Relay Appearance Detection Technology Based On Image Processing

In the field of industrial products, the traditional artificial method detection exists a series of problems such as high labor intensity, low work efficiency and high cost. These questions can be well improved by the introduction of computer image processing technology. But in terms of automobile relay products, image detection technology is not yet universal. Relay appearance is the identity of the product labeling. It can be used to identify the classification of the relay. So the detection process is important and essential before products manufactured.Based on the research of the latest research results and progress in the field of product detecting at home and abroad. By analysis of image detection system of different applications. We can use to detect the appearance of relays. The main research content of this system includes three parts. That are image preprocessing, edge detection and image recognition.(a) Image preprocessing. Due to different environments, the appearance of the image may be captured with noise and uneven illumination, etc. Hence We need to do some pre-processing for the image data. It can be done with noise removal filter, gray balance and image enhancement. Thereby the image quality of the relay appearance can be improved and become clearer. Finally, the image can be converted to only black and white binary image through binarization. After that, it can be easier to detecting feature information.(b) Edge detection. On purpose to extract the product labeling contour of relay appearance. On the basis of Canny detection method, the noise impact can be eliminated by the use of data preprocessing. The false edges can be removed while the true contours of rough edges can be retain. And the contour lines can be thick and clear.(c) Image recognition. In order to identify the relay product model. With the use of mathematical morphology processing. The product labeling area can be found. And then the rectangle contour can be searched and draw from out. As follow, the product labeling area can be divided and saved. Finally, the product labeling can be matched and identified based on SIFT feature point matching algorithm.In this paper, on the basis of simple and efficient code in the Open CV library. Visual Studio 2010 under Windows system can be chosen to be the platform of software development. Through experimental tests, it includes better recognition effect. So it can satisfy the requirements of the relay appearance detection system. For the following other products detecting, it can provide a certain reference.