小波域局部贝叶斯阈值的SAR图像斑点抑制算法

Wavelet domain Bayesian shrinkage despeckling algorithm for SAR image

提出了一种抑制SAR图像斑点噪声的小波域贝叶斯软阈值方法。该算法不同于有偏的去除乘性噪声的同态滤波算法,而是将噪声转化为局部平稳的加性白噪声。在非下采样小波子带上,视数给定时该算法可以简洁有效地估计局部加性噪声方差。在实验中,该算法同Kuan,Lee和Argenti等算法作了比较,结果表明,在常用性能指标上所提算法优于其它算法。

A novel SAR despeckling method using Bayesian shrinkage in local wavelet coefficient domain is presented. Instead of using the homomorphic filter which is used to get rid of multiple noise and is a biased estimator, this algorithm translates speckle into locally stationary additive white noise. In undecimated wavelet subbands, for a given look number this algorithm can concisely and efficiently estimate local additive noise variance. The proposed algorithm is compared with Kuan, Lee and Argenti algorithms in an experiment which resuits show this algorithm is better in common quality index.