Study On The Defect Inspection Of Touch Switch Based On Machine Vision

Tact switches are widely used in audio products, digital products, telecontroller, communication products, household appliances, medical equipment, fitness equipment and other industries which are one of the important electronic components of the electronic information industry. Detection of the tact switch appearance is a very important step because it directly affects the life and reliability of the product. In China it is relied on the workers’ eye examination to take the defects inspection in the production process. But there are problems like more items to be checked, smaller size of production, higher precision request which can easily make the workers visual fatigue and lead to error detection or omission, thus make it hard to improve the speed of subsequent sorting process and the quality. Bringing the machine vision technology to the tact switch appearance detection cam make the process automatic and intelligent and this can make a great practical value and social benefits.This paper takes a certain type of liquid crystal tact switch(sizes of 6.4*5.2*1.5mm) as the research object and designs a tact switch defect detection device based on machine vision technology. Then study the principles and algorithms of detecting defects of touch switch according to the image acquisition scheme of the device. The main contents and results are as follows:(1) Designed an integrated automatic equipment for tact switch charged and appearance testing, which consists of a feeding table, rotary table, motor, MCU controlled board, a mechanical transmission device, the image acquisition device and the collecting box etc.(2) Make a scheme to determine the image acquisition combined with the requirements of laboratory detection, selecting the appropriate camera and light source to acquire the image of the tact switch. To get the binary image of the tact switch, there is a need to take the pretreatment of the acquired image like light compensation, gray processing, filtering, threshold segmentation and morphological closing operation.(3) Presents a new defect recognition method for tact switch. Firstly get the image contour point set using the improved contour tracking method and get the bounding rectangle of the tact switch using rotating calipers algorithm of minimum area of external rectangle. Then take the spatial transform of the image using the fast rotation method and carry on the interpolation process using bilinear interpolation method, extracting the region of ROI(Region Of Interest) after correction. Get the statistical white pixel area of the four pin ROI region and compare it with the standard threshold. Finally select the eligible and defective switch as samples to test the algorithm. The test results show that this method can meet the requirements of the switch appearance detection. The stability and repeatability experiments proved that the measurement system can meet the detecting requirements.(4) Implemented the tact switch defect detection software based on Windows platform using Visual C++ 6.0 development environment, and took the experiments of the detection algorithm related to tact switch and the element image filtering.