Component image acquisition system and method applicable to automated optical inspection

Abstract

A component image acquisition system and method applicable to automatic optical inspection are disclosed. The image processing server uses an object recognition model to identify electronic components in the image to be processed, finds the four unique electronic components on the board, and calculates the center image coordinates of the four electronic components. The center image coordinates of the components are then grouped with the standard coordinates of the centers of nearby components into four coordinate groups. A coordinate transformation algorithm is used to calculate four sets of coordinate transformation coefficients. The maximum and minimum values ​​of each coordinate transformation coefficient group are removed and then averaged to calculate the coordinate transformation coefficient. The image to be processed is converted into a transformed image using the coordinate transformation algorithm of the coordinate transformation coefficients. At least one component image is acquired based on the center coordinates and acquisition range of each acquired information. This achieves the technical effect of providing component images with the correct acquisition range.

Description

Technical Field

[0001] An image acquisition system and method thereof, particularly a component image acquisition system and method thereof applicable to automated optical inspection. Background Technology

[0002] In the field of PCB manufacturing, many inspection methods are gradually shifting to optical inspection methods to facilitate the analysis of production data and the improvement of manufacturing processes. When analyzing specific problems, due to the relationship between the area and volume of electronic components on the PCB, it is necessary to further acquire high-resolution images of the electronic components in order to analyze them. Existing technologies usually acquire high-resolution images at fixed positions.

[0003] However, if the circuit board placed in the optical inspection device is offset or deflected, the high-resolution image obtained may not be able to show the complete electronic components, which will affect the analysis process and lead to incorrect detection results. Therefore, relevant solutions need to be proposed.

[0004] In summary, it is evident that existing technologies have long suffered from the problem of obtaining erroneous analysis results due to board offset or deflection when acquiring high-resolution images at fixed positions. Therefore, it is necessary to propose improved technical methods to solve this problem. Summary of the Invention

[0005] In view of the problem that existing technologies for high-resolution image acquisition at fixed positions can lead to erroneous analysis results due to board offset or deflection, this invention discloses a component image acquisition system and method suitable for automated optical inspection, wherein:

[0006] The component image acquisition system disclosed in this invention, applicable to automated optical inspection, includes: an automated optical inspection (AOI) machine and an image processing server. The image processing server further includes: a standard coordinate database, an image acquisition database, a query module, a component identification module, a conversion calculation module, and a component image acquisition module.

[0007] An automated optical inspection machine acquires an image of the complete circuit board to be processed, provides the image of the circuit board and the circuit board model to an image processing server, and obtains an image of at least one acquisition element from the image processing server for subsequent inspection.

[0008] The image processing server acquires the image to be processed and the circuit board model from the automatic optical inspection machine, providing at least one image of an acquired component to the machine. The image processing server's standard coordinate database stores the circuit board model and the standard coordinates of four component centers. The image acquisition database stores the circuit board model and multiple acquisition information entries, each containing the acquisition center coordinates and the acquisition range. The image processing server's query module retrieves the four corresponding standard coordinates from the standard coordinate database based on the circuit board model, and also retrieves the corresponding acquisition information from the image acquisition database based on the circuit board model. The image processing server's component recognition module uses an object recognition model to identify the electronic components in the image to be processed. The system identifies the four unique electronic components on the circuit board and calculates their center image coordinates. These coordinates are then grouped with the standard coordinates of nearby component centers into four coordinate groups. The image processing server's conversion calculation module selects three of these four coordinate groups and uses a coordinate transformation algorithm to calculate four sets of coordinate transformation coefficients. The maximum and minimum values ​​of each coefficient are removed, and the average is calculated to obtain the final coordinate transformation coefficient. The image processing server's component image acquisition module uses the coordinate transformation algorithm to convert the image to be processed into a transformed image. Based on the center coordinates and acquisition range of each acquired information, at least one component image is acquired.

