A System For Automatic Inspection Of Printed Circuit Board Defects Based On Machine Vision

The defects inspection of printed circuit board(PCB) is a necessary process during PCB assembly, which is done manually all the time. Since people’s ability is limited, the missinspection is incidental and inefficient. With the development of PCB manufacturing technology, PCB going toward larger area, higher density and higher precision. For such kind of PCB, it’s impossible to achieve the goal. In order to meet the increasingly inspection requirements and find PCB defects timely and effective, using machine vision technology to realize PCB defect detection has become the main direction. Most of them are detected by comparing images and standard images to identify PCB defects. However, this method can only identify the existence of defects, and can not determine the categories of defects and the specific affiliation of the defects and the PCB components.To the above problems, this paper is designed about system for automatic inspection of printed circuit board defects based on machine vision. And this system includes image acquisition and image processing, and designed to reach the acquisition and position of image,and the inspection of defects. In view of the system to the PCB flaw detection content and the detection precision specific request, realize the function of each module of the system and build a system of hardware platform and related software, based on the above correlation algorithm research.The main research work and innovation points of the system are summarized as follows:(1) In order to improve the efficiency of the detection of PCB diagonal positioning holes, an improved Hough transform algorithm is proposed. The improved algorithm starts with the round parameter equation, mapping image space to parameter space, then reducing dimensions and accumulator with multidimensional array instead of multiple cycles of voting statistics to get circle radius and calculate the coordinates of circle center. The feasibility and effectiveness of the circle detection algorithm are proved by comparison and analysis of several experiments with standard Hough transform.(2) In the process of extracting the component image and the image of the pad, the paper presents a Region of interest(ROI) extraction algorithm, which is based on the unique Designator, Footprint, Mid_X,Mid_Y, Rotation and Value in the PCB component information document. According to the information of the document, it can automatically determine the image area of the components and parts, which greatly reduces the time of image processing and improve the detection accuracy.(3) In this paper, a method of automatic detection of defects is proposed. This method is introduced to the location and acquisition process of PCB components and components, not only can clear defects, also points out the defects and the PCB components of specific affiliation, and mainly through the following to achieve: At first, the PCB image is corrected by the positioning hole on the PCB diagonal line to solve the problem of different degree of geometric distortion in the image acquisition process. The bottom left corner of the PCB positioning hole center and the PCB element document reference origin agreement;Secondly, analysis of relevant information within the document, locate and get to image and image pad components in the lower left corner of the positioning hole center at the origin;At last, after image processing algorithm for the analysis of the previous step to get the components related to the image, the image can be reflected in the image to reflect the useful information. And compares it with the experience data, thus obtains the test result. Through the PCB component pad welding leakage less tin and wrong of resistance welding defects detection verified the method’s accuracy and efficiency of and for the method provides feasible basis.