纹理特征向量与最大化熵法相结合的SAR影像非监督变化检测

摘要

Generally,spatial-contextual information would be used in change detection because there is significant speckle noise in synthetic aperture radar (SAR)images.In this paper,using the rich texture information of SAR images,an unsupervised change detection approach to high-resolution SAR images based on texture feature vector and maximum entropy principle is proposed.The difference image is generated by using the 32-dimensional texture feature vector of gray-level co-occurrence matrix (GLCM). And the automatic threshold is obtained by maximum entropy principle.In this method,the appropriate window size to change detection is 11 × 11 according to the regression analysis of window size and precision index.The experimental results show that the proposed approach is better could both reduce the influence of speckle noise and improve the detection accuracy of high-resolution SAR image effectively;and it is better than Markov random field.%合成孔径雷达(SAR)影像具有明显的斑点噪声,在变化检测中,一般需要考虑空间邻域信息.本文结合SAR影像丰富的纹理信息,提出一种考虑空间邻域信息的高分辨率SAR影像非监督变化检测方法,用基于灰度共生矩阵(GLCM)的32维纹理特征向量构造差异影像.通过最大化熵法自动选取阈值,对精度指标随窗口大小的变化进行回归分析,得到适合于变化检测的窗口为11×11.试验表明,本文方法优于马尔科夫随机场法,可以减小斑点噪声的影响,有效提高高分辨率SAR影像变化检测的精度.