Design and research of PCB automatic detection system based on image processing (Part 2)

3. System software design

The software design of the PCB defect automatic detection system based on image processing is of course its core. The system software design mainly realizes the functions of image data acquisition, image data processing, and image data analysis units in the computer system, and also realizes the function of the control unit of the two-dimensional motion platform, and is responsible for interacting with the operator. The system software structure is shown in the figure.

3.1 System Algorithm Flow

The system software process is divided into manual detection and automatic detection. Manual detection requires the operator to click the corresponding operation process on the human-computer interaction interface. Automatic detection can realize one-click automatic detection, directly perform PCB defect detection and obtain results according to the process set with pre-set parameters. The algorithm process is shown in the figure.

Manual detection can directly move the CCD camera to the main part of the PCB board to be tested by controlling the stepper motor movement when collecting images as needed. When performing image processing, the image processing algorithm suitable for it can also be selected according to the image quality to make the system interactive. After the automatic detection initialization parameters are set, defect detection can be achieved with one click to obtain the test results, which reduces the complexity of operation and greatly improves the speed of detection, making the system automatic, simple to operate, and fast. This article combines the two to make the PCB defect automatic detection system more excellent and more practical.

3.2 Defect Detection

The current printed circuit board defect detection methods are mainly divided into three categories: reference comparison method, non-reference comparison method and hybrid method. The reference comparison method compares the image to be tested with the reference image feature by feature; the non-reference comparison method does not require any reference image, but only judges whether there is a defect based on the previously designed rules and standards. If it does not meet the standards, it is considered defective; the hybrid method is a comprehensive application of the reference comparison method and the non-reference comparison method. This article mainly uses the reference comparison method to compare and analyze the PCB image with the standard image to determine whether the PCB board has defects.

3.3 Defect Identification

Typical defects on bare PCB boards in actual production include short circuits, open circuits, protrusions, depressions, voids, etc.

After the defect is acquired, the defect type cannot be determined, and defect identification is required. Defect identification is based on the different features of various defects. Commonly used image features include histogram statistical features, texture features, and binary image features. Because the grayscale levels of PCB images are not rich and the circuit patterns are all geometric patterns, this paper uses binary image features to identify defects. The main defect features for classification and identification of typical defects such as short circuits, open circuits, protrusions, pits, and voids are: (1) The number of connected domains in the defect image is different from that in the standard image; (2) The area of ​​the background connected domain of the defect image is different from that of the standard image; Combining the above features 1 and 2, the open circuit, short circuit, protrusion, void, and pit defects can be identified, as shown in Table 1.

For a defect image with only a single defect, the detection process is as follows:

(1) First, the defect image is segmented by threshold, the obtained binary image is added to the binary image of the standard image and the average is taken to obtain the location of the defect connected domain, and the defect is marked by color coding.

(2) Extract contours of the threshold segmented images of the defect image and the standard image respectively; (3) Calculate the number of connected domains of the contour extracted images of the defect image and the standard image respectively to obtain the number of connected domains; (3) Calculate the area of ​​the background connected domain of the contour extracted images of the defect image and the standard image respectively to obtain the size of the background connected domain area; (4) Determine the defect type according to Table 1, mark the defect according to the defect position obtained in (1) and display the defect image.

3.4 Results Analysis

The reference comparison method is used to identify defects on PCB boards through comparative analysis. First, XOR operation is performed on the PCB board to extract defect features; then binary mathematical morphology processing is performed on it to remove false defects; then two image recognition methods are used: one is to highlight defects through comparative operation and then pseudo-color processing to facilitate manual visual identification of defect types and locations; the other is to perform tree-like hierarchical judgment of defects through the number of target area features, target area area features and closed features of defect boundaries, thereby realizing the automatic identification of common open circuits, short circuits, protrusions, pits and voids on PCB boards.

4. Conclusion

This paper designs an automatic detection system for printed circuit board (PCB) defects based on computer vision and image processing, and verifies its functions. The experimental results show that the system has a friendly interface, simple operation, simple detection method, rapid detection process, and accurate detection results. The system provides a good solution for the detection of PCB defects and has important application value.