SAR Image Denoising Based On Undecimated Wavelet Packet Decomposition

Due to the own special advantages, Synthetic Aperture Radar (SAR) has a great potential in both military and civil fields. But SAR images, as the results by a coherent imaging system, are much degraded by speckle, which considerably affects object detection and identification in SAR images and SAR image segmentation. Therefore, it is of great importance to suppress the speckle in SAR Images and improve SAR image quality.This thesis is focused on the speckle suppression algorithms based upon undecimated wavelet packet transform (uWPT). The intention of despeckling is to suppress speckle while preserving the details of the image, such as texture and edges. On the basis of fully reviewing various SAR speckle suppression methods, we presented a new speckle suppression algorithm based on uWPT. Because speckle in SAR images is mainly concentrated on the high frequency region, wavelet transform can not provide fine 2D frequency segmentation to represent the details in SAR images efficiently and thus uWPT is used to decompose SAR images. After the decomposition, the lowpass subband image was regarded as the local mean image of the SAR image. Spatially adaptive shrinkage was applied to the rest subband images for speckle suppression. Finally, the despeckled image was recovered from the filtered subband images by the inverse uWPT. The results showed that the proposed algorithm outperforms the Kuan algorithm and the -WMAP algorithm using undecimated wavelet transform in performance.