Speckle Reduction And Data Compression In SAR Images

SAR images get wide application and rapid development because they can be obtained any-time, any-weather and with high resolution. Since each SAR image inevitably contains coherent speckle noise, the relationship between adjacent pixels reduced severely, and speckle noise is the key issue which restricts SAR image compression.Due to the SAR coherent imaging characteristics, the images contain multiplicative speckle noise. Therefore, logarithm the multiplicative model of speckle noise, and the multiplicative noise become additive white Gaussian noise which easy to filter. The paper presents a new non-linear thresholding function, and the shape of the new thresholding function can be changed visually by adjusting parameters. It takes the Donoho hard-soft threshold as two kind of special situations. Simulation experiments confirm that the improved method can remove the noise effectively, and overcome the shortcomings of hard and soft threshold with oscillation and borderline blurred. The visual effects is superior to the traditional method of the soft and hard threshold.Compared with natural images, SAR images have many characteristics of more dynamic range and rich texture detail, etc. On the basis of standard EZW algorithm, through codes the lowest frequency sub-band coefficient, introduces the Lift Wavelet Transform, and improves the process of successive approximation quantization, the paper presents an improved EZW algorithm. The experiment results show that the reconstruction image of the algorithm can still maintain certain texture details, and the visual effect performs better than that of EZW and SPIHT. So the algorithm is an efficient image compression algorithm.