CN117761060A - A visual inspection system and its inspection method - Google Patents

Abstract

The invention discloses a visual detection system and a detection method thereof, which include placing a base plate. A placement frame is provided at the top of the placement base plate. A linear movement module for adjusting the vertical shooting position is vertically provided on the front of the placement frame. The linear movement module There is a positioning component on one side of the group, and a shooting component for visual inspection is set on the front of the linear movement module. The shooting component includes a camera, a shading mechanism and a detection and analysis system. The detection and analysis system includes an image input unit, a defect positioning unit, and algorithm processing. unit, processing decision unit and output unit. After the shooting component of the present invention has completed locating the defects, it can analyze the defects through the defect analysis of the filtering module and the dividing module. In this way, after the defect analysis, the defects can be judged more accurately, and therefore the defects will not be misidentified. judgment, making the detection and judgment of materials more accurate and further reducing the loss of material detection and judgment.

Description

A visual inspection system and its inspection method

Technical field

The present invention relates to the field of visual detection systems, and in particular to a visual detection system and a detection method thereof.

Background technique

When using receiving equipment to collect materials, you first need to use a visual inspection system to detect the materials. After the inspection is completed, the materials are packaged and collected. However, with the traditional visual inspection system, when performing visual inspection, Some optical adjustments can be made to improve the clarity of the photographed image, and then find defects in the material. Although this detection method can clearly locate the defects in the material, it does not analyze and classify the types of defects. In actual practice, During the production process, some material defects can be collected and packaged without affecting the use of the product. Therefore, the traditional visual inspection system does not conduct defect classification analysis, which leads to misjudgment of some qualified products. In this way This resulted in the loss of some materials.

Contents of the invention

The purpose of the present invention is to provide a visual detection system and a detection method thereof to solve the problems raised in the above background technology.

In order to achieve the above object, the present invention provides the following technical solution: a visual inspection system, including a placement base plate, a placement frame is provided at the top of the placement base plate, and a placement frame is vertically provided on the front of the placement frame for adjusting the vertical shooting position. A linear movement module. One side of the linear movement module is provided with a positioning component for positioning the moving position. The front of the linear movement module is provided with a shooting component for visual inspection. The shooting component includes a camera and a shading mechanism. And a detection and analysis system for defect analysis, the detection and analysis system includes an image input unit, a defect positioning unit, an algorithm processing unit, a processing judgment unit and an output unit.

Preferably, the placement frame includes an N-shaped frame. The bottoms of both arms of the N-shaped frame are fixedly connected with mounting plates. The mounting plates are fixedly connected to the top of the placement bottom plate. The front side of the N-shaped frame is provided with a connection. The sliding frame, the positioning component includes a positioning insert provided on one side of the connecting sliding frame.

Preferably, the connecting carriage is slidably sleeved in the middle of the linear movement module. The top and bottom of one side of the linear movement module are respectively provided with light sensors used to cooperate with the positioning inserts to achieve positioning. The top of the moving module is provided with a driving motor for driving the linear moving module to move.

Preferably, the front side of the linear movement module is provided with a connection mechanism for connecting the shooting assembly. The connection mechanism includes a connection plate fixedly connected to the front side of the linear movement module. A connection block is provided on the front side of the middle part of the connection plate. One end of the connecting block away from the connecting plate is fixedly connected to the back of the camera. The front of the bottom of the connecting plate is provided with a connecting frame for connecting the light shielding mechanism.

Preferably, the light shielding mechanism includes a light shielding frame fixedly connected to the bottom end of the connecting frame, a plurality of light shielding plates are provided in the middle of the light shielding frame, and a light shielding plate is provided in the middle of the light shielding frame and directly below the camera to limit the photographing field of view. The field of view box for the range.

Preferably, a lens group for adjusting the focal length of the camera is provided between the bottom end of the camera and the top of the field of view frame.

Preferably, the image input unit includes an image input module and a device positioning module, used for image transmission and positioning of the detection area;

The defect positioning unit includes a grayscale comparison module and a defect classification module, and the defect classification module includes an instance segmentation part and a defect classification part for determining and classifying defects;

The algorithm processing unit includes a post-processing algorithm module for first optimization processing of defects;

The processing and determination unit includes a filtering module and a dividing module, used for classifying defects;

The output unit includes a defect classification output module, a defect size output module and a defect coordinate position output module, and is used for outputting defects.

