Image classification based on beta distribution for SAR image
K. Arai, Yasunori Terayama, Tsuyoshi Arata · 2002
A new method for SAR image classification is proposed. The method is based on maximum likelihood decision rule with texture features and takes into account the probability density function of texture features. The experimental results show the proposed method is superior to the existing maximum likelihood method with multivariate normal distribution. 2.28 to 5.16% of improvements are observed with real SAR image. Effects of local least square estimator, sigma and weighting filters for speckle noise reduction on classification performance are clarified. The results show that 7.1 to 12.04 % of improvements on the classification performance are observed.