基于单亲遗传算法的SAR图像混合像元分解

Decomposition of mixed pixels of SAR remote sensing images based on Partheno-Genetic Algorithm

已有的遥感影像混合像元分解理论方法都要求遥感影像的通道数目大于地物种类,而合成孔径雷达(SAR)的自身特点决定了SAR图像不可能有过多的通道数目,为解决SAR图像地物种类大于通道数目情况下的混合像元分解问题,本文基于单亲遗传算法提出了一种新的混合像元分解方法,创建了一种新的染色体编码方式及进化迭代方式,新算法很好地实现混合像元的分解,可以分解出比通道数目更多的地物种类.并从北京地区ENVISAT-ASAR图像中截取天安门附近区域作为数据源进行实验,实验结果表明了本文算法的正确性和有效性.

The existing theories and methods used for remote sensing images＇ mixed pixel decomposition all require that the number of image channels should be more than that of the ground types, but the characteristics of the Synthetic Aperture Radar （SAR） hinder the SAR image from having excessive channels. To solve the problem in which the SAR images＇ mixed pixel decomposition in circumstance of ground types being more than channels number, we propose a new method for mixed pixels＇ decomposition based on Partheno-Genetic Algorithm. In our algorithm, we establish a new chromosome coding method and an evolution iterative method. The new algorithm can realize the ideal mixed pixel decomposition and decompose more ground types than the number of channels. We intercept the area of the Tiananmen vicinity from the Beijing ENVISAT-ASAR image as our experiment data source and compare the result with the real ground feature. The experiment result shows that our algorithm is correct and effective.