Intelligent portable 3D anti-overexposure part defect detection device and method

Abstract

The invention relates to an intelligent portable 3D anti-overexposure part defect detection device and method, and belongs to the field of industrial visual inspection. According to the technical scheme, the device is composed of a protection box, a detection table telescopic system, a projection-imaging system, a workpiece rotating system, a main control system and an indicator lamp module. The protection box comprises a protection box cover and a protection box body; the detection table telescopic system comprises a guide rail frame, a detection table, a plane mirror mounting groove, a handle and a limiting module; the projection-imaging system comprises an industrial projector, a camera and a plane mirror; the workpiece rotating system comprises a workpiece rotating disc and a rotating disc driving module; the main control system comprises a control module, a reset button and an upper computer; the indicating lamp module comprises a standby indicating lamp and a working indicating lamp. The device is intelligent and high in integration level, the problem of easy overexposure of fringe projection three-dimensional measurement is avoided through the double-light-path design, and the hardware cost is reduced.

Description

Technical Field

[0001] The present invention relates to a device for detecting part defects, which utilizes a dual-light path system combined with monocular structured light technology to achieve three-dimensional reconstruction of highly reflective parts, and is used in the field of industrial visual inspection. Background Art

[0002] With the rapid development of computer vision, optical sensing, and signal processing technologies, 3D reconstruction has been widely applied in various fields, such as industrial inspection, medical imaging, virtual reality, augmented reality, robotic navigation, and cultural heritage preservation. Currently, 3D reconstruction methods based on structured light projection have attracted widespread attention due to their high precision, strong real-time performance, low cost, and easy integration and implementation.

[0003] The industrial sector requires high-precision and high-efficiency part defect detection. Traditional detection methods such as manual inspection and contact inspection have problems such as low detection efficiency, high cost, and easy wear of workpieces. Currently, many non-contact part defect detection devices have emerged on the market. However, for metal parts with highly reflective surfaces, fringe projection three-dimensional measurement can easily lead to overexposure of the captured image, making it difficult to achieve high-precision defect detection. Existing solutions to this problem mostly utilize multiple exposure methods, adaptive fringe projection, and deep learning-based methods. However, the algorithm architecture of these methods is complex and cannot meet the needs of fast online detection; or they use hardware-assisted methods, but the hardware structure is complex and the detection cost is high. Therefore, there is an urgent need to design a non-contact part defect detection device with a simple execution architecture, low cost, intelligence, and the ability to effectively solve the problem of image overexposure. Summary of the Invention

[0004] In order to overcome the shortcomings of the existing technology, an intelligent portable 3D anti-overexposure part defect detection device and method are provided, which solves the problem of overexposure of captured images caused by fringe projection 3D measurement and achieves the purpose of 3D reconstruction of highly reflective parts.

[0005] The technical solution of the intelligent portable 3D overexposure prevention part defect detection device and method of the present invention includes: An intelligent portable 3D anti-overexposure part defect detection device, comprising a protective box, a detection table telescopic system, a projection-imaging system, a workpiece rotation system, a main control system, and an indicator light module; the protective box comprises a protective box cover and a protective box body, and the protective box cover and the protective box body are flexibly opened and closed by a hinge connection structure; the detection table telescopic system comprises a guide rail frame, a detection table, a plane mirror mounting slot, a handle, and a limit module, the guide rail frame is located inside the protective box body, and is interconnected with the detection table through the limit module, the handle adopts a recessed structure, is located on the front side of the detection table, and is used to control the movement of the detection table, the plane mirror mounting slot is located on the right side of the detection table, and a plane mirror is installed on the plane mirror mounting slot; the projection-imaging system comprises a protective box cover and a protective box body, and a hinge connection structure is used to connect the protective box cover and the detection table to the detection table. -The imaging system includes an industrial projector, a camera, and a plane mirror. The industrial projector and the camera are installed on the upper surface of the guide rail frame according to the principle of similar triangles, and the plane mirror is installed on the plane mirror mounting slot; the workpiece rotation system consists of a workpiece rotating disk and a rotating disk drive module. The workpiece rotating disk is installed above the detection table, and the rotating disk drive module is located inside the detection table; the indicator light module includes a standby indicator light and a working indicator light. Both the standby indicator light and the working indicator light are arranged on the right side of the protective box. The main control system includes a control module, a reset button, and a host computer. The control module is fixed on the upper surface of the guide rail frame, the reset button is arranged on the left side of the working indicator light, and the host computer is connected to the device via a data cable.

