Ultra-thin vapor chamber detection and positioning device and method based on computer vision

Abstract

The present invention discloses an ultra-thin heat spreader detection and positioning device and method based on computer vision, wherein the computer system comprises target detection and classification, image processing, positioning point acquisition and angle correction; the ultra-thin heat spreader visual detection and positioning mechanism comprises a visual processing module, a camera, a lens and a light source; the feeding mechanism comprises a motor screw module, a shooting platform and a material box, in which the ultra-thin heat spreader is stored; the industrial robot module is responsible for moving to a designated point solved by the visual system to grab the ultra-thin heat spreader; the computer system detects whether there is an ultra-thin heat spreader in the material box and positions the detected ultra-thin heat spreader, and guides the industrial robot to grab it, so as to realize the automated transformation of the ultra-thin heat spreader for detecting and positioning materials in the process of liquid injection and vacuum extraction (abbreviated as one removal), quickly and accurately detect and position the ultra-thin heat spreader in the material box and guide the industrial robot to carry it, thereby improving the production efficiency and benefits of the product.

Description

Technical Field

[0001] The present invention relates to detection and positioning of an ultra-thin vapor chamber, and in particular to a device and method for detecting and positioning an ultra-thin vapor chamber based on computer vision. Background Art

[0002] As electronic products develop towards high performance and miniaturization, ultra-thin heat spreaders are widely used in various thin and portable electronic products, and with the increase in the number of electronic products such as mobile phones and tablets, the demand for ultra-thin heat spreaders has also increased significantly. In the liquid injection and one-step removal process of the ultra-thin heat spreader, positioning is the key technology that determines whether these two processes can proceed smoothly. In these two processes, positioning is divided into the determination of the positioning point and the solution and correction of the rotation angle. The determination of the positioning point ensures that the liquid injection tube of the ultra-thin heat spreader is aligned with the liquid injection needle in the liquid injection process and the vacuum cavity in the one-step removal process; the solution and correction of the rotation angle ensure that the liquid injection tube of the ultra-thin heat spreader always remains vertically upward and will not be inverted. Failure to position will cause the liquid injection needle to be unable to enter the ultra-thin heat spreader for the liquid injection process, resulting in the failure of the liquid injection process, and failure to position will cause the liquid injection tube of the ultra-thin heat spreader to be unable to smoothly enter the vacuum cavity, resulting in the inability to proceed with the one-step removal process. At present, the positioning process of ultra-thin vapor chamber injection and removal process is mainly manual operation. This repetitive and high-intensity work will lead to missed injection and double injection during manual injection, and the probability of double removal during removal will increase. At the same time, with the reduction of the labor force, the labor cost is also increasing. Therefore, there is an urgent need for a device and method that can realize automatic detection and positioning of ultra-thin vapor chamber injection and removal process. Applying computer vision to the ultra-thin vapor chamber injection and removal process can realize efficient and accurate detection and positioning process. Summary of the invention

[0003] The purpose of the present invention is to address the above-mentioned problems and to provide a computer vision-based ultra-thin heat spreader detection and positioning device and method, so as to realize the automated detection and positioning process of the ultra-thin heat spreader during the liquid injection and removal processes, thereby replacing manual labor to improve the production efficiency and benefits of the product.

[0004] To achieve the above object, the present invention adopts the following technical solution:

[0005] An ultra-thin vapor chamber detection and positioning device based on computer vision, including a visual device and a computer system;

[0006] The visual device includes a feeding module, an ultra-thin heat spreader visual detection and positioning mechanism, a workbench and an industrial robot module; the workbench is fixed to the ground by anchor bolts, and the ultra-thin heat spreader visual detection and positioning mechanism is installed on the working panel of the workbench.

[0007] The feeding module is used to place the material box and transport the ultra-thin heat spreader in the material box through the motor screw, ensuring that the ultra-thin heat spreader visual inspection and positioning mechanism can vertically collect complete ultra-thin heat spreader image information and transmit it to the computer system.

[0008] The computer system includes an industrial computer, which is connected to the ultra-thin heat spreader visual detection and positioning mechanism through a network cable; the industrial computer includes a target detection module, an image processing module, and a positioning point acquisition and angle correction module.

