection result.

[0020] Abnormal information analysis strategies include The position determination step is used to convert the coordinates in the image to be analyzed into relative position coordinates, and to partition the reference plane, each partition corresponds to the image to be analyzed under different illumination angles, and obtain the abnormal area, and calculate the coordinates of its center point to determine the partition to which the point belongs; the absolute coordinates in the image to be analyzed are converted into relative position coordinates to ensure that a unified reference coordinate system is established for all images to be analyzed, and partition processing is performed according to the reference image, each partition corresponds to the image to be analyzed under different lighting angles, and in the converted relative position coordinates, the abnormal area is identified and obtained, the center point of the abnormal area is determined, and its coordinates are calculated, and the partition described by the point is determined according to the coordinates of the center point, so as to understand the specific position of the abnormal area in the image; The size calculation step is to perform proportional scaling based on the relative sizes of the safety buckle reference image and the image to be analyzed, and calculate the size of the abnormal area; the safety buckle reference image is used as a reference, and proportional scaling is performed based on the relative sizes of the image to be analyzed and the reference image, and the actual size of the abnormal area is calculated in the image after the scaling is completed; In the type identification step, a safety buckle defect type library is established, and the features of the abnormal area are matched with the features in the type library to obtain the defect type; a database containing various safety buckle defect types is established, and the defect types include scratches, cracks, dents, bulges, etc. Features are extracted from the abnormal area, and the features include shape, color, etc. The extracted features are matched with the features in the defect type library, and the defect type of the defect area is determined based on the matching results; the database is established in a way that the possible defect types of the safety buckle, such as scratches, cracks, dents, bulges, etc., are clearly defined, and at the same time, for each defect type, its key features are determined, including shape, color, texture, size, etc., and safety buckle defect data is obtained from historical records and quality inspection processes to ensure that the collected data covers various defect types. types and instances of different severities to improve the generalization ability of the defect type library. After the collected data is in a unified format, the data is preprocessed, and image features are automatically extracted using deep learning modules such as convolutional neural networks. The database is used to store defect types, features, data samples and other information. The preprocessed data and the extracted features are poured into the database to form a complete defect type database. The feature matching algorithm is used to match the features in the new data with the features in the type database, and the feature matching algorithm considers the similarity of the features and the feature weights, etc. The extracted abnormal area features are input into these classification models for training and prediction. Compared with the traditional feature matching-based method, the deep learning classification model can automatically learn more complex defect feature representations and improve the accuracy of defect type recognition.

[0021] The abnormal information includes defect location, defect size and defect type, and the defect type includes scratches and pits.

[0022] In the depth analysis step, the partition to which the abnormal area belongs is used as the detection image, the abnormal area in the detection image is obtained, and several edge points are drawn to obtain the edge trajectory line, and any point in the edge trajectory line and the surrounding area is selected as the abnormal item point, and the point outside the edge trajectory line is selected as the non-defect item point, and the depth of the abnormal item point and the non-defect item point is compared to obtain the depth value of the abnormal item point; the detection image is extracted from the partition to which the abnormal feature belongs in order to perform a more detailed analysis, according to the abnormal detection step, the partition to which the abnormal area belongs is determined, and the image containing the abnormal area is extracted from the partition as the detection image, the edge points of the abnormal area are identified and drawn to form an edge trajectory line, abnormal item points are selected in the edge trajectory line and the surrounding area, and non-defect item points are selected outside the edge trajectory line, In the edge trajectory line and its internal area, several points are randomly or according to certain rules selected as abnormal points, and several points outside the edge trajectory line are selected as non-defective points. These points should represent the normal state of the safety buckle surface. The depth characteristics of the abnormal area are evaluated by comparing the depth values ​​of the abnormal points with those of the non-defective points. The depth information of each selected point is obtained using 2.5D imaging technology, and the average depth value or maximum depth value of the abnormal points is calculated and compared with the depth value of the non-defective points. The depth estimation algorithm is used to perform a three-dimensional evaluation of the abnormal area, and a depth map is obtained based on 2.5D imaging. The depth information is semantically segmented using networks such as PSPNet. Combined with point cloud data processing, the actual depth of the pit or scratch is calculated, and the severity and type of the defect are quantified.

[0023] When the depth value is compared with the preset value, if the depth value is less than the preset value, it is a scratch. When the depth value is greater than the preset value, the output is a pit. The depth value is compared with the preset value. If the depth value is less than the preset value, it is judged as a scratch (because scratches are usually shallow, and when the depth value is negative, there is a bulge on the surface of the safety buckle, which may be the edge of the scratch). When the depth values ​​are all negative, it is a bulge. If the depth value is greater than the preset value, it is judged as a pit (because pits are usually deeper).