[0006] An intelligent portable 3D anti-overexposure part defect detection device, wherein the limit module in the telescopic system of the detection platform consists of a lock A with an elastic rod A, a guide rail A, a lock B with an elastic rod B, and a guide rail B. The limit module has a bilaterally symmetrical mechanical structure, wherein the guide rail A has a plurality of slots, the elastic rod A is an elastic slender metal strip, the lock A moves in the guide rail A, and when stationary, the lock A is embedded in the slot to realize the limit function; the guide rail B has a plurality of slots, the elastic rod B is an elastic slender metal strip, the lock B moves in the guide rail B, and when stationary, the lock B is embedded in the slot to realize the limit function.

[0007] An intelligent portable 3D anti-overexposure part defect detection device, wherein the rotating disk drive module in the workpiece rotation system is composed of a pressure sensor, a motor, and a coupling. The motor is connected to the workpiece rotating disk through the coupling. The center of the workpiece rotating disk and the output shaft of the motor are on the same vertical line. The pressure sensor is installed below the workpiece rotating disk to detect the pressure value above the workpiece rotating disk. The main control system controls the start and stop of the motor according to the pressure value. The motor drives the workpiece rotating disk to control the rotation angle of the workpiece to be tested. The projection-imaging system is used to obtain workpiece view information. The camera and industrial projector are located in front of the workpiece to be tested. The plane mirror is located on the side of the workpiece to be measured, and the camera obtains dual-view information of the workpiece to be measured in real time; the main control system is used to control the operation of the entire device and process and analyze image information. The control module in the main control system is composed of a data storage module, a data sending module, a synchronization trigger module, and a main controller. The data storage module is used to cache received data, and the main controller is used for data processing, analysis, judgment, and issuing instructions to control the operation status of the entire device. The data sending module is used to send image information to the host computer, and the synchronization trigger module is used to control the camera and industrial projector to synchronize shooting and projection. The reset button controls the main control system to return to standby state.

[0008] A detection method for an intelligent portable 3D anti-overexposure part defect detection device is used. When inspecting a workpiece, a frame of image captured by the camera contains image information of the workpiece from two different perspectives. The right view is the view directly obtained by the camera, and the left view is the view indirectly obtained through the plane mirror imaging. The camera corresponding to the left view is set as a virtual camera. The camera and virtual camera are calibrated to obtain their respective intrinsic parameter matrices. 、 , the external parameter matrix 、 ; A fringe image is projected onto the surface of the object. The camera captures the image and obtains the corresponding grayscale value distribution map. The grayscale value is determined to be 255 pixel by pixel. If it is equal to 255, the point is a saturated pixel. The ratio of saturated pixels is counted. When the ratio is greater than a threshold, the workpiece is determined to be a metal workpiece and the automatic adjustment device working mode is dual-view detection mode. When the ratio is less than the threshold, the workpiece is determined to be a general workpiece and the working mode is adjusted to single-view detection mode. In single-view detection mode, the right view information is used to restore the three-dimensional shape of the workpiece to be tested and detect defect information, and the left view information is not used; In the dual-view detection mode, the original image obtained by the camera is processed as follows: the right view information is recorded as image 1, the left view information is mirrored and recorded as image 2, and a mask of the overexposed area of ​​image 1 is generated. , is the mask area number, are the mask region coordinates, is the gray value of the current pixel. In image 1, select the mask The coordinates of a pixel on the edge are the anchor point coordinates , and generate a 5×5 window area with the anchor point coordinates as the base point, and select an anchor point coordinate in image 2 , the same size window area is generated in Image 2, and the current row is based on Calculate the matching cost of each pixel in image 2. The smaller the matching cost, the better the matching cost. The higher the matching degree, the In pixels The matching window is centered. , For image 1 and image 2 , Gray value at the coordinate point, is parallax, is the weight function, , weight function Masked by overexposed area and the gradient weighting factor Jointly decided, the calculation formula is in, To avoid the minimum value with zero denominator, is the gradient vector of image 1 at the current pixel coordinate, To obtain, and are the horizontal gradient and the vertical gradient respectively, is the average gradient within the window, Ask for, is the number of pixels in the window, set to =25, Represents the gradient. According to the above formula, the matching information of the edge pixels of the mask of Image 1 is obtained. Then, the mask of Image 1 is mapped to the corresponding area of ​​Image 2. Image 1 is compensated by the information of the corresponding area of ​​Image 2. Finally, the workpiece is 3D reconstructed and the defect information is obtained through the phase shift method.

