结合相干斑抑制的小波域SAR图像网格编码量化

本文结合SAR图像相干斑的抑制,研究了一种在小波域对带噪SAR图像做网格编码量化的新方法。首先将SAR图像在小波域内实施软阀值去噪声,然后根据SAR图像在小波域中各子带系数固有的树结构关系对其进行零树分类,对分类后的重要性小波系数进行网格编码量化,利用卷积编码和信号空间扩展来增大量化信号间的欧氏距离,并用维特比算法寻找最优量化序列。该方法综合了小波相干斑抑制、零树编码、网格编码量化技术,不仅利用了信号小波变换域的空间相关性,而且也较好地利用了信号间的时间相关性。在压缩的同时进行了相干斑抑制,在SAR图像的压缩中取得了很好的效果。

TCQ Of SAR Wavelet Image Combined with Speckle Suppression

In this paper, combining with speckle noise suppression, A new method which performs TCQ of noised SAR image in wavelet domain is proposed. First, the wavelet domain soft-thresholding technique is implemented to reduce the speckle noise before compression is performed. And then, the zerotree classification utilizes correlation of inherent tree structure in subbands. At last, TCQ is applied in significant coefficients. Convolutional coding and signal space expanding are used for increasing Euclidian distance between signals. Viterbi algorithm is used to find a optimized survived quantized sequence. This method utilizes not only the correlation in wavelet transform domain, but also time correlation of signal. Image compression combined with speckle reduction. Simulation states this method has advantages in SAR image compression.