Edge detection of high resolution remote sensing imagery using wavelet

In this paper, the 3-level B-spline wavelet transform is applied to extract different types of edges, according to the singularity exponent of edges from high resolution remote sensing imagery. The gradient algorithm, Laplacian algorithm, and Robert algorithm are classic linear algorithms. However these algorithms have an acute function on edges and are vulnerable to noise. In the process of edge detection using wavelets, edge information at different scales is obtained according to the wavelet's multi-scale character. We may integrate multi-scale edge information to create high accuracy and different types of edges with a pixel width. We verify the 3-level B-spline wavelet transform is asymptotically optimum in practical application such as feature extraction. This paper gives a fast algorithm in decomposition, time response and frequency analysis. More importantly, in this paper, the algorithm can also be used to determinate the singularity exponent of edges so as to identify different types of edges. According to the different needs, we can output the different types of edges that serve as a more effective representation on which subsequent localization and recognition tasks are based.