[0009] The beneficial effects of the present invention are: 1. A monocular camera is used to acquire images of different perspectives through dual optical paths, which solves the problem of image overexposure during the imaging process, simplifies the system architecture, and reduces hardware costs; 2. The limiting function of the limiting module is utilized to realize the control of the distance between the object being measured and the camera, and flexibly adjust the field of view of the camera shooting; 3. The system is integrated in a protective box and has intelligent control functions, which can adapt to the application requirements of different working scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 It is a schematic diagram of the overall structure of the present invention.

[0011] Figure 2 This is a structural diagram of the workpiece rotation system of the present invention.

[0012] Figure 3 It is a schematic structural diagram of the limiting module of the present invention.

[0013] Figure 4 Schematic diagram of the main control system.

[0014] In the figure: 1. Protective box cover, 2. Protective box body, 3. Guide rail frame, 4. Testing table, 5. Plane mirror mounting slot, 6. Handle, 7. Industrial projector, 8. Camera, 9. Plane mirror, 10. Workpiece rotating disk, 11. Pressure sensor, 12. Motor, 13. Control module, 14. Standby indicator light, 15. Working indicator light, 16. Reset button, 17. Limit module, 18. Lock A, 19. Elastic rod A, 20. Guide rail A, 21. Guide rail B, 22. Rotating disk drive module, 23. Coupling, 24. Main controller, 25. Data storage module, 26. Data sending module, 27. Host computer, 28. Synchronous trigger module, 29. Lock B, 30. Elastic rod B. DETAILED DESCRIPTION

[0015] The following is combined with Figure 1 、 2 , 3, and 4 further illustrate the present invention.