[0024] The method also includes a verification step, in which, during the lighting angle adjustment process, a visual camera is used to obtain an image of a reference surface of the safety buckle as a comparison image, an image to be analyzed under the same lighting angle is obtained, and the image to be analyzed and the comparison image are both converted to a standard image to determine whether the features in the image to be analyzed and the comparison image are consistent, and if they are inconsistent, an abnormality is detected; during the lighting angle adjustment process, an image of a reference surface of the safety buckle is obtained as a comparison image, and an image to be analyzed is obtained under the same lighting angle, and the image to be analyzed and the comparison image are both converted to a unified standard image for comparison and analysis, so as to determine whether the features in the image to be analyzed and the comparison image are consistent. Whether the features are consistent, extract the key features in the image to be analyzed and the comparison image, these features may include edges, textures, shapes, etc., use feature matching algorithms (such as SIFT, SURF, ORB, etc.) to compare the feature points in the image to be analyzed and the comparison image, and judge whether the features in the two images are consistent according to the matching degree of the feature points. If the features in the image to be analyzed and the comparison image are inconsistent, it is detected as an abnormality. If the feature consistency judgment result shows that there are significant differences in the features of the two images (such as feature point mismatch, shape change, etc.), it is judged as an abnormality, outputs the abnormality detection result, and may trigger further detection or processing procedures.

[0025] It also includes a model construction step, using the AI ​​algorithm to train the historical defective images and historical qualified images of the safety buckle in the safety buckle detection model, and presetting the number of iterations, iterative optimization until the number of iterations is reached, and obtaining a preliminary detection model after the training is completed, and obtaining preliminary detection results, and obtaining the safety buckle abnormality type according to the abnormality detection step and judging whether there is an error, if there is an error, modifying the detection error data to continue training until the detection result is correct, otherwise, no modification is made; collecting the historical defective images and historical qualified images of the safety buckle to ensure that they can accurately reflect the status of the safety buckle, annotating the images, and clearly distinguishing between defective and qualified safety buckles; using a convolutional neural network to build a safety buckle detection model, setting the input layer, hidden layer and The output layer and other structures are used to input the labeled image data into the model for training, and the number of iterations is preset. The model parameters are optimized through multiple iterations. In each iteration, the model adjusts its internal parameters according to the input data, and can accurately identify the state of the safety buckle. After completing the preset number of iterations, a preliminary detection model is obtained, and the preliminary detection model is used to predict the new safety buckle image information to obtain the preliminary detection result; according to the preliminary detection result, the abnormal type of the safety buckle is obtained, and it is determined whether the abnormal detection result is incorrect, that is, whether it is consistent with the actual safety buckle state; if the abnormal detection result is found to be incorrect, the detection error data is modified, the modified data is re-added to the training set, the model is continued to be trained, and the above process is repeated until the detection result is correct.

[0026] Another implementation method: During the model building process, a more advanced deep learning architecture can be selected, such as a Transformer-based visual model (such as ViT); when processing image tasks, the self-attention mechanism can be used to better capture long-distance dependencies in the image, which may have a better effect on the recognition of complex defect features on the surface of the safety buckle; during the training process, transfer learning technology is introduced; a model pre-trained on a large-scale image dataset (such as ImageNet) can be used as a basis, and then fine-tuned for the safety buckle detection task; in this way, the general image features learned by the pre-trained model can be utilized to reduce the number of samples required for training and improve the training efficiency and generalization ability of the model.

[0027] A safety buckle full appearance visual inspection system combining 2.5D imaging and AI. An image analysis module is used to obtain the safety buckle surface image captured by the visual camera based on the 2.5D lighting technology as a reference image, and obtain the safety buckle upper reference surface according to the reference image; An abnormality acquisition module, wherein the abnormality acquisition step is used to acquire edge lines and abnormal areas in the reference image through semantic segmentation, and to acquire pre-detection results of the abnormal areas through a safety buckle detection model, wherein the pre-detection results include the position, size and type of the abnormal areas; A lighting shooting control module is used to obtain the pre-detection result and is provided with an illumination database. According to the pre-detection result, a corresponding illumination sequence is retrieved from the illumination database. The illumination sequence is preset with an illumination instruction. According to the illumination instruction, the angle and parameters of the illumination device are adjusted. The lighting angle and the shooting angle are adjusted through a path planning strategy to shoot the surface of the safety buckle, and the safety buckle images under different illumination angles and lighting angles are obtained as images to be analyzed. The illumination device includes a lighting device and a shooting device. The anomaly detection module analyzes the abnormal area on the surface of the safety buckle using the abnormal information analysis strategy of the image to be analyzed under different lighting directions to obtain the detection results.

[0028] The above are only preferred embodiments of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A safety buckle full appearance visual inspection method combining 2.5D imaging and AI, characterized in that: An image analysis step is used to obtain a safety buckle surface image captured by a visual camera based on a 2.5D lighting technology as a reference image, and obtain a reference surface on the safety buckle according to the reference image; An abnormality acquisition step, the abnormality acquisition step is used to obtain edge lines and abnormal areas in the reference image through semantic segmentation, and obtain pre-detection results of the abnormal area through a safety buckle detection model, the pre-detection results including the position, size and type of the abnormal area; The lighting shooting control step is used to obtain the pre-detection result, and is provided with an illumination database. According to the pre-detection result, the corresponding illumination sequence is retrieved from the illumination database, and the illumination sequence is preset with an illumination instruction. According to the illumination instruction, the angle and parameters of the illumination device are adjusted, and the lighting angle and the shooting angle are adjusted through the path planning strategy to shoot the surface of the safety buckle, and the safety buckle images under different illumination angles and lighting angles are obtained as the images to be analyzed, and the illumination device includes a lighting device and a shooting device; In the anomaly detection step, the image to be analyzed under different lighting directions is analyzed through an abnormal information analysis strategy to obtain the abnormal area on the surface of the safety buckle and obtain the detection result.

