ABSTRACT
According to the current mathematic models the subpixel level edge detection methods can be divided into three kinds mainly: fitting method, interpolation method and moment method. The moment is the most widely used sub-pixel level edge detection method. The Zernike image moment is based on the integral operation so it has stronger anti-noise ability compared with the differential operation ( Sobel and Robers operator etc. ). The moment method mainly including two kinds: spatial moment and Zernike moment. The spatial moment concept was first put forward by Lyvers[1,2] which used six image geometric moments deduced four step edge parameters, this method has redundancy information because lack of the orthogonality. Ghosal and Mehrotal presents an edge detection method based on orthogonal Zernike image moment[3~5]. The domestic scholars also performed related research, paper [6] proposed a height step edge detection which improve the calculation time and sometime lose minimal image detail information.