[0009] The component image acquisition method for automated optical inspection disclosed in this invention includes the following steps:

[0010] First, the automated optical inspection machine acquires an image of the complete circuit board to be processed, providing the image and the circuit board model to the image processing server. Next, the image processing server has a standard coordinate database and an image acquisition database. The standard coordinate database stores the circuit board model and the standard coordinates of the four component centers. The image acquisition database stores the circuit board model and multiple acquisition information entries, each including the acquisition center coordinates and the acquisition range. Then, the image processing server retrieves the four corresponding standard coordinates from the standard coordinate database based on the circuit board model, and retrieves the corresponding acquisition information from the image acquisition database based on the circuit board model. Finally, the image processing server uses an object recognition model to identify the electronic components in the image to find the unique electronic components on the circuit board. Four electronic components are processed, and their center image coordinates are calculated. These center image coordinates are then grouped with the standard coordinates of nearby component centers into four coordinate groups. Next, the image processing server selects three coordinate groups from these four groups and uses a coordinate transformation algorithm to calculate four sets of coordinate transformation coefficients. The maximum and minimum values ​​of each coordinate transformation coefficient are removed, and the average is calculated to obtain the final coordinate transformation coefficient. The image processing server then uses the coordinate transformation algorithm to convert the image to a transformed image. Based on the center coordinates and acquisition range of each acquired information, at least one acquired component image is obtained. Finally, the image processing server provides at least one acquired component image to an automated optical inspection machine for subsequent inspection.

[0011] The system and method disclosed in this invention are as described above. The image processing server uses an object recognition model to identify electronic components in the image to be processed, in order to find the four unique electronic components in the circuit board and calculate the image coordinates of the center of the four electronic components. Then, the image coordinates of the center of the components and the standard coordinates of the centers of nearby components are grouped into four coordinate groups. A coordinate transformation algorithm is used to calculate four sets of coordinate transformation coefficients. The maximum and minimum values ​​of each coordinate transformation coefficient group are removed and then averaged to calculate the coordinate transformation coefficient. The image to be processed is converted into a transformed image using the coordinate transformation algorithm of the coordinate transformation coefficients. Then, at least one image of the acquired component is obtained based on the center coordinates and acquisition range of each acquired information.

[0012] Through the above-described technical means, the present invention can achieve the technical effect of providing an image of the acquisition element within the correct range. Attached Figure Description

[0013] Figure 1 The diagram illustrates a system block diagram of an automatic optical inspection component image acquisition system according to the present invention.

[0014] Figure 2The illustration is a schematic diagram of the image to be processed for the acquisition of component images applicable to automatic optical inspection according to the present invention.

[0015] Figure 3A The illustration shows a schematic diagram of an electronic component for identification using the automatic optical inspection of components according to the present invention.

[0016] Figure 3B The illustration shows a standard electronic component for image acquisition using automatic optical inspection, as described in this invention.

[0017] Figure 4A as well as Figure 4B The diagram illustrates a flowchart of the component image acquisition method applicable to automatic optical inspection according to the present invention.

[0018] The annotations in the attached figures are explained as follows:

[0019] 10: Automatic optical inspection machine

[0020] 20: Image processing server

[0021] 21: Standard Coordinate Database

[0022] 22: Image Acquisition Database

[0023] 23: Query Module

[0024] 24: Component Identification Module

[0025] 25: Conversion Calculation Module

[0026] 26: Component Image Acquisition Module

[0027] 30: Circuit board

[0028] 31: First electronic component

[0029] 32: Second electronic component

[0030] 33: Third electronic component

[0031] 34: Fourth electronic component

[0032] 35: First Standard Electronic Components

[0033] 36: Second Standard Electronic Components

[0034] 37: Third Standard Electronic Components

[0035] 38: Fourth Standard Electronic Components

[0036] 40: Image to be processed

[0037] Step 501: The automated optical inspection machine acquires an image to be processed containing the complete circuit board, and provides the image to be processed and the circuit board model to the image processing server.

[0038] Step 502: The image processing server has a standard coordinate database and an image acquisition database.

[0039] Step 503: Standard coordinate database corresponds to the storage board model and the standard coordinates of the four component centers.

[0040] Step 504: Obtain the image from the database corresponding to the storage board model and multiple acquisition information, each of which includes the acquisition center coordinates and the acquisition range.

[0041] Step 505: The image processing server retrieves the four corresponding standard coordinates from the standard coordinate database based on the motherboard model, and retrieves the corresponding acquisition information from the image acquisition database based on the motherboard model.

