基于模糊分析的多光谱遥感图像的纹理特征

来源期刊：中南大学学报(自然科学版)2004年第2期

论文作者：林剑 鲍光淑 林强

文章页码：281 - 284

关键词：多光谱遥感图像;纹理;不确定性;光谱

Key words：multi-spectral remote sensing; texture; uncertainty; spectrum

摘    要：模糊分析的方法是用均一表面不确定性对原始图像进行模糊纹理滤波,在滤波图像上计算空间均一不确定性,对不确定性进行模糊纹理光谱分析,其光谱曲线直观地反映了多光谱遥感图像的纹理特征;通过采用不同的测量窗口,在不同的类别提取纹理样品进行实验。研究结果表明:多光谱遥感图像在小区域纹理特征不稳定,不同波段的纹理特征不同,不同类别的最小测量区域不同;模糊纹理分析的方法可用于图像分割。

Abstract: This paper summarizes the general algorithms of describing texture feature of multi-spectral remote sensing, and pointed out their problems. Fuzzy filtering with uncertainty of the uniform surface is applied to the algorithm, special uniform uncertainty is calculated in the filtered image, and fuzzy uncertainty texture spectrum is analyzed. The spectrum shows texture features of multi-spectral remote sensing. Experiments with multi-window and multi-class indicate that the feature in the little region is uncertain, different spectral bands have different texture features, the least measure region is different, and the fuzzy analysis algorithm can be used to segment multi-spectral remote sensing images by texture feature.