[0009] The ultra-thin vapor chamber visual detection and positioning mechanism comprises an image acquisition system and a visual instrument device; the image acquisition system is installed on the visual instrument device;

[0010] The image acquisition system includes a camera, a lens, a light source and a light source controller;

[0011] The visual instrument device includes a built-in dual-axis guide rail, a slider, a light source bracket, a Z-shaped hand-tightening bolt, a crossbeam, a side plate, a knob, a screw, an optical axis, a screw nut, a camera clamp, a fine-tuning slide, a side plate fixing plate, and a silicone pad;

[0012] The visual instrument device is installed on the working panel through the threaded holes below the built-in dual-axis guide rail optical rail. The crossbeam and the fine-tuning slide are respectively installed on two sliders, so that the shooting height and the light source height can be adjusted by sliding the sliders on the guide rails; the light source bracket is installed on the crossbeam through Z-type hand-screwed bolts, and the Z-type hand-screwed bolts are used to adjust the width between the two light source brackets to achieve the positioning of the light source. After the positioning is completed, it is fixed to the light source bracket by screws; the side panel fixing plate is connected to the slider, the side panel is installed on the side panel fixing plate, and the two camera clamps are installed between the two optical axes and the lead screw. The lead screw drives the lead screw nut to rotate through the knob to adjust the distance between the two camera clamps, so that the silicone pad clamps the camera, and the camera is leveled by adjusting the fine-tuning slide.

[0013] The feeding module includes a motor, a positioning plate, a clamping cylinder, a material box, an ultra-thin heat spreader, a positioning pin, a guide rail and a slide; the guide rail is installed on the working panel through a threaded hole, the motor and the guide rail are connected through a coupling and fixed with bolts, the positioning plate is installed on the slide, so that the ultra-thin heat spreader can be transported along the guide rail through the slide, the material box is positioned by the positioning pin installed on the positioning plate and clamped by the clamping cylinders on both sides, and the groove opened in the material box is used to place the ultra-thin heat spreader;

[0014] The workbench comprises a work panel and a bracket; the bracket is fixed to the ground through anchor bolts, and the work panel is installed on the bracket through bolt holes.

[0015] The industrial robot module includes a robot arm base, a robot arm, a vacuum suction cup mounting plate and a vacuum suction cup; the robot arm base is fixed to the ground by anchor bolts, the robot arm is connected to the base by bolts, the vacuum suction cup mounting plate is installed on the sixth axis of the robot arm, and the vacuum suction cup is fixed on the mounting plate.

[0016] The specific steps of the ultra-thin vapor chamber detection and positioning method based on computer vision are as follows:

[0017] Step 1: The target detection and classification module includes feature recognition, material detection and quadrant classification; the image collected by the camera in the ultra-thin vapor chamber visual positioning mechanism is transmitted to the computer system, and the Yolov5 network is used to identify the characteristics of the ultra-thin vapor chamber injection port. Whether the ultra-thin vapor chamber is stored in the material box is detected based on whether the characteristics are recognized. After the existence of the ultra-thin vapor chamber is detected, quadrant classification is performed according to the direction of the injection port;

[0018] Step 2: The image processing module includes binarization, connected domain segmentation, image filling, contour extraction and minimum bounding rectangle; the required part of the image is extracted by binarization, and the image is segmented into the injection port feature area and the cover plate area by connected domain segmentation, and image filling, contour extraction and minimum bounding rectangle are solved for each segmented area;

[0019] Step 3: The positioning point acquisition and angle correction module includes the solution of the midpoint equation, the solution of the bottom contour line equation, the positioning point solution and the rotation angle solution and correction part; according to step 2, after solving the minimum circumscribed rectangle for each segmented area, the minimum circumscribed rectangle is used to calculate the midpoint of the injection port feature area and the midpoint equation and the bottom contour line equation of the cover plate area and solve the intersection of the two straight line equations, and the midpoint of the intersection and the midpoint of the injection port feature area are solved. The midpoint coordinates finally solved are the positioning point of the ultra-thin heat spreader; then, according to the deflection angle obtained by the minimum circumscribed rectangle of the cover plate area and the classification result of the Yolov5 network before, the deflection angle is corrected to obtain the rotation angle for turning the ultra-thin heat spreader straight. According to the obtained positioning point and rotation angle, the industrial robot is guided to the positioning point, the sixth axis is rotated according to the rotation angle, and finally the ultra-thin heat spreader is transported by the vacuum suction cup installed on the connector.