[0042] Step 506: The image processing server uses an object recognition model to identify the electronic components in the image to be processed, in order to find the four unique electronic components on the circuit board and calculate the image coordinates of the center of the four electronic components. Then, the image coordinates of the center of the components are grouped with the standard coordinates of the centers of nearby components into four coordinate groups.

[0043] Step 507: The image processing server selects three coordinate groups from the four coordinate groups, uses a coordinate transformation algorithm to calculate four sets of coordinate transformation coefficients, and then averages the results of each set of coordinate transformation coefficients after removing the maximum and minimum values ​​to obtain the final coordinate transformation coefficient.

[0044] Step 508: The image processing server uses a coordinate transformation algorithm with coordinate transformation coefficients to convert the image to be processed into a transformed image, and then obtains at least one image of an acquisition element based on the center coordinates and acquisition range in each acquisition information.

[0045] Step 509: The image processing server provides at least one image of the acquired component to the automated optical inspection machine for subsequent inspection. Detailed Implementation

[0046] The following will describe in detail the implementation of the present invention with reference to the accompanying drawings and embodiments, thereby enabling a full understanding of how the present invention uses technical means to solve technical problems and achieve technical effects, and allowing for its implementation.

[0047] The following section will first describe the component image acquisition system for automated optical inspection disclosed in this invention, and please refer to [reference needed]. Figure 1 As shown, Figure 1 The diagram illustrates a system block diagram of an automatic optical inspection component image acquisition system according to the present invention.

[0048] The component image acquisition system disclosed in this invention is applicable to automated optical inspection and includes: an automated optical inspection (AOI) machine 10 and an image processing server 20. The image processing server 20 further includes: a standard coordinate database 21, an image acquisition database 22, a query module 23, a component identification module 24, a conversion calculation module 25, and a component image acquisition module 26.

[0049] The automated optical inspection machine 10 is a piece of equipment used in the electronics manufacturing industry. It is primarily used to inspect the appearance defects and manufacturing quality of printed circuit boards (PCBs), semiconductor components, or other industrial products. The automated optical inspection machine 10 acquires a processing image 40 containing the complete circuit board 30. Please refer to the diagram of the processing image 40. Figure 2 As shown, Figure 2 The illustration is a schematic diagram of the image to be processed for the acquisition of component images applicable to automatic optical inspection according to the present invention.

[0050] When the automatic optical inspection machine 10 inspects the circuit board 30, the circuit board model of the circuit board 30 can be pre-input by an external device. The automatic optical inspection machine 10 obtains the circuit board model of the circuit board 30 by scanning the barcode on the circuit board 30, or by recognizing the circuit board model at a designated position on the circuit board 30 through optical character recognition (OCR) technology. This is only an example and is not intended to limit the application scope of the present invention. The automatic optical inspection machine 10 can then provide the image to be processed 40 and the circuit board model to the image processing server 20.

[0051] The image processing server 20 pre-establishes a standard coordinate database 21 and an image acquisition database 22. The standard coordinate database 21 corresponds to the storage board model and the standard coordinates of the four component centers. The image acquisition database 22 corresponds to the storage board model and multiple acquisition information. Each acquisition information includes the acquisition center coordinates and the acquisition range. The aforementioned standard coordinates of the four component centers are based on the standard image of the board model (for example, the center point of the board is located at the center point of the standard image, and the board has no deflection or offset; this is only an example and does not limit the application scope of the present invention). Then, the four electronic components that are uniquely present on the board are selected. The component center coordinates of the four selected electronic components are the component center standard coordinates. The acquisition center coordinates are also based on the standard image of the board model to select the electronic components to be detected. The center coordinates of the selected electronic components are used as the acquisition center coordinates. Then, the range containing the electronic components is selected as the acquisition range. It is worth noting that the acquisition range can be a rectangle, circle, polygon, etc., which are geometric shapes; this is only an example and does not limit the application scope of the present invention.

[0052] When the image processing server 20 obtains the image 40 to be processed and the board model from the automatic optical inspection machine 10, the query module 23 can query the standard coordinates corresponding to the board model from the standard coordinate database 21 based on the board model, and query the acquisition information corresponding to the board model from the image acquisition database 22 based on the board model.

