Abstract:

The precision of localizing the traditional edge detection algorithms is very low, and the efficiency of these is very low too, and these algorithms are very sensitive to the noise. It can not meet the demand of industrial production. Because of this, this paper puts forward a subpixel edge detection algorithm based on fitting ——the subpixel edge detection algorithm based on multinomial five times fitting. This algorithm firstly fetches some points near the edge which is waiting for measuring, then tries to get the grey level of these point, and fits to the grey level curve through the multinomial curve five times, and  tries to get the point at which the two steps of derivatives of multinomial five times equals zero, and this point is also the subpixel edge position. We carry out the experiment to the cubic polynomial, and compare this algorithm to the subpixel edge detection algorithms which are partly based on two times difference, LOG operator and space square. The result shows that the proposed algorithm has shorter running time and higher efficiency.

Key words: measurement of the parameter of cutter;edge detection;subpixel