SAR Image De-noising And Segmentation

With the rapid development of satellite astronomy technique, Synthetic aperture radar (SAR) is widely applied, and more and more people pay attention to the SAR image processing. In the thesis, new methods on noise-smoothing and edge detection is put forward for the contradiction between de-noising and edge detection. Because Wavelet transform has good localizing quality at time domain and frequency domain simultaneously, and has the characteristic of multi-resolution analysis, so it can do well in the image processing. Therefore, wavelet transform is the means of the SAR image processing in the thesis.Because of the backscatters, SAR images are contaminated by speckle noises which lower image quality and mask image structure. Therefore, noise-smoothing is the first step in the image processing. To solve the problem of losing part of edge in the SAR image de-noising, the edge of image can be preserves beforehand. In the new method, edge points are distinguished from others based on modulus maxima, then based on wavelet transform to de-noise image points except the edge points. The method removes speckle noises of SAR image in effect, at the same time remaining the edge better.Image segmentation, as the key step from image processing to image analysis, lays the foundation for target recognition, this paper studies image segmentation technique. At the regional segmentation, the small areas are eliminated from segmentation results of the maximum variance method; at the edge detection, the method based on stationary wavelet transform, seeks modulus maxima by multiplying adjacent wavelet modulus. The edge judged from modulus maxima not only suppresses the speckle noise but also improves the veracity of edge position. The results prove the good performance of the proposed new method.