Preferably, the device positioning module and the gray-scale comparison module perform positioning of the device detection area and gray-scale comparison of positioning defects based on traditional vision, and the defect classification module performs instance segmentation and defect classification based on deep learning.

Preferably, the processing and decision unit is based on a rule decision algorithm.

A detection method of a visual detection system, using the visual detection system, the detection method includes the following steps;

The first step is to transport the material to be inspected directly below the field of view frame. At this time, the drive motor starts and controls the linear movement module to move in the vertical direction to find the shooting position of the material. At the same time, the linear movement module moves in the vertical direction. During the group movement, the camera is driven to move synchronously, and the photo location is positioned through the cooperation between the positioning insert and the light sensor;

In the second step, after the position is adjusted, start the camera, adjust the lens group, and confirm the shooting focal length before shooting the material to be inspected;

In the third step, the detection and analysis system performs detection and analysis of the photos taken under the signal transmission of the image input unit. If there is a defect in the image, the detection area is first positioned through the device positioning module, and then the defect positioning unit is used to perform grayscale determination of the defect. After comparison, deep learning is then used to segment and classify defect instances, and then the algorithm processing unit and processing decision unit are used to further calculate and divide the defects to obtain the classification of the defect, the size of the defect, and the coordinate position of the defect;

In the fourth step, the obtained defect classification, defect size and defect coordinate position are output through the output unit. At this time, the visual inspection of a group of materials is completed.

Technical effects and advantages of the present invention:

The shooting component of the present invention includes a camera, a shading mechanism and a detection and analysis system for defect analysis, and an algorithm processing unit and a processing and determination unit are provided at the detection and analysis system. After the defects are positioned, the filter module and the dividing module Defect analysis allows defects to be analyzed, so that after defect analysis, defects can be judged more accurately, so it will not cause misjudgments of defects, making the detection and judgment of materials more accurate, and further reducing the errors in material detection and judgment. loss.

Description of drawings

Figure 1 is a schematic structural diagram of the visual inspection system of the present invention.

Figure 2 is a schematic diagram of the structural connection between the linear movement module and the shooting assembly of the present invention.

Figure 3 is a schematic structural diagram of the linear movement module of the present invention.

Figure 4 is a structural side view of the linear movement module of the present invention.

Figure 5 is a structural front view of the linear movement module of the present invention.

FIG6 is a calculation flow chart of the rule determination algorithm of the present invention.

In the picture: 1. Place the bottom plate; 2. N-shaped frame; 201. Mounting plate; 3. Linear movement module; 301. Drive motor; 302. Connecting slide; 303. Connecting plate; 304. Connecting block; 305. Connection 306. Light sensor; 307. Positioning insert; 4. Camera; 401. Lens set; 5. Shade frame; 501. Shade plate; 502. Field of view frame.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some 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 fall within the scope of protection of the present invention.

Embodiment 1, the present invention provides a visual inspection system as shown in Figures 1-6, including a placement base plate 1. A placement frame is provided at the top of the placement base plate 1. The front of the placement frame is vertically provided with a vertical camera for adjusting vertical shooting. Position linear movement module 3. One side of the linear movement module 3 is provided with a positioning component for mobile position positioning. The front of the linear movement module 3 is provided with a shooting component for visual inspection. The shooting component includes a camera 4 and a shading mechanism. and a detection and analysis system for defect analysis. The detection and analysis system includes an image input unit, a defect positioning unit, an algorithm processing unit, a processing judgment unit and an output unit.

Specifically, the placement frame includes an N-shaped frame 2. The bottoms of both arms of the N-shaped frame 2 are fixedly connected with mounting plates 201. The mounting plates 201 are fixedly connected to the top of the placing bottom plate 1. A connecting slide is provided on the front of the N-shaped frame 2. 302. The positioning component includes a positioning insert 307 provided on one side of the connecting carriage 302.

It should be noted that the N-shaped frame 2, as the entire hypothetical mechanism, provides an erection height for the shooting component. Through the setting of this structure, the shooting component can be placed in a suspended state, so that the shooting can be adjusted when taking pictures. The setting height of the components makes it possible to more flexibly adjust the shooting height when taking pictures, making it easier to find the shooting teaching aids, and also allows materials of different sizes to be focused and photographed.

