基于灰度共生矩阵的纹理提取方法研究

摘要

With the enhancement of spatial resolution of remote sensing image, texture features play increasingly important role in remote sensing image.In the paper we study nine texture descriptors including gray level co-occurrence matrix ( GLCM) homogeneity, contrast, dissimilarity, angular second moment (asm), correlation, entropy, and gray level difference vector (GLDV) contrast, mean, angular second moment, etc., and find that both GLCM-asm and GLDV-asm are the more effective ways for texture extraction of the high resolution remote sensing image.%随着遥感影像空间分辨率的提高，纹理特征在遥感影像中的作用越来越重要。对灰度共生矩阵的GLCM （灰度共生矩阵）同质性、GLCM对比度、GLCM相异性、GLCM角二阶距、GLCM相关性、GLCM熵、GLDV（归一化灰度）反差、GLDV均值、GLDV角二阶矩等九种纹理描述子进行研究，发现采用GLCM角二阶矩、GLDV角二阶矩这两种纹理描述子对高分辨率遥感影像的纹理提取具有较好效果。