Chip is the basic component of circuit, has a wide range of applications. There are many shortcomings in the traditional manual inspection, so the visual inspection size measurement technology arises at the historic moment. Visual inspection has its own characteristics of high sensitivity, high precision and high durability, which has a great help to improve the level of industrial automation and industrial production efficiency. According to the requirements of the subject, with the single chip microcomputer chip as the research object, to detect the size, appearance, defects, quality, length, width, diameter, height, precision, missing material, characters, scratches, pins, circularity, burr on the two-dimensional plane of the single chip microcomputer chip as the research object, design the hardware scheme of the single chip chip detection system based on visual detection. The hardware components include the choice of light source and lighting mode, as well as the choice of camera and lens. After the completion of the hardware platform construction, the calibration board used in camera calibration was made and about 30 calibration pictures were taken on the debugging hardware platform.

Using MATLABR2016A as the software processing platform of the system, on the one hand, the MATLAB calibration box is used to calibrate the calibration diagram to do the camera calibration, on the other hand, the image processing code and measurement code for microcontroller chip size measurement are written. Among them, image processing mainly includes image filtering, binarization processing and edge extraction. After the size measurement experiment of the single chip microcomputer chip is completed, the experimental results are compared with the real size. It can be seen that the visual detection-based external size measurement system of electronic components meets the goal set by the subject.

Key words

Chip size, appearance, defect, quality, length, width, diameter, height, accuracy, missing material, character, scratch, pin, roundness, Burr detection

1. Overall system design The equipment required for hardware design of this system includes light source, camera, lens and computer, the platform required for software design is MATLAB R2016A, and the single chip microcomputer chip is selected as the electronic component. The implementation steps of the overall system plan are as follows: First of all, select the type of light source suitable for the measurement of the size of electronic components and the corresponding lighting mode. After completing the lighting design, select the appropriate industrial camera and lens, fix the lighting source, camera and lens on the bracket, and adjust the distance between the camera and the measured object, the magnification of the camera lens and the aperture size. Ensure that the single chip microcomputer chip to be tested can be completely and clearly presented in the computer imaging screen, so that the establishment of the system hardware platform is completed. Subsequently, the grayscale image of the microcontroller chip with high contrast is obtained from the imaging screen as the image to be processed in the software part of the size measurement system [5]. After a series of image processing methods, such as image enhancement, image filtering, binarization, edge detection, etc. are used on the MATLAB software platform, the length and width of the chip are obtained by pixel point calculation.

2. Hardware system design Lighting scheme design is particularly important for the visual system. On the one hand, it can improve the clarity of the acquired images, and on the other hand, reduce the interference of ambient light on the measurement system. By referring to the data and comparing the size of the components actually measured in this topic, the comprehensive performance of LED lamp is the most superior; At the same time, through the actual comparison of various lighting methods in the laboratory environment, considering the lighting effect and operability, this paper finally chose forward lighting, back lighting and ring lighting. Compared with CMOS camera, CCD camera has higher resolution and sensitivity, this system finally adopts large image MER-132-30GM field array CCD camera, it supports horizontal and vertical mirroring, can be directly connected to HALCON, MERLIC, LabVIEW and other third-party software; The lens is M0824-MPW2 lens produced by COMPUTAR Company [6]. Now the mainstream visual processing software are: OpenCV, LabVIEW, HALCON, eVision, HexSight, RVB and MATLAB. This topic chooses the R2016a version of MATLAB as the system software design platform. Compared with other image processing software, MATLAB has low learning cost and integrates many image processing functions. It is suitable for undergraduates to choose as the development software of graduation project. After the hardware is selected, the hardware can be assembled to build a simple platform and then the hardware and software debugging and subsequent work can begin.

This chapter will focus on the image preprocessing and edge extraction process, including image filtering, image binarization and edge detection as well as the alternative edge detection processing methods, at the same time in these several processes to compare the effects of different algorithms, and select the most suitable method for microcontroller chip size measurement. By comparing median filtering, mean filtering and Gaussian filtering, it can be concluded that the weight distribution of each point in the neighborhood of Gaussian filtering is similar to that of normal distribution, so Gaussian filtering has a very good suppression effect on noise obeying normal distribution. Because the image quality of the collected microcontroller chip is good, the three filtering effects are not too obvious, and Gaussian filtering is finally adopted. The filtering effect is shown in Figure 3.

The gray level is reversed to prepare the binary image. For the binarization processing of single chip microcomputer chip, this topic failed to obtain ideal binarization image results after using the maximum inter-class variance method, average value method and histogram threshold method. Due to the small sample size of the image to be tested, the threshold value was finally determined according to the criteria of human experience selection method [7] [8]. Determine whether the threshold is 200.

After the above series of connected domain processing steps are completed, the minimum external rectangle function in MATLAB software can be used to calculate the external rectangle of the chip to be tested, and the coordinates of four corners of the single chip microcomputer can be obtained to lay the foundation for the final size measurement.

4. Electronic component size measurement experiment and analysis with checkerboard calibration method to the single chip image of the size into the real size. The operation of checkerboard calibration method is between the camera self-calibration method and the traditional camera calibration method without the complicated steps of the traditional camera calibration. This system uses MATLAB calibration toolbox to calibrate the camera. The pixel coordinates are converted into physical coordinates by using the internal and external parameters of the camera, and the dimensions of the two lengths and widths of the microcontroller chip are calculated and the average value is obtained. In this experiment, the length and width of four different models of microcontroller chips were measured, and the error calculation was made on the measurement results of the four length widths [9] [10]. Using the two-sided method of size calculation to calculate the average value to reduce the error, but by comparing the actual measurement value of the microcontroller chip with the system measurement value, it is found that there are still slight errors. For microcontroller length and width of the final value of the calculation only rely on the simple bilateral average method, can only limited to reduce the error can not be completely eliminated error.

conclusion

Dimension detection is widely used in industry, and electronic components are the categories of industrial objects to be detected in a large proportion. The size detection system using machine vision technology has replaced the previous manual measurement, saved a lot of manpower and material resources, and added power to the development of industrial detection technology. The measurement object of the size measurement system of electronic components based on visual inspection is the single-chip chip. The system mainly includes camera calibration, image acquisition, image preprocessing and edge extraction, chip length and width measurement and other steps. The focus of the system lies in two parts: image processing and size measurement. After selection, the focus is on the comparative study of image processing methods, including image filtering, image binarization and edge extraction steps. The image filtering is the comparative analysis of Gaussian filtering, mean filtering and median filtering, image binarization is the comparative analysis of histogram threshold method, maximum inter-class variance method and average method, and finally combined with the three methods using the manual empirical selection method to determine the threshold of image segmentation. Using the method of connected domain processing, the image is filled with holes, boundary smoothing and area filtering. Finally, the microcontroller chip size measurement and error calculation and analysis are completed.