Multiscale edge detection in SAR images

Edge detection plays an important role in automatic process of SAR images. Single scale edge detector is difficult to take account of noise suppression, the integrity of edges and the veracity of position. A multiscale edge detect method is proposed. At first, multiscale Gaussian edge detector is constructed. According to transfer characteristic across scales of the wavelet modules of signal edge and noise edge, the property of edges in different scale is combined and a multiscale edge fusion algorithm is proposed consisting of edge transfer, edge inherit and edge growth. The result of experiment shows that this algorithm can get ride of the affect of noise,the edges fused have precise position and intact contour.