光谱影像的灰度