Further, the connecting carriage 302 is slidably sleeved in the middle of the linear movement module 3. The top and bottom of one side of the linear movement module 3 are respectively provided with light sensors 306 used to cooperate with the positioning insert 307 to achieve positioning. The linear movement The top of the module 3 is provided with a driving motor 301 for driving the linear motion module 3 to move.

It should be noted that the linear movement module 3 used in this solution is a screw-shaped linear module, which cooperates with the connecting carriage 302 to enable the linear movement module to move linearly under the support of the connecting carriage 302 After 3 is started, the linear movement module 3 can move in the vertical direction according to the rotation control direction of the drive motor 301, so that driven by the movement of the linear movement module 3, the shooting assembly can adjust the vertical setting.

The light sensor 306 cooperates with the positioning insert 307 for position sensing and positioning. When the positioning insert 307 passes through the light sensor 306 at the corresponding position, the position can be learned, and the position of the positioning insert 307 follows the connection carriage. 302 does not move, and the light sensor 306 follows the linear movement module 3 to move in the vertical direction, so that it can reach the position where the linear movement module 3 moves vertically.

Specifically, the front of the linear movement module 3 is provided with a connection mechanism for connecting the shooting components. The connection mechanism includes a connection plate 303 fixedly connected to the front of the linear movement module 3. A connection block 304 is provided on the front of the middle part of the connection plate 303. One end of the block 304 away from the connecting plate 303 is fixedly connected to the back of the camera 4. The front of the bottom of the connecting plate 303 is provided with a connecting frame 305 for connecting the light shielding mechanism.

It should be noted that the connecting plate 303 is fixedly connected to the bottom of the front of the linear motion module 3, so that the connecting plate 303 moves synchronously with the linear motion module 3, and the connecting block 304 is fixedly connected to the middle of the front of the connecting plate 303, so that through The connection block 304 connects and fixes the camera 4 so that the camera 4 can move synchronously with the linear movement module 3 .

Further, the light shielding mechanism includes a light shielding frame 5 fixedly connected to the bottom end of the connecting frame 305. A plurality of light shielding plates 501 are provided in the middle of the light shielding frame 5. A light shielding plate 501 is provided in the middle of the light shielding frame 5 and directly below the camera 4 to limit the photographing field of view. Range of view range box 502.

It should be noted that in this solution, the light shielding frame 5 is fixed through the connecting frame 305, so that the light shielding frame 5 can follow the linear movement module 3 for synchronous adjustment and movement. During the movement of the light shielding frame 5, the view range frame 502 is driven. Move synchronously to ensure that after the material to be detected moves to just below the field of view frame 502, the material to be detected can be better covered by moving the field of view frame 502, so that the camera 4 can shoot more accurately. This can reduce shooting interference from foreign objects and make subsequent image analysis more accurate.

Furthermore, a lens group 401 for adjusting the focal length of the camera 4 is provided between the bottom end of the camera 4 and the top end of the field of view frame 502 .

It should be noted that the lens group 401 used in this solution is the MVL-KF0814M-12MPE industrial lens, which has a large aperture of F1.4, 1.1", and 12 million resolution. It is a C-mount industrial lens with high resolution and consistent picture clarity. It has the advantages of high sensitivity, low distortion, excellent optical performance under different object distances, and ultra-high image peripheral brightness ratio. Through the setting of the lens group 401, the camera 4 can take more clear and accurate photos of boring photos to be detected.

Specifically, the image input unit includes an image input module and a device positioning module, which are used for image transmission and positioning of the detection area;

It should be noted that a communication interface is provided on the top of the camera 4. After the communication interface is electrically connected to the industrial computer, the image input unit can be used to transmit the image information captured by the camera 4 to the industrial computer for detection. When performing image detection and analysis, the image detection area is positioned through the device positioning module. The positioning position is set in the device positioning module in advance. After the setting is completed, the positioning can be accurately To the area to be detected, it can quickly locate the detection area, quickly eliminate interference, and further improve the accuracy of detection.

The defect positioning unit includes a grayscale comparison module and a defect classification module, and the defect classification module includes an instance segmentation part and a defect classification part, which are used to determine and classify defects;

It should be noted that the defect classification module performs deep learning to implement instance segmentation and defect classification. In this way, defect types can be separated, making the defect type analysis clearer, so that during subsequent defect analysis, it can The form and type of defects can be learned more accurately, making the judgment of material qualification more accurate.