[0020] Step 4: After moving this ultra-thin heat spreader, the motor of the feeding module will be driven to rotate, thereby driving the slide to move forward, allowing the camera to capture the image of the next ultra-thin heat spreader, and then repeat steps one to three until the slide gradually moves to the end point, completing the grabbing of all ultra-thin heat spreaders in the material box.

[0021] In the step 1, a single-channel industrial camera is used as the camera, a telecentric lens is used as the lens to reduce camera distortion, a white surface light source is used as the light source, and the image captured by the camera is transmitted to the computer system, and the image is processed by software.

[0022] In the step 1, the collected ultra-thin vapor chamber image is put into the Yolov5 network for model inference and prediction. The identified features (liquid injection port of the ultra-thin vapor chamber) are marked with frame lines on the obtained output image. Whether the ultra-thin vapor chamber exists in the captured image is determined based on whether the frame lines exist in the output image, so as to achieve the purpose of feature recognition and material detection;

[0023] Based on the identified injection port features, the Yolov5 network will also perform quadrant classification according to the orientation of the injection port features. It will be divided into quadrants one, two, three, and four according to the conventional quadrant division method. It will determine the orientation of the injection port in the current image and divide it into the corresponding quadrant category.

[0024] In the step 2, the collected ultra-thin heat spreader image is subjected to global binary threshold segmentation processing to obtain a binary image of only the ultra-thin heat spreader area, and then the obtained binary image is divided into connected domains to divide the liquid injection port features into area 1 and the cover plate area into area 2.

[0025] In the step 2, image filling is performed on the injection port feature area and the cover plate area respectively to ensure that there are no defects such as holes or gaps on the images of the two areas, and to avoid the situation where multiple contours appear during contour extraction. After the image filling is completed, contour extraction is performed on the two areas respectively to extract the outer contours of the injection port feature and the cover plate, and then the minimum circumscribed rectangles are solved for the two outer contours respectively to obtain two minimum circumscribed rectangles.

[0026] In step three, the minimum circumscribed rectangle solved for the injection port feature can be used to calculate the coordinates of the center point of the rectangle and the midpoint coordinates of the top contour line using the coordinates of the upper left corner of the rectangle and the width and height parameters of the rectangle. The equation of the perpendicular bisector in the vertical direction can be solved based on the coordinates of the two points.

[0027] In step three, for the minimum circumscribed rectangle of the cover area, the coordinates of the lower left and lower right corners of the rectangle can be solved by using the coordinates of the upper left corner of the rectangle and the width and height coefficients of the rectangle. The bottom contour line equation of the rectangle can be calculated based on the coordinates of these two points.

[0028] In the step three, the coordinates of the intersection of the two straight lines can be solved according to the equation of the perpendicular bisector in the vertical direction obtained from the liquid injection port feature and the equation of the bottom contour line obtained from the cover plate area, and the midpoint coordinates obtained from the liquid injection port feature and the intersection coordinates are used to solve the midpoint to obtain the coordinates of the positioning point; at the same time, the deflection angle solved according to the minimum circumscribed rectangle of the cover plate area is combined with the classification result of the Yolov5 network to correct the deflection angle and obtain the rotation angle for turning the ultra-thin heat spreader straight.

[0029] In the step three, according to the obtained positioning point and rotation angle, the industrial robot is guided to reach the positioning point, the sixth axis is rotated according to the rotation angle, and finally the ultra-thin heat spreader is transported using the vacuum suction cup installed on the connecting piece.

[0030] In the step 4, when the robotic arm grabs the ultra-thin heat spreader at the positioning point, an electrical signal will be sent out. After receiving the electrical signal, the controller will send a signal to the motor driver to make the motor start to rotate, thereby driving the slide to move, and the material box also moves forward on the guide rail. When the Yolov5 network detects the presence of an ultra-thin heat spreader in the image taken by the camera again, the motor stops, and the material box also stops moving forward. At the same time, the industrial computer starts to repeat steps 1 to 3 to solve the positioning point and rotation angle of the ultra-thin heat spreader, and then guide the robotic arm to grab, and continue the cycle until the slide moves to the end position, that is, the grabbing of all ultra-thin heat spreaders in the material box is completed.

