The Research On Sub-Pixel Edge Detection Method In Work Piece Image

Along with the rapid development of automation industrial production,the testing of product appearance quality is based on computer detection. Image detection technique has been a new testing method which based on machine vision has the characteristics include the non-contact detection, high precision and short compute time. It can be meet the requirement of industrial production. The key of improve the efficiency in industrial production is finding a kind of fast and effective detection method.Several kinds of classic edge detection algorithm with pixel level are researched in this paper, including Sobel operator, Prewitt operator, Laplace operator and Canny operator. After that the advantages and disadvantages of edge detection have been analyzed, and improving Canny operator according to it's drawback. And then several sub-pixel edge detection methods were researched. It proposed a new edge detection method as the basic of sub-pixel technology with a high precision.A novel sub-pixel edge detection technique which based on the features of high precision and shorter computing time is proposed in paper. The image must be processed by pretreatment before the edge detection. Canny edge detection operator is used in the pixel level edge detection. In the gray image, a rectangular area by 5×5 pixels around every pixel level edge points is detected, and a second-degree curve is fit using the difference of gray value. So the first derivative of this curve is the coordinates of sub-pixel level edge detection in the image. The experimental results indicate that this method have higher anti-noise performance, and high inspection accuracy. It can be used in the process of real-time detection system.