[0053] Next, the component recognition module 24 uses an object recognition model to identify the electronic components in the image 40 to be processed, in order to find the four unique electronic components in the board 30 and calculate the image coordinates of the center of the four electronic components. Then, the image coordinates of the center of the components are grouped with the standard coordinates of the center of the components that are close to each other into four groups. The aforementioned object recognition model is, for example, YOLO (You Only Look Once) model, Faster R-CNN model, SSD (Single Shot MultiBox Detector, SSD) model, etc., which are only examples here and are not intended to limit the application scope of the present invention. The determination of whether the image coordinates of the center of the components are close to the standard coordinates of the center of the components is as follows: first, calculate the coordinate distance from the image coordinates of the center of the components to the standard coordinates of the center of the components. When the coordinate distance is less than or equal to the distance threshold, the image coordinates of the center of the components are close to the standard coordinates of the center of the components. When the coordinate distance is greater than the distance threshold, the image coordinates of the center of the components are not close to the standard coordinates of the center of the components. This is only an example here and is not intended to limit the application scope of the present invention.

[0054] Please also refer to Figure 3A as well as Figure 3B As shown, Figure 3AThe illustration shows a schematic diagram of an electronic component for identification using automatic optical inspection of the component image acquisition method according to the present invention. Figure 3B The illustration shows a standard electronic component for image acquisition using automatic optical inspection, as described in this invention.

[0055] Figure 3A The system displays the first electronic component 31, the second electronic component 32, the third electronic component 33, and the fourth electronic component 34 identified by the component identification module 24. Figure 3B The system presents a first standard electronic component 35, a second standard electronic component 36, a third standard electronic component 37, and a fourth standard electronic component 38. The coordinate distances from the image coordinates of the center of the first electronic component 31 to the standard coordinates of the center of the first standard electronic component 35 are less than or equal to a distance threshold; the coordinate distances from the image coordinates of the center of the second electronic component 32 to the standard coordinates of the center of the second standard electronic component 36 are less than or equal to a distance threshold; the coordinate distances from the image coordinates of the center of the third electronic component 33 to the standard coordinates of the center of the third standard electronic component 37 are less than or equal to a distance threshold; and the coordinate distances from the image coordinates of the center of the fourth standard electronic component 34 to the standard coordinates of the center of the center of the fourth standard electronic component 38 are less than or equal to a distance threshold. Conversely, the coordinate distances from the image coordinates of the center of the first electronic component 31 to the standard coordinates of the center of the second standard electronic component 36 are greater than a distance threshold; and the coordinate distances from the image coordinates of the center of the first electronic component 31 to the standard coordinates of the center of the third standard electronic component 37 are also less than or equal to a distance threshold. If the distance between the components is greater than the distance threshold, the coordinate distance from the center image coordinates of the second electronic component 32 and the third standard electronic component 37 to the standard center coordinates is greater than the distance threshold. Similarly, the coordinate distance from the center image coordinates of the third electronic component 33 and the fourth standard electronic component 38 to the standard center coordinates is greater than the distance threshold. The coordinate distances from the center image coordinates of the other electronic components to the standard center coordinates of the standard electronic components can be deduced in the same way, and will not be elaborated here. Therefore, the center image coordinates of the first electronic component 31 and the standard center coordinates of the first standard electronic component 35 are in the first coordinate group, the center image coordinates of the second electronic component 32 and the standard center coordinates of the second standard electronic component 36 are in the second coordinate group, the center image coordinates of the third electronic component 33 and the standard center coordinates of the third standard electronic component 37 are in the third coordinate group, and the center image coordinates of the fourth electronic component 34 and the standard center coordinates of the fourth standard electronic component 38 are in the fourth coordinate group.