Furthermore, the device positioning module and the gray-scale comparison module perform positioning of the device detection area and gray-scale comparison of defects based on traditional vision, and the defect classification module performs instance segmentation and defect classification based on deep learning.

The algorithm processing unit includes a post-processing algorithm module for the first optimization processing of defects;

It should be noted that post-processing algorithms refer to algorithms that further process and optimize results in the final stage of data processing. It is often used to improve the quality of data, improve the performance of a model, or increase the accuracy of predictions. Post-processing algorithms can be applied to various fields, such as image processing, natural language processing, machine learning, etc. The specific implementation of post-processing algorithms depends on the characteristics and requirements of the task. It can include a series of steps such as denoising, smoothing, normalization, filtering, clustering, classification, etc. These steps are designed to fix potential errors, reduce noise, adjust predictions, or provide better interpretability. In image processing, post-processing algorithms can be used to remove artifacts from images, enhance details, adjust the brightness and contrast of images, etc. In natural language processing, post-processing algorithms can be used to correct grammatical errors, deal with ambiguous word meanings, and improve the fluency of translation. In general, post-processing algorithms play a key role in the final stage of data processing. It can help us improve the performance of data and models, and improve our understanding and utilization of data.

The processing and judgment unit includes a filtering module and a dividing module, which are used to classify defects;

Further, the processing and determination unit is based on a rule determination algorithm. First, the detection part is divided into multi-area blocks through the filtering module for screening and filtering of qualified areas. If it is a qualified area, the output is filtered. If not, it is divided by dots and lines, respectively by length. Rules, area rules and number rules are used to classify defects, determine the size of defects and determine the coordinate position of defects.

It should be noted that the length rule is based on the length corresponding to the defective product as the rule standard, and the area rule is based on the defective area corresponding to the defective product as the rule standard. The length rule and area rule are used to determine the size and defect of the defect. Classification, and the number rule is based on the divided detection area, and the number of defective areas determined to be corresponding to the positioning of defect coordinates.

The output unit includes a defect classification output module, a defect size output module and a defect coordinate position output module, which are used for defect output.

Embodiment 2: The present invention provides a detection method of a visual detection system, which adopts a visual detection system. The detection method includes the following steps;

In the first step, the material to be detected is first transported to just below the field of view frame 502. At this time, the drive motor 301 is started and controls the linear movement module 3 to move in the vertical direction to find the shooting position of the material. During the movement of the linear movement module 3, the camera 4 is driven to move synchronously, and the photographing position is positioned through the cooperation between the positioning insert 307 and the light sensor 306;

It should be noted that this step is the preparation work before taking pictures. After adjustment through this step, the positioning of the material to be detected can be more accurately achieved.

In the second step, after the position is adjusted, start the camera 4, adjust the lens group 401, confirm the shooting focal length, and then shoot the material to be inspected;

It should be noted that this step is to obtain the image information of the material to be detected through the camera 4, and then analyze the image after obtaining the image information.

In the third step, the detection and analysis system performs detection and analysis of the photos taken under the signal transmission of the image input unit. If there is a defect in the image, the detection area is first positioned through the device positioning module, and then the defect positioning unit is used to perform grayscale determination of the defect. After comparison, deep learning is then used to segment and classify defect instances, and then the algorithm processing unit and processing decision unit are used to further calculate and divide the defects to obtain the classification of the defect, the size of the defect, and the coordinate position of the defect;

In the fourth step, the obtained defect classification, defect size and defect coordinate position are output through the output unit. At this time, the visual inspection of a group of materials is completed.

It should be noted that by outputting the defects corresponding to the classification number, conducting analysis and comparison, and screening out unqualified products, such classification and screening can make the detection and screening more accurate.

Finally, it should be noted that the above are only preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, it is still The technical solutions described in the foregoing embodiments may be modified, or equivalent substitutions may be made to some of the technical features. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included in within the protection scope of the present invention.

Claims (10)

1. The utility model provides a visual inspection system, includes places bottom plate (1), its characterized in that, the top of placing bottom plate (1) is provided with the rack, the front of rack is vertical to be provided with and is used for adjusting vertical rectilinear movement module (3) of shooting position, one side of rectilinear movement module (3) is provided with the locating component that is used for removing the position location, the front of rectilinear movement module (3) is provided with visual inspection's shooting subassembly, shooting subassembly includes camera (4), shading mechanism and is used for defect analysis's detection analysis system, detection analysis system includes image input unit, defect location unit, algorithm processing unit, processing determination unit and output unit.

