论文部分内容阅读
H/α-Wishart分类方法是目前常用且较为有效的极化SAR影像分类方法,但其分类精度还有待改善。研究一种基于遗传算法的极化SAR影像的分类方法,该方法根据极化SAR影像Cloude特征分解的特征值,采用H/α平面进行初分类,然后采用遗传算法迭代进行再次分类。针对遗传算法“早熟”和收敛速度慢的问题,结合H/α平面图对遗传算法的变异算子进行了改进,以利用极化散射机理缩小变异范围,改善算法收敛速度。采用NASA-JPL实验室的极化SAR数据以及中国电子科技集团38研究X波段原型样机的高分辨率极化SAR数据进行实验,结果表明:该方法极化SAR影像分类精度优于H/α-Wishart分类方法。
The H / α-Wishart classification method is a commonly used and effective method for polarimetric SAR image classification, but its classification accuracy needs to be improved. A classification method of polarimetric SAR images based on genetic algorithm is studied. According to the eigenvalues of the Cloude eigenvalues of the polarimetric SAR images, this method uses H / α plane to classify the eigenvalues, and then classifies them again by genetic algorithm iteration. Aiming at the problem of genetic algorithm “precocious ” and slow convergence rate, the mutation operator of genetic algorithm is improved by combining H / α plan to reduce the variation range and improve the convergence speed of the algorithm by using polarization scattering mechanism. The experimental results of NASA-JPL polarimetric SAR data and China Electronics Technology Group38 high resolution polarimetric SAR data of X-band prototyping prototype show that the accuracy of this method is better than that of H / α- Wishart classification method.