[0056] Next, the transformation calculation module 25 of the image processing server 20 selects three coordinate groups from the four coordinate groups, uses a coordinate transformation algorithm to calculate four sets of coordinate transformation coefficients, removes the maximum and minimum values ​​from each coordinate transformation coefficient set, and then averages them to calculate the coordinate transformation coefficients. The aforementioned coordinate transformation algorithm is shown below:

[0057]

[0058] Where θ is the deflection angle of the board 30 in the image to be processed 40, s is the scaling factor of the board 30 in the image to be processed 40, dx is the offset of the center position of the board 30 in the image to be processed from the center position of the image 40 on the x-axis, dy is the offset of the center position of the board 30 in the image to be processed from the center position of the image 40 on the y-axis, (x1, y1), (x2, y2) and (x3, y3) are the image coordinates of the center of the component in each coordinate group, and (x′1, y′1), (x′2, y′2) and (x′3, y′3) are the standard coordinates of the center of the component in each coordinate group.

[0059] Specifically, the first coordinate group, the second coordinate group, and the third coordinate group are selected and the coordinate transformation algorithm is used to calculate the first coordinate transformation coefficient group; the second coordinate group, the third coordinate group, and the fourth coordinate group are selected and the coordinate transformation algorithm is used to calculate the second coordinate transformation coefficient group; the third coordinate group, the fourth coordinate group, and the first coordinate group are selected and the coordinate transformation algorithm is used to calculate the third coordinate transformation coefficient group; and the first coordinate group, the second coordinate group, and the fourth coordinate group are selected and the coordinate transformation algorithm is used to calculate the fourth coordinate transformation coefficient group.

[0060] The deflection angle θ of the machine board 30 in the image to be processed 40 is at its maximum value in the first coordinate transformation coefficient group and at its minimum value in the second coordinate transformation coefficient group. This is calculated by averaging the deflection angles θ of the second and third coordinate transformation coefficient groups. Similarly, the scaling factor s of the machine board 30 in the image to be processed 40 is at its maximum value in the third coordinate transformation coefficient group and at its minimum value in the first coordinate transformation coefficient group. This is calculated by averaging the scaling factors s of the second and third coordinate transformation coefficient groups. The x-axis offset between the center position of the machine board 30 and the center position of the image to be processed 40 is also considered. The value is at its maximum in the fourth coordinate transformation coefficient group and at its minimum in the second coordinate transformation coefficient group. That is, the x-axis offset of the coordinate transformation coefficient is calculated by averaging the x-axis offset of the first coordinate transformation coefficient group and the x-axis offset of the third coordinate transformation coefficient group. Also, the y-axis offset of the center position of the machine board 30 in the image to be processed from the center position of the image to be processed 40 is at its maximum in the third coordinate transformation coefficient group and at its minimum in the first coordinate transformation coefficient group. That is, the y-axis offset of the coordinate transformation coefficient is calculated by averaging the y-axis offset of the second coordinate transformation coefficient group and the y-axis offset of the fourth coordinate transformation coefficient group. This is only an example and is not intended to limit the scope of application of the present invention.

[0061] Next, the component image acquisition module 26 uses a coordinate transformation algorithm with coordinate transformation coefficients to convert the image to be processed 40 into a transformed image. Then, based on the center coordinates and acquisition range of each acquisition information, at least one acquired component image is acquired. The image processing server 20 then provides at least one acquired component image to the automatic optical inspection machine 10. The automatic optical inspection machine 10 acquires at least one acquired component image for subsequent inspection.

[0062] Next, the operation method of the present invention will be described below, and please refer to the following: Figure 4A as well as Figure 4B As shown, Figure 4A as well as Figure 4B The diagram illustrates a flowchart of the component image acquisition method applicable to automatic optical inspection according to the present invention.

[0063] The component image acquisition method for automated optical inspection disclosed in this invention includes the following steps:

[0064] First, the automated optical inspection machine acquires an image of the complete circuit board to be processed, providing the image and the circuit board model to the image processing server (step 501). Next, the image processing server has a standard coordinate database and an image acquisition database (step 502). Then, the standard coordinate database stores the circuit board model and the standard coordinates of the four component centers (step 503). Next, the image acquisition database stores the circuit board model and multiple acquisition information entries, each containing the acquisition center coordinates and the acquisition range (step 504). Next, the image processing server retrieves the four corresponding standard coordinates from the standard coordinate database based on the circuit board model, and retrieves the corresponding acquisition information from the image acquisition database based on the circuit board model (step 505). Finally, the image processing server uses an object recognition model to identify the electronic components in the image to be processed, in order to find the unique component in the circuit board. The image processing server calculates the center image coordinates of the four electronic components and groups them into four coordinate groups with the standard coordinates of the centers of nearby components (step 506). Next, the image processing server selects three coordinate groups from the four coordinate groups and uses a coordinate transformation algorithm to calculate four sets of coordinate transformation coefficients. The coordinate transformation coefficients are then averaged after removing the maximum and minimum values ​​of each coordinate transformation coefficient (step 507). Next, the image processing server uses the coordinate transformation algorithm of the coordinate transformation coefficients to convert the image to be processed into a transformed image. Then, it acquires at least one image of an acquired component based on the center coordinates and acquisition range of each acquired information (step 508). Finally, the image processing server provides at least one image of an acquired component to an automated optical inspection machine for subsequent inspection (step 509).