[0031] Compared with the prior art, the present invention has the following advantages:

[0032] 1) The method of the present invention combines the Yolov5 network with traditional machine vision to solve the problem of poor stability of traditional visual algorithms in industrial environments due to factors such as noise and ambient light sources. By introducing a convolutional neural network, the robustness and universality of the method of the present invention can be improved, so that it can be applied to the automated detection and positioning process of most ultra-thin heat sink filling and one-step removal processes.

[0033] 2) The method of the present invention adopts the Yolov5 network to classify the orientation of the ultra-thin heat sink liquid injection port features, and corrects the rotation angle obtained by applying the minimum circumscribed rectangle algorithm to the cover plate area according to the classification results, so as to achieve the purpose of compensating for the defects of the traditional visual algorithm through the Yolov5 network.

[0034] 3) The method of the present invention adopts a self-designed ultra-thin heat spreader positioning point solution algorithm for the structure of the ultra-thin heat spreader, which can ensure that the industrial robot always ensures that the injection tube and the injection needle in the injection process and the vacuum chamber in the removal process are aligned during the subsequent grabbing of the ultra-thin heat spreader for injection and removal processes.

[0035] 4) The method of the present invention realizes the automated liquid injection and removal process of the ultra-thin heat sink by combining with an industrial robot, replacing the traditional manual operation, improving the production efficiency of the product and saving the production cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A flow chart of the method for detecting and positioning an ultra-thin vapor chamber according to the present invention;

[0037] Figure 2 This is an overall structural diagram of the ultra-thin vapor chamber detection and positioning device of the present invention;

[0038] Figure 3 It is a structural diagram of the feeding module of the present invention;

[0039] Figure 4 It is a structural diagram of the visual detection and positioning mechanism of the ultra-thin vapor chamber of the present invention;

[0040] Figure 5 It is a structural diagram of a workbench of the present invention;

[0041] Figure 6 It is a structural diagram of the industrial robot module of the present invention.

[0042] Figure 2 , Figure 3 , Figure 4 , Figure 5 and Figure 6 As shown in the figure: 1 is a feeding module, 2 is a visual inspection and positioning mechanism for an ultra-thin heat spreader, 3 is a workbench, and 4 is an industrial robot module; 1-1 is a motor, 1-2 is a positioning plate, 1-3 is a clamping cylinder, 1-4 is a material box, 1-5 is an ultra-thin heat spreader, 1-6 is a positioning pin, 1-7 is a guide rail, 1-8 is a slide table, 2-1 is a built-in dual-axis guide rail, 2-2 is a slider, 2-3 is a light source bracket, 2-4 is a Z-shaped hand-tightening bolt, 2-5 is a beam, 2 -6 is the side panel, 2-7 is the knob, 2-8 is the screw, 2-9 is the optical axis, 2-10 is the screw nut, 2-11 is the camera clamp, 2-12 is the fine-tuning slide, 2-13 is the side panel fixing plate, 2-14 is the camera, 2-15 is the silicone pad, 2-16 is the lens, 2-17 is the light source, 3-1 is the working panel, 3-2 is the bracket, 4-1 is the robot arm base, 4-2 is the robot arm, 4-3 is the vacuum suction cup mounting plate, and 4-4 is the vacuum suction cup. DETAILED DESCRIPTION

[0043] In order to better understand the present invention, the present invention is further described below in conjunction with the accompanying drawings, but the embodiments of the present invention are not limited thereto. In addition, 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.

[0044] The ultra-thin heat spreader detection and positioning device based on computer vision is composed of two parts: a visual device and a computer system.

[0045] like Figure 2 As shown, an embodiment of the present invention provides an ultra-thin vapor chamber detection and positioning device based on computer vision, comprising: Figure 3 The feeding module shown, Figure 4 The ultra-thin heat spreader visual inspection positioning mechanism shown, Figure 5 The workbench shown and Figure 6 The industrial robot module and computer system shown are used to implement the ultra-thin heat sink detection and positioning method based on computer vision.

