Multiscale Edge Detection Method

With the development of society, Image edge as the basis of image identification authentication is more and more important. Wavelet analysis which is a tool of multi-scale analysis focus on the overall outline in large scale and focus on the details in small scale, thus achieve very good results. The main work is as follows:(1)In this paper, the wavelet analysis theory, wavelet transform based on lifting scheme and image edge-extraction basis are described in detail. Experiment analysis on the traditional method of edge detection was done.(2)In this paper, according to the edge extraction algorithms based on point correlation, the edge extraction algorithms based on both interval and spatially correlation are proposed. The methods help to weaken the affect which is aroused by migration in the course of wavelet transform. Experimental results show that both methods are good at the edge detection.(3)In this paper, the advantages and disadvantages between the first and second generation wavelets were printed out. We gave the lifting process of the bi-orthogonal wavelet flitter and apply to the edge detection. Experiments show that the edge detection based on lifting wavelet has the same effect as the traditional methods, but has superiority on computational complexity, computational speed and storage space.