[0016] The protective box is the outer protective structure of the entire device. The protective box cover 1 can be opened as needed to meet the debugging requirements of the internal industrial projector 7 and camera 8; when the protective box cover 1 is closed, it can effectively achieve dust protection for the industrial projector 7 and camera 8; When the device is to be stored or transported, the plane mirror mounting slot 5 is in an empty state, the detection platform 4 is retracted to the innermost position of the protective box 2, the motor 12 is in a stopped state, the standby indicator light 14 and the working indicator light 15 are both off, and the protective box cover 1 is closed, so that the entire device can be stored and transported; When the device is powered on, the standby indicator light 14 lights up, and the device is in the standby state. The limit module 17 limits the detection platform 4. According to the detection requirements, the operator operates the handle 6 to pull the detection platform 4 outward. Under the action of the elastic rod A19 and the elastic rod B30, the lock head A18 and the lock head B29 pass through the slots along the guide rails A20 and B21 in sequence. After reaching the target position, the handle 6 is stopped, and the lock heads A18 and B29 are embedded in the corresponding slots to lock the detection platform 4. The plane mirror 9 is installed on the plane mirror mounting slot 5. The industrial projector 7, camera 8, and plane mirror 9 constitute the projection-imaging system of the device. Among them, the camera 8 and industrial projector 7 are located in front of the workpiece to be measured, and the plane mirror 9 is located on the right side of the workpiece to be measured. The entire detection process is divided into a calibration phase and a measurement phase; During the calibration phase, a pinhole imaging model is established. ,in, is the three-dimensional coordinate of any point on the object to be measured, is the image coordinate of the point in the image coordinate system, is the scale factor, is the external parameter matrix, the rotation matrix and translation matrices To describe the parameter matrix from the world coordinate system to the camera coordinate system, is the camera intrinsic parameter matrix, which can be expressed as , and are the focal lengths in the horizontal and vertical directions, and The camera 8 and the virtual camera are calibrated to obtain their respective intrinsic parameter matrices. 、 and external reference 、 , establish the geometric correspondence between components; During the measurement phase, the pressure sensor 11 mounted below the workpiece rotating disk 10 detects the current pressure value and transmits it to the control module 13. The data storage module 25 in the control module 13 stores the pressure value information. The main controller 24 analyzes and determines the current pressure value information. When a workpiece is placed on the workpiece rotating disk 10, the main controller 24 monitors the sudden change in the pressure value information and determines whether it is a human error every 5 seconds. If the pressure value drops back to the initial value within 5 seconds, the main controller 24 determines that it is an error. If the pressure value remains unchanged after 5 seconds, the main controller 24 wakes up the device and puts it into working state. At this time, the working indicator light 15 lights up and the standby indicator light 14 goes out. After entering the working state, the host computer 27 performs an initial judgment to confirm the type of workpiece. The specific steps are as follows: the industrial projector 7 projects a stripe image onto the surface of the object, the camera 8 takes a frame of image, and the image information is sent to the host computer 27 through the data sending module 26 in the control module 13. The host computer 27 counts the proportion of saturated pixels. When the proportion is greater than the threshold, it is determined that the workpiece is a metal workpiece. The main controller 24 adjusts the device detection mode to the dual-view detection mode. When the proportion is less than the threshold, it is determined that the workpiece is a general workpiece and the detection mode is adjusted to the single-view detection mode. After the initial determination is completed, the main controller 24 controls the motor 12 to start and drive the workpiece rotating disk 10 to rotate. If the workpiece to be tested is removed in advance before the detection is completed, the device returns to the standby state and the image data transmission is terminated. In single-view detection mode, only the right view information is used, and the industrial projector 7 projects a sinusoidal grating with a continuously changing phase onto the workpiece to be tested. The camera 8 captures the current grating image in real time and sends it to the host computer 27 via the data sending module 26. The host computer 27 performs three-dimensional reconstruction and defect analysis of the workpiece; In dual-view detection mode, the industrial projector 7 projects a sinusoidal grating with a continuously changing phase onto the workpiece to be tested. Camera 8 synchronously captures the current raster image, mirrors the left view information of the image and records it as image 1, and records the right view information as image 2, and generates the over-exposure area mask of image 1 ,in is the mask area number, are the mask region coordinates, The grayscale value of the pixel at the current coordinate is then selected as an anchor point at the edge of the mask area in image 1. A window of 5×5 is constructed based on this point, and the anchor coordinates in image 2 are covered with a window of the same size. ,pass Find the matching cost of the two coordinates, where , For image 1 and image 2 , Gray value at the coordinate point, is parallax, Represented in image coordinates The parallax is The matching cost when the value is smaller, the higher the matching degree is. is the weight function, , weight function Masked by overexposed area and the gradient weighting factor Jointly decided, the calculation formula is in, To avoid the minimum value with zero denominator, is the gradient vector of image 1 at the current pixel coordinate, and the calculation formula is in and are the horizontal gradient and the vertical gradient respectively, is the average gradient within the window, Ask for, is the number of pixels in the window, here =25, Represents the gradient. Pixels with higher gradients correspond to areas with rich texture details in the image and have higher weights. Pixels with lower gradients correspond to smooth areas in the image and have lower weights. Move the window in image 2 to obtain the point Corresponding respectively of ,in , then in image 2 For image 1 Repeat the above steps to obtain the matching information of all pixel points. Solve the truncated phase and obtain the truncated phase distribution diagram of image 1 and image 2 、 , through matching information, the mask area of ​​image 1 Mapped to image 2, using Phase information compensation of corresponding area The missing phase information is used to construct the three-dimensional morphology of the workpiece to be measured and identify the surface defect information of the workpiece; After the workpiece rotating disk 10 rotates 360 degrees, the camera 8 obtains all surface information of the workpiece to be measured and transmits it to the host computer 27 through the data sending module 26. At this time, the main controller 24 controls the motor 12 to stop running, and then waits for the operator to remove the workpiece that has been measured from the workpiece rotating disk 10. When the workpiece is separated from the workpiece rotating disk 10, the pressure value decreases. After the main controller 24 detects the decrease in pressure value, it issues a command to control the device to switch back to the standby state. At this time, the standby indicator light 14 lights up and the working indicator light 15 goes out. During the detection process, if the reset button 16 is pressed, the main controller 24 will issue a command to make the data storage module 25 clear all the data of this detection and control the device to enter the standby state. After the inspection is completed, the power supply is disconnected, the plane mirror 9 is removed from the plane mirror mounting groove 5 , and the operator operates the handle 6 to control the inspection platform 4 to retract into the protective box 2 .