[0065] In summary, the image processing server uses an object recognition model to identify electronic components in the image to be processed, in order to find the four unique electronic components on the circuit board and calculate the image coordinates of the center of each of the four electronic components. Then, the image coordinates of the center of each component and the standard coordinates of the centers of nearby components are grouped into four coordinate groups. A coordinate transformation algorithm is used to calculate four sets of coordinate transformation coefficients. The maximum and minimum values ​​of each coordinate transformation coefficient group are removed and then averaged to calculate the coordinate transformation coefficient. The image to be processed is converted into a transformed image using the coordinate transformation algorithm of the coordinate transformation coefficients. Finally, at least one image of the acquired component is obtained based on the center coordinates and acquisition range of each acquired information.

[0066] This technology can solve the problem of incorrect analysis results caused by the offset or deflection of the board when acquiring high-resolution images at a fixed position, thus achieving the technical effect of acquiring images of components within the correct range.

[0067] While the embodiments disclosed in this invention are as described above, the content is not intended to directly limit the scope of patent protection for this invention. Anyone skilled in the art to which this invention pertains may make minor modifications to the form and details of the implementation without departing from the spirit and scope disclosed herein. The scope of patent protection for this invention shall still be determined by the scope defined in the appended claims.

Claims

1. A component image acquisition system suitable for automated optical inspection, characterized in that, Include: An automated optical inspection machine acquires an image to be processed containing the complete circuit board, provides the image to be processed and the circuit board model, and obtains at least one image of the acquisition element for subsequent inspection; and An image processing server acquires the image to be processed and the circuit board model from the automated optical inspection machine, and provides the at least one image of the acquisition element to the automated optical inspection machine. The image processing server further comprises: A standard coordinate database stores the circuit board model and the standard coordinates of the centers of the four components. The image acquisition database stores the circuit board model and multiple acquisition information, each of which includes the acquisition center coordinates and the acquisition range. The query module retrieves four corresponding standard coordinates from the standard coordinate database based on the board model, and retrieves the corresponding acquisition information from the image acquisition database based on the board model. The component recognition module uses an object recognition model to identify the electronic components in the image to be processed, so as to find the four unique electronic components in the circuit board and calculate the image coordinates of the center of the four electronic components. Then, the image coordinates of the center of the components and the standard coordinates of the centers of nearby components are grouped into four coordinate groups. The transformation calculation module selects three coordinate groups from four coordinate groups, uses a coordinate transformation algorithm to calculate four sets of coordinate transformation coefficients, removes the maximum and minimum values ​​from each coordinate transformation coefficient group, and then averages them to calculate the coordinate transformation coefficients. and The component image acquisition module converts the image to be processed into a transformed image using a coordinate transformation algorithm with coordinate transformation coefficients, and then acquires the at least one acquired component image based on the acquisition center coordinates and the acquisition range in each acquisition information.

2. The component image acquisition system applicable to automatic optical inspection as described in claim 1, characterized in that, The formula for the coordinate transformation algorithm is as follows: Wherein, θ is the deflection angle of the board in the image to be processed, s is the scaling factor of the board in the image to be processed, dx is the x-axis offset of the center position of the board in the image to be processed from the center position of the image to be processed, dy is the y-axis offset of the center position of the board in the image to be processed from the center position of the image to be processed, (x1, y1), (x2, y2) and (x3, y3) are the image coordinates of the center of the component in each coordinate group, and (x′1, y′1), (x′2, y′2) and (x′3, y′3) are the standard coordinates of the center of the component in each coordinate group.

