Research On SAR Image Edge Detection Based On Multiscale Statistical Analysis

In the recent years, great progress has been made in Synthetic Aperture Radar(SAR) technology. But according to the characteristics of SAR imaging mechanisms, geometric distortion and a form of multiplicative noise, known as coherent speckle, generally corrupt the resulting images. Thus it is very difficult for traditional methods to process SAR images. Based on statistical characters of SAR imaging, the edge detecting are deeply studied in this dissertation and some valuable results are obtained.The author studied and compared some typical edge detection methods, especially the Touzi Ratio algorithm and Maximal Likelihood(ML) edge detection algorithm. Statistical theory is used to analyze the SAR image data. An adaptive edge detection algorithm is proposed, which can change the edge detection window and threshold automatically according to the local changes of fields. The performance of this algorithm is better than conventional algorithms.According to the transfer characteristic across scales of the wavelet module of signal edge and noise, we proposed an edge enhancement method and combined the property of edges in different scale. A multiscale edge fusion algorithm consisting of edge transfer, edge inherit and edge growth is proposed. The result of experiment shows that this algorithm can get ride of the impact of noise and the edges have precise position and intact contour.