Claims

1. An intelligent portable 3D overexposure-proof part defect detection device, characterized by: The invention comprises a protective box, a detection table telescopic system, a projection-imaging system, a workpiece rotation system, a main control system, and an indicator light module; the protective box comprises a protective box cover (1) and a protective box body (2), and the protective box cover (1) and the protective box body (2) are flexibly opened and closed by a hinge connection structure; the detection table telescopic system comprises a guide rail frame (3), a detection table (4), a plane mirror mounting groove (5), a handle (6), and a limit module (17); the guide rail frame (3) is located inside the protective box body (2) and is connected to the detection table (4) through the limit module (17); the handle (6) adopts a recessed structure and is located at the front side of the detection table (4) for controlling the movement of the detection table (4); the plane mirror mounting groove (5) is located on the right side of the detection table (4), and a plane mirror (9) is installed on the plane mirror mounting groove (5); the projection-imaging system comprises an industrial projector (7), a camera (8), and a plane mirror (9) , the industrial projector (7) and the camera (8) are installed on the upper surface of the guide rail frame (3) according to the principle of similar triangles, and the plane mirror (9) is installed on the plane mirror mounting groove (5); the workpiece rotation system consists of a workpiece rotating disk (10) and a rotating disk driving module (22), the workpiece rotating disk (10) is installed above the detection table (4), and the rotating disk driving module (22) is located inside the detection table (4); the indicator light module includes a standby indicator light (14) and a working indicator light (15), and the standby indicator light (14) and the working indicator light (15) are both arranged on the right side of the protective box (2), and the main control system includes a control module (13), a reset button (16), and a host computer (27), the control module (13) is fixed on the upper surface of the guide rail frame (3), the reset button (16) is arranged on the left side of the working indicator light (15), and the host computer (27) is connected to the device through a data cable.

2. The intelligent portable 3D overexposure prevention part defect detection device according to claim 1, characterized in that: The limit module (17) in the telescopic system of the detection platform is composed of a lock head A (18) with an elastic rod A (19), a guide rail A (20), a lock head B (29) with an elastic rod B (30), and a guide rail B (21). The limit module (17) has a bilaterally symmetrical mechanical structure. The guide rail A (20) has a plurality of slots. The elastic rod A (19) is a slender metal strip with elasticity. The lock head A (18) moves in the guide rail A (20). When the lock head A (18) is stationary, the lock head A (18) is embedded in the slot to realize the limit function. The guide rail B (21) has a plurality of slots. The elastic rod B (30) is a slender metal strip with elasticity. The lock head B (29) moves in the guide rail B (21). When the lock head B (29) is stationary, the lock head B (29) is embedded in the slot to realize the limit function.

