SAR图像的二维灰熵模型快速分割方法

Fast SAR image segmentation method based on the two-dimensional grey entropy model

为提高图像分割速度和抵抗噪声的能力,综合利用小波分析、遗传算法、图像熵和灰色理论,提出一种基于二维灰熵模型的快速SAR图像分割方法.该方法首先对待分割图像进行小波变换,将表征图像概貌特征的低频信息重构为概貌图像,表征图像细节和边缘的部分高频信息重构为梯度图像,然后构造两者的概貌-梯度共生矩阵模型,根据最大熵原理设计二维灰色熵模型作为遗传算法的适应度函数.最后,利用遗传算法高效、并行的寻优能力,通过选择、交叉和变异等遗传操作快速逼近最佳阈值.实验表明,该方法不仅在图像分割过程中能够滤除SAR图像中的噪声,而且分割速度明显加快.

In order to speed up the segmentation procedure and solve the problem of noise-sensibility in SAR image segmentation, the paper suggests a fast SAR image segmentation method based on the 2D grey entropy model, which integrates the wavelet transform, Genetic Algorithm （GA）, image entropy and grey theory. In the method, after an approximation image and a gradient image are deduced from the original image respectively via the wavelet transform, their concurrence matrix is constructed. On the basis of the matrix, a 2D grey entropy based fitness function is designed for GA. And then, after the operations of selection, crossover and mutation, the best threshold is obtained. Finally, our experimental results indicate that the method not only ignores the disturbance of inherent speckle in the SAR image, hut also shortens the segmenting time obviously.