2. According to claim 1, a method for visual inspection of the safety buckle appearance by combining 2.5D imaging and AI, characterized in that: The abnormal information analysis strategy includes The position determination step is used to convert the coordinates in the image to be analyzed into relative position coordinates, and to partition the reference surface, each partition corresponding to the image to be analyzed under different illumination angles, and obtain the abnormal area, and calculate the coordinates of its center point to determine the partition to which the point belongs; The size calculation step is to calculate the size of the abnormal area by scaling the relative sizes of the safety buckle reference image and the image to be analyzed; In the type identification step, a safety buckle defect type library is established, and the features of the abnormal area are matched with the features in the type library to obtain the defect type.

3. The method for visual inspection of the safety buckle appearance by combining 2.5D imaging and AI according to claim 2 is characterized in that: The method also includes a depth analysis step, in which the partition to which the abnormal area belongs is used as a detection image, the abnormal area in the detection image is obtained, and a number of edge points are drawn to obtain an edge trajectory line, and any point in the edge trajectory line and the surrounding area is selected as an abnormal item point, and the point outside the edge trajectory line is selected as a non-defect item point, and the depth of the abnormal item point and the non-defect item point is compared to obtain the depth value of the abnormal item point.

4. The method for visually inspecting the overall appearance of a safety buckle by combining 2.5D imaging and AI according to claim 3 is characterized in that: When the depth value is compared with a preset value, if the depth value is less than the preset value, it is a scratch, and when the depth value is greater than the preset value, it is output as a pit.

5. The method for visual inspection of the safety buckle appearance by combining 2.5D imaging and AI according to claim 1 is characterized in that: It also includes a path planning strategy, in which at least two groups of reference planes are arranged on the safety buckle, and the two groups of reference planes are arranged continuously. The angle and parameters of the irradiation equipment are obtained, the angle of the irradiation equipment is defined as the detection angle, the edge line of the safety buckle and the reference plane are determined, and the lighting angle is adjusted so that the light shoots the reference plane, the inclined shooting plane and the edge line of the safety buckle in sequence.

6. The method for visually inspecting the overall appearance of a safety buckle by combining 2.5D imaging and AI according to claim 5, characterized in that: The light trajectory planning step also includes a trajectory priority step. The trajectory priority step determines that one of the groups of reference planes is the first reference plane. During the shooting process, the first reference plane is given priority, and the inclined shooting plane, another reference plane and the safety buckle edge line are photographed in turn to ensure that the last group of shots are on the safety buckle edge line where the first reference plane is located.

7. The method for visually inspecting the overall appearance of a safety buckle by combining 2.5D imaging and AI according to claim 1, characterized in that: The method also includes a verification step. During the lighting angle adjustment process, a visual camera is used to obtain an image of a reference surface of the safety buckle as a comparison image, and an image to be analyzed under the same lighting angle is obtained. Both the image to be analyzed and the comparison image are converted into a standard image to determine whether the features in the image to be analyzed and the comparison image are consistent. If they are inconsistent, an abnormality is detected.

8. The method for visual inspection of the safety buckle appearance by combining 2.5D imaging and AI according to claim 1 is characterized in that: It also includes a model building step, using a deep learning algorithm to build a safety buckle detection model, and iterative training using historical safety buckle abnormal images and historical abnormal information under different lighting conditions to complete the construction of the safety buckle detection model.

9. A safety buckle full appearance visual inspection system combining 2.5D imaging and AI, characterized in that: An image analysis module is used to obtain the safety buckle surface image captured by the visual camera based on the 2.5D lighting technology as a reference image, and obtain the safety buckle upper reference surface according to the reference image; An abnormality acquisition module, the abnormality acquisition module is used to obtain edge lines and abnormal areas in the reference image through semantic segmentation, and obtain pre-detection results of the abnormal area through the safety buckle detection model, the pre-detection results including the position, size and type of the abnormal area; A lighting shooting control module is used to obtain the pre-detection result and is provided with an illumination database. According to the pre-detection result, a corresponding illumination sequence is retrieved from the illumination database. The illumination sequence is preset with an illumination instruction. According to the illumination instruction, the angle and parameters of the illumination device are adjusted. The lighting angle and the shooting angle are adjusted through a path planning strategy to shoot the surface of the safety buckle, and the safety buckle images under different illumination angles and lighting angles are obtained as images to be analyzed. The illumination device includes a lighting device and a shooting device. The anomaly detection module analyzes the abnormal area on the surface of the safety buckle using the abnormal information analysis strategy of the image to be analyzed under different lighting directions to obtain the detection results.

Images