3. The intelligent portable 3D overexposure prevention part defect detection device according to claim 1, characterized in that: The rotating disk driving module (22) in the workpiece rotating system is composed of a pressure sensor (11), a motor (12), and a coupling (23). The motor (12) is connected to the workpiece rotating disk (10) through the coupling (23). The center of the workpiece rotating disk (10) and the output shaft of the motor (12) are on the same line. The pressure sensor (11) is installed below the workpiece rotating disk (10) and is used to detect the pressure value above the workpiece rotating disk (10). The main control system controls the start and stop of the motor (12) according to the pressure value. The motor (12) drives the workpiece rotating disk (10) to control the rotation angle of the workpiece to be measured. The projection-imaging system is used to obtain workpiece view information. The camera (8) and the industrial projector (7) are located in front of the workpiece to be measured. The plane mirror ( 9) is located on the side of the workpiece to be measured, and the camera (8) obtains the dual-view information of the workpiece to be measured in real time; the main control system is used to control the operation of the entire device and process and analyze image information. The control module (13) in the main control system is composed of a data storage module (25), a data sending module (26), a synchronization trigger module (28), and a main controller (24). The data storage module (25) is used to cache received data, and the main controller (24) is used to process, analyze, judge and issue instructions for data, thereby controlling the operation status of the entire device. The data sending module (26) is used to send image information to the host computer, and the synchronization trigger module (28) is used to control the camera (8) and the industrial projector (7) to synchronously shoot and project. The reset button (16) controls the main control system to return to the standby state.

4. A detection method using the intelligent portable 3D overexposure prevention part defect detection device according to claim 1, characterized in that: When inspecting a workpiece, a frame of image captured by the camera (8) contains image information of the workpiece at two different viewing angles. The right view is the view directly obtained by the camera (8), and the left view is the view indirectly obtained by imaging through the plane mirror (9). Let the left view be the virtual camera view, calibrate the camera (8) and the virtual camera, and obtain their respective intrinsic parameter matrices. 、 , the external parameter matrix 、 ; A stripe image is projected onto the surface of the object, and a camera (8) captures the image and obtains a corresponding grayscale value distribution map. The grayscale value is determined pixel by pixel to be equal to 255. If it is equal to 255, the point is a saturated pixel point. The ratio of saturated pixels is counted. When the ratio is greater than a threshold, the workpiece is determined to be a metal workpiece, and the working mode of the automatic adjustment device is the dual-view detection mode. When the ratio is less than the threshold, the workpiece is determined to be a general workpiece, and the working mode is adjusted to the single-view detection mode. In single-view detection mode, the right view information is used to restore the three-dimensional shape of the workpiece to be tested and detect defect information, and the left view information is not used; In the dual-view detection mode, the original image obtained by the camera (8) is processed as follows: the right view information is recorded as image 1, the left view information is mirror-flipped and recorded as image 2, and a mask of the over-exposed area of ​​image 1 is generated. , is the mask area number, are the mask region coordinates, is the gray value of the current pixel. In image 1, select the mask The coordinates of a pixel point on the edge are the anchor point coordinates , and generate a 5×5 window area with the anchor point coordinates as the base point, and select an anchor point coordinate in image 2 , the same size window area is generated in Image 2, and the current row is based on Calculate the matching cost of each pixel in image 2. The smaller the matching cost, the better the matching cost. The higher the matching degree, the In pixels The matching window is centered. , For image 1 and image 2 , Gray value at the coordinate point, is parallax, is the weight function, , weight function Masked by overexposed area and the gradient weighting factor Jointly decided, the calculation formula is in, To avoid the minimum value with zero denominator, is the gradient vector of image 1 at the current pixel coordinate, To obtain, and are the horizontal gradient and the vertical gradient respectively, is the average gradient within the window, Ask for, is the number of pixels in the window, set to =25, Represents the gradient. According to the above formula, the matching information of the edge pixels of the mask of Image 1 is obtained. Then, the mask of Image 1 is mapped to the corresponding area of ​​Image 2. Image 1 is compensated by the information of the corresponding area of ​​Image 2. Finally, the workpiece is 3D reconstructed and the defect information is obtained through the phase shift method.

Images