[0046] The specific structure is as follows:

[0047] like Figure 3 As shown, the feeding module 1 includes a motor 1-1, a positioning plate 1-2, a clamping cylinder 1-3, a material box 1-4, an ultra-thin heat spreader 1-5, a positioning pin 1-6, a guide rail 1-7 and a slide 1-8; the guide rail 1-7 is installed on the working panel 3-2 through a threaded hole, the motor 1-1 and the guide rail 1-7 are connected through a coupling and fixed with bolts, the positioning plate 1-2 is installed on the slide 1-8, so that the ultra-thin heat spreader 1-5 can be transported along the guide rail 1-7 through the slide 1-8, the material box 1-4 is positioned by the positioning pin 1-6 installed on the positioning plate 1-2 and is clamped by the clamping cylinders 1-3 on both sides, and the groove opened in the material box 1-4 is used to place the ultra-thin heat spreader 1-5;

[0048] like Figure 4As shown, the ultra-thin heat spreader visual inspection positioning mechanism 2 includes an image acquisition system and a visual instrument device; the image acquisition system is installed on the visual instrument device; the image acquisition system includes a camera 2-14, a lens 2-16, a light source 2-17 and a light source controller; the visual instrument device includes a built-in dual-axis guide rail 2-1, a slider 2-2, a light source bracket 2-3, a Z-type hand-tightening bolt 2-4, a crossbeam 2-5, a side plate 2-6, a knob 2-7, a screw 2-8, an optical axis 2-9, a screw nut 2-10, a camera clamp 2-11, a fine-tuning slide 2-12, a side plate fixing plate 2-13, and a silicone pad 2-15; the visual instrument device is installed on the working panel 3-2 through the threaded holes below the built-in dual-axis guide rail 2-1, and the crossbeam 2-5 and the fine-tuning slide 2-12 are respectively installed on the two sliders 2-2, so that the visual instrument device can be installed on the working panel 3-2 through the threaded holes below the built-in dual-axis guide rail 2-1, and the crossbeam 2-5 and the fine-tuning slide 2-12 are respectively installed on the two sliders 2-2, so that the visual instrument device can be installed on the working panel 3-2 through the threaded holes below the built-in dual-axis guide rail 2-1. The shooting height and the light source height are adjusted by sliding the block 2-2 on the guide rail 2-1; the light source bracket 2-3 is installed on the crossbeam 2-5 by means of the Z-type hand-tightening bolt 2-4, and the width between the two light source brackets 2-3 is adjusted by means of the Z-type hand-tightening bolt 2-4, so as to realize the positioning of the light source 2-17, and after the positioning is completed, it is fixed to the light source bracket 2-3 by means of screws; the side panel fixing plate 2-13 is connected to the slider 2-2, the side panel 2-6 is installed on the side panel fixing plate 2-13, and the two camera clamps 2-11 are installed between the two optical axes 2-9 and the screw rod 2-8, and the screw rod 2-8 is driven by the knob 2-7 to rotate the screw rod nut 2-10 to adjust the distance between the two camera clamps 2-11, so that the silicone pad 2-15 clamps the camera 2-14, and the level of the camera 2-14 is ensured by adjusting the fine-tuning slide 2-12.

[0049] like Figure 5 As shown, the workbench 3 includes a work panel 3-1 and a bracket 3-2; the bracket 3-2 is fixed to the ground through anchor bolts, and the work panel 3-1 is installed on the bracket 3-2 through bolt holes.

[0050] like Figure 6 As shown, the industrial robot module 4 includes a robot arm base 4-1, a robot arm 4-2, a vacuum suction cup mounting plate 4-3 and a vacuum suction cup 4-4; the robot arm base 4-1 is fixed to the ground by anchor bolts, the robot arm 4-2 is connected to the base 4-1 by bolts, the vacuum suction cup mounting plate 4-3 is installed on the sixth axis of the robot arm 4-1, and the vacuum suction cup 4-4 is fixed on the mounting plate 4-3.

[0051] The computer system mainly refers to an industrial control computer. The industrial computer is connected to the camera through a network cable. The industrial computer executes the ultra-thin heat sink detection and positioning method based on computer vision, including a target detection module, an image processing module, and a positioning point acquisition and angle correction module. Figure 1 As shown, the detection and positioning method includes the following steps:

[0052] Step 1: The camera uses a single-channel industrial camera, the lens uses a telecentric lens to reduce camera distortion, and the light source uses a white surface light source. The pictures collected by the camera are transmitted to the computer system, and the images are processed by the software.