2. A visual inspection system according to claim 1, wherein the placement frame comprises an N-shaped frame (2), the bottoms of the two arms of the N-shaped frame (2) are fixedly connected with a mounting plate (201), the mounting plate (201) is fixedly connected to the top end of the placement base plate (1), the front surface of the N-shaped frame (2) is provided with a connecting carriage (302), and the positioning assembly comprises a positioning insert (307) arranged on one side of the connecting carriage (302).

3. The visual inspection system according to claim 2, wherein the connecting carriage (302) is slidably sleeved in the middle of the linear movement module (3), light sensors (306) for matching with the positioning insertion sheet (307) to realize positioning are respectively arranged at the top and the bottom of one side of the linear movement module (3), and a driving motor (301) for driving the linear movement module (3) to move is arranged at the top of the linear movement module (3).

4. The visual inspection system according to claim 1, wherein the front surface of the linear movement module (3) is provided with a connecting mechanism for connecting the shooting assembly, the connecting mechanism comprises a connecting plate (303) fixedly connected to the front surface of the linear movement module (3), the front surface of the middle part of the connecting plate (303) is provided with a connecting block (304), one end, far away from the connecting plate (303), of the connecting block (304) is fixedly connected with the back surface of the shooting camera (4), and the front surface of the bottom of the connecting plate (303) is provided with a connecting frame (305) for connecting the shading mechanism.

5. The visual inspection system according to claim 4, wherein the light shielding mechanism comprises a light shielding frame (5) fixedly connected to the bottom end of the connecting frame (305), a plurality of light shielding plates (501) are arranged in the middle of the light shielding frame (5), and a view field range frame (502) for limiting a photographing view field range is arranged in the middle of the light shielding frame (5) and under the photographing camera (4).

6. A visual inspection system according to claim 5, characterized in that a lens group (401) for adjusting the focal length of the beat camera (4) is arranged between the bottom end of the shooting camera (4) and the top end of the field of view frame (502).

7. A visual inspection system according to claim 1, wherein the image input unit comprises an image input module and a device positioning module for transmission of images and positioning of inspection areas;

the defect positioning unit comprises a gray scale comparison module and a defect classification module, and the defect classification module comprises an example division part and a defect classification part which are used for judging and classifying defects;

the algorithm processing unit comprises a post-processing algorithm module and is used for performing first optimization processing on the defects;

the processing judging unit comprises a filtering module and a dividing module, and is used for dividing defects;

the output unit comprises a defect classification output module, a defect size output module and a defect coordinate position output module, and is used for outputting defects.

8. The vision inspection system of claim 7, wherein the device localization module and the gray scale comparison module perform localization of device inspection areas and gray scale comparison of localized defects based on conventional vision, and the defect classification module performs instance segmentation and defect classification based on deep learning.

9. A visual inspection system as set forth in claim 7, wherein said process determination unit is based on a rule determination algorithm.

10. A detection method of a visual detection system, characterized in that a visual detection system according to any one of claims 1-9 is used, said detection method comprising the following steps;

firstly, conveying a material to be detected to the position right below a visual field range frame (502), starting a driving motor (301) at the moment, controlling a linear moving module (3) to move in the vertical direction, finding the shooting position of the material, simultaneously driving a shooting camera (4) to synchronously move in the moving process of the linear moving module (3), and positioning the shooting position through the cooperation between a positioning inserting sheet (307) and a light sensor (306);

secondly, after the position is adjusted, starting a camera (4), adjusting a lens group (401), and after confirming the shooting focal length, shooting the material to be detected;

thirdly, the detection analysis system performs detection analysis of the shot photo under the signal transmission of the image input unit, if the image has defects, the device positioning module is used for positioning a detection area, then the defect positioning unit is used for gray scale comparison of the defects, then the deep learning is used for example segmentation and classification of the defects, and then the algorithm processing unit and the processing judgment unit are used for further performing calculation and division of the defects in sequence to obtain classification of the defects, the sizes of the defects and the coordinate positions of the defects;

and fourthly, outputting the obtained defect classification, defect size and defect coordinate position through an output unit, and completing visual detection of a group of materials.