3. The component image acquisition system applicable to automatic optical inspection as described in claim 1, characterized in that, The component identification module groups the component center image coordinates with the standard coordinates of similar component centers, calculates the coordinate distance between the component center image coordinates and the standard coordinates of the component center, and when the coordinate distance is less than or equal to a distance threshold, the component center image coordinates and the standard coordinates of the component center are considered similar coordinates, so that the component center image coordinates and the standard coordinates of the component center are grouped into the same coordinate group.

4. The component image acquisition system applicable to automatic optical inspection as described in claim 1, characterized in that, The object recognition model includes the YOLO model, the Faster R-CNN model, and the SSD model.

5. The component image acquisition system applicable to automatic optical inspection as described in claim 1, characterized in that, The automatic optical inspection machine receives the board model number from an external device. The automatic optical inspection machine obtains the board model number by scanning the barcode on the board, or by using optical character recognition technology to identify the board model number at a designated location on the board.

6. A method for acquiring component images suitable for automated optical inspection, characterized in that, Includes the following steps: An automated optical inspection machine acquires an image to be processed containing the complete circuit board, and provides the image to be processed and the circuit board model to an image processing server; The image processing server has a standard coordinate database and an image acquisition database; The standard coordinate database stores the circuit board model and the standard coordinates of the centers of the four components. The image acquisition database stores the motherboard model and multiple acquisition information, each acquisition information including the acquisition center coordinates and the acquisition range; The image processing server retrieves four corresponding standard coordinates from the standard coordinate database based on the motherboard model, and retrieves the corresponding acquisition information from the image acquisition database based on the motherboard model. The image processing server uses an object recognition model to identify the electronic components in the image to be processed, so as to find the four unique electronic components in the circuit board and calculate the image coordinates of the center of the four electronic components. Then, the image coordinates of the center of the components and the standard coordinates of the centers of nearby components are grouped into four coordinate groups. The image processing server selects three coordinate groups from the four coordinate groups, uses a coordinate transformation algorithm to calculate four coordinate transformation coefficient groups, removes the maximum and minimum values ​​from each coordinate transformation coefficient group, and then averages them to calculate the coordinate transformation coefficient. The image processing server converts the image to be processed into a transformed image using a coordinate transformation algorithm with coordinate transformation coefficients, and then obtains at least one image of an acquisition element based on the acquisition center coordinates and the acquisition range in each acquisition information. and The image processing server provides the at least one image of the acquired element to the automated optical inspection machine for subsequent inspection.

7. The component image acquisition method applicable to automatic optical inspection as described in claim 6, characterized in that, The formula for the coordinate transformation algorithm is as follows: Wherein, θ is the deflection angle of the board in the image to be processed, s is the scaling factor of the board in the image to be processed, dx is the x-axis offset of the center position of the board in the image to be processed from the center position of the image to be processed, dy is the y-axis offset of the center position of the board in the image to be processed from the center position of the image to be processed, (x1, y1), (x2, y2) and (x3, y3) are the image coordinates of the center of the component in each coordinate group, and (x′1, y′1), (x′2, y′2) and (x′3, y′3) are the standard coordinates of the center of the component in each coordinate group.

8. The component image acquisition method applicable to automatic optical inspection as described in claim 6, characterized in that, The image processing server groups the image coordinates of the component center with the standard coordinates of the component center, calculates the coordinate distance between the image coordinates of the component center and the standard coordinates of the component center, and when the coordinate distance is less than or equal to a distance threshold, the image coordinates of the component center and the standard coordinates of the component center are considered to be similar coordinates, so that the image coordinates of the component center and the standard coordinates of the component center are grouped into the same coordinate group.

9. The component image acquisition method applicable to automatic optical inspection as described in claim 6, characterized in that, The object recognition model includes the YOLO model, the Faster R-CNN model, and the SSD model.

10. The component image acquisition method applicable to automatic optical inspection as described in claim 6, characterized in that, The automatic optical inspection machine receives the board model number from an external device. The automatic optical inspection machine obtains the board model number by scanning the barcode on the board, or by using optical character recognition technology to identify the board model number at a designated location on the board.

Images