[0053] Step 2: The collected ultra-thin heat spreader image is put into the Yolov5 network for model inference and prediction. The identified features (liquid injection port of the ultra-thin heat spreader) will be marked with frame lines on the output image. Whether there is an ultra-thin heat spreader in the captured image is determined based on whether there is a frame line in the output image, so as to achieve the purpose of feature recognition and material detection.

[0054] Step 3: Based on the identified injection port features, the Yolov5 network will also perform quadrant segmentation according to the orientation of the injection port features. It is divided into quadrants one, two, three, and four according to the conventional quadrant division method. The orientation of the injection port in the current image is determined and divided into the corresponding quadrant category.

[0055] Step 4: After saving the detection and classification results, perform global binary threshold segmentation processing on the collected ultra-thin vapor chamber image f(x, y) to obtain a binary image g(x, y) of only the ultra-thin vapor chamber area. grag To set the segmentation threshold:

[0056] Secondly, the connected domain partitioning is used for the binary image, and the injection port feature is divided into region 1, and the cover plate region is divided into region 2.

[0057] Step 5: Perform image filling on the injection port feature area and the cover plate area respectively to ensure that there are no defects such as holes or gaps on the images of the two areas to avoid multiple contours during contour extraction. After completing the image filling, perform contour extraction on the two areas respectively to extract the outer contours of the injection port feature and the cover plate, and then solve the minimum circumscribed rectangles of the two outer contours respectively to obtain two minimum circumscribed rectangles.

[0058] Step 6: For the minimum circumscribed rectangle solved for the injection port feature, the coordinates of the center point of the rectangle (x0, y0) can be calculated using the coordinates of the upper left corner of the rectangle (x0, y0) and the width w1 and height h1 coefficients of the rectangle. c1 ,y c1 ) and the midpoint coordinates of the top contour line (x1, y1). According to the coordinates of the two points, the vertical bisector equation L1 can be solved: k1x+b1=y.

[0059] Center point coordinates Top contour midpoint coordinates Perpendicular bisector equation parameters

[0060] Step 7: For the minimum circumscribed rectangle of the cover area, the coordinates of the upper left corner of the rectangle (x2, y2) and the width w2 and height h2 coefficients of the rectangle can be used to solve the coordinates of the lower left corner (x3, y3) and the lower right corner (x4, y4) of the rectangle. Based on the coordinates of these two points, the bottom contour line equation of the rectangle can be calculated: k2x+b2=y.

[0061] Bottom contour equation parameters

[0062] Step 8: Based on the vertical midline equation L1 obtained from the injection port characteristics and the bottom contour line equation L2 obtained from the cover plate area, the intersection coordinates of the two straight lines (x i ,y i ), and use the midpoint coordinates (x c1 ,y c1 ) and the intersection coordinates (x i ,y i ) to solve the midpoint and obtain the coordinates of the positioning point (x A ,y A ); At the same time, the deflection angle is corrected according to the deflection angle solved by the minimum circumscribed rectangle of the cover area and the classification result of the Yolov5 network to obtain the rotation angle for turning the ultra-thin heat spreader straight.

[0063] Intersection coordinates Anchor point coordinates

[0064] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. An ultra-thin vapor chamber detection and positioning method based on computer vision, using an ultra-thin vapor chamber detection and positioning device based on computer vision, including a visual device and a computer system; The visual device comprises a feeding module (1), an ultra-thin vapor chamber visual detection and positioning mechanism (2), a workbench (3), and an industrial robot module (4); the workbench (3) is fixed to the ground via anchor bolts, and the ultra-thin vapor chamber visual detection and positioning mechanism (2) is mounted on a working panel of the workbench (3); The feeding module (1) is used to place the material box and to feed the ultra-thin heat spreader (1-5) in the material box through a motor screw, so as to ensure that the ultra-thin heat spreader visual detection and positioning mechanism (2) can vertically collect complete ultra-thin heat spreader image information and transmit it to the computer system; The computer system comprises an industrial computer, which is connected to the ultra-thin heat-sink visual detection and positioning mechanism (2) through a network cable; the industrial computer comprises a target detection module, an image processing module, and a positioning point acquisition and angle correction module; Features: The following steps are involved: Step 1: The target detection module of the industrial computer performs feature recognition, material detection and quadrant classification; the image collected by the ultra-thin vapor chamber visual positioning mechanism (2) is transmitted to the computer system, and the filling port feature of the ultra-thin vapor chamber (1-5) is identified using the Yolov5 network. Whether the ultra-thin vapor chamber (1-5) is stored in the feeding module (1) is detected based on whether the feature is recognized. After the presence of the ultra-thin vapor chamber (1-5) is detected, quadrant classification is performed based on the direction of the filling port; Step 2: The image processing module of the industrial computer uses binarization to extract the required part of the image, and divides the image into the injection port feature area and the cover plate area through connected domain segmentation, and performs image filling, contour extraction and solution of the minimum circumscribed rectangle for each segmented area; Step 3: The positioning point obtaining and angle correction module of the industrial computer solves the minimum circumscribed rectangle for each segmented area according to step 2, and then uses the minimum circumscribed rectangle to calculate the midpoint of the injection port feature area and the midpoint equation and the bottom contour line equation of the cover plate area, and solves the intersection of the two straight line equations, and solves the midpoint of the intersection and the midpoint of the injection port feature area. The midpoint coordinates finally solved are the positioning point of the ultra-thin heat spreader; then, based on the deflection angle obtained by the minimum circumscribed rectangle of the cover plate area and the classification result of the Yolov5 network, the deflection angle is corrected to obtain the rotation angle for turning the ultra-thin heat spreader straight; based on the obtained positioning point and rotation angle, the industrial robot module (4) is guided to carry the ultra-thin heat spreader (1-5); Step 4: After moving this ultra-thin vapor chamber, drive the feeding module (1) so that the ultra-thin vapor chamber visual detection and positioning mechanism (2) captures the image of the next ultra-thin vapor chamber, and then repeats steps 1 to 3 until all ultra-thin vapor chambers in the feeding module (1) are grabbed.

2. The method for detecting and positioning an ultra-thin vapor chamber based on computer vision according to claim 1, characterized in that: In the step 1, the collected ultra-thin vapor chamber image is put into the Yolov5 network for model inference and prediction. The identified feature, i.e., the liquid injection port of the ultra-thin vapor chamber, is marked with a frame line on the obtained output image. Whether the ultra-thin vapor chamber exists in the captured image is determined based on whether the frame line exists in the output image, so as to achieve the purpose of feature recognition and material detection. Based on the identified injection port features, the Yolov5 network will also perform quadrant classification according to the orientation of the injection port features. It will be divided into quadrants one, two, three, and four according to the conventional quadrant division method. It will determine the orientation of the injection port in the current image and divide it into the corresponding quadrant category.

3. The method for detecting and positioning an ultra-thin vapor chamber based on computer vision according to claim 1, characterized in that: In the step 2, the collected ultra-thin vapor chamber image is binarized to obtain a binarized image of only the ultra-thin vapor chamber area, and then the obtained binarized image is divided into a liquid injection port feature area and a cover plate area by using a connected domain division method; The image filling and contour extraction are performed on the injection port feature area and the cover plate area respectively, and the outer contour of the injection port feature and the outer contour of the cover plate are extracted. Then, the minimum circumscribed rectangle of the two outer contours is solved respectively, and two minimum circumscribed rectangles are solved respectively.

4. The method for detecting and positioning an ultra-thin vapor chamber based on computer vision according to claim 1, characterized in that: In the step 3, for the minimum circumscribed rectangle solved for the injection port feature, the coordinates of the center point of the rectangle and the midpoint coordinates of the top contour line are calculated using the coordinates of the upper left corner of the rectangle and the length and width parameters of the rectangle. The perpendicular bisector equation in the vertical direction can be solved based on the coordinates of the two points. For the minimum circumscribed rectangle solved for the cover plate, the bottom contour equation of the rectangle is calculated using the coordinates of the upper left corner of the rectangle and the length and width parameters of the rectangle; The coordinates of the intersection of the two straight lines are solved according to the equation of the perpendicular bisector obtained from the injection port feature and the equation of the bottom contour line obtained from the cover plate area. The midpoint coordinates obtained from the injection port feature and the intersection coordinates are used to solve the midpoint to obtain the coordinates of the positioning point. At the same time, the deflection angle solved according to the minimum circumscribed rectangle of the cover plate area is combined with the classification result of the Yolov5 network to correct the deflection angle and obtain the rotation angle for turning the ultra-thin heat spreader straight.

5. The method for detecting and positioning an ultra-thin vapor chamber based on computer vision according to claim 1, characterized in that: The ultra-thin vapor chamber visual detection and positioning mechanism (2) comprises an image acquisition system and a visual instrument device; the image acquisition system is installed on the visual instrument device; The image acquisition system includes a camera (2-14), a lens (2-16), a light source (2-17) and a light source controller; The visual instrument device includes a built-in dual-axis guide rail (2-1), a slider (2-2), a light source bracket (2-3), a Z-shaped hand-tightening bolt (2-4), a crossbeam (2-5), a side plate (2-6), a knob (2-7), a screw (2-8), an optical axis (2-9), a screw nut (2-10), a camera clamp (2-11), a fine-tuning slide (2-12), a side plate fixing plate (2-13), and a silicone pad (2-15); The visual instrument device is installed on the working panel (3-1) through the threaded hole below the built-in double-axis guide rail (2-1), and the crossbeam (2-5) and the fine-tuning slide (2-12) are respectively installed on two sliders (2-2). The shooting height and the light source height are adjusted by sliding the sliders (2-2) on the guide rail (2-1); the light source bracket (2-3) is installed on the crossbeam (2-5) through the Z-shaped hand-tightening bolt (2-4), and the width between the two light source brackets (2-3) is adjusted by using the Z-shaped hand-tightening bolt (2-4), so as to realize the positioning of the light source (2-17). After the positioning is completed, The screws are fixed on the light source bracket (2-3); the side plate fixing plate (2-13) is connected to the slider (2-2); the side plate (2-6) is installed on the side plate fixing plate (2-13); the two camera clamps (2-11) are installed between the two optical axes (2-9) and the screw rod (2-8); the screw rod (2-8) drives the screw nut (2-10) to rotate through the knob (2-7) to adjust the distance between the two camera clamps (2-11), so that the silicone pad (2-15) clamps the camera (2-14); and the camera (2-14) is kept level by adjusting the fine-tuning slide (2-12).

6. The method for detecting and positioning an ultra-thin vapor chamber based on computer vision according to claim 1, characterized in that: The feeding module (1) comprises a motor (1-1), a positioning plate (1-2), a clamping cylinder (1-3), a material box (1-4), an ultra-thin soaking plate (1-5), a positioning pin (1-6), a guide rail (1-7) and a slide table (1-8); the guide rail (1-7) is mounted on the workbench (3) through a threaded hole, the motor (1-1) and the guide rail (1-7) are connected through a coupling and fixed by bolts, the positioning plate (1-2) is mounted on the slide table (1-8), so that the ultra-thin soaking plate (1-5) can be transported along the guide rail (1-7) through the slide table (1-8), the material box (1-4) is positioned by the positioning pin (1-6) mounted on the positioning plate (1-2) and is clamped by the clamping cylinders (1-3) on both sides, and the groove provided in the material box (1-4) is used to place the ultra-thin soaking plate (1-5).

7. The method for detecting and positioning an ultra-thin vapor chamber based on computer vision according to claim 1, characterized in that: The workbench (3) comprises a work panel (3-1) and a bracket (3-2); the bracket (3-2) is fixed to the ground via anchor bolts, and the work panel (3-1) is mounted on the bracket (3-2) via bolt holes.

8. The method for detecting and positioning an ultra-thin vapor chamber based on computer vision according to claim 1, characterized in that: The industrial robot module (4) comprises a robot arm base (4-1), a robot arm (4-2), a vacuum suction cup mounting plate (4-3) and a vacuum suction cup (4-4); the robot arm base (4-1) is fixed to the ground via anchor bolts, the robot arm (4-2) is connected to the base (4-1) via bolts, the vacuum suction cup mounting plate (4-3) is mounted on the sixth axis of the robot arm (4-2), and the vacuum suction cup (4-4) is fixed on the vacuum suction cup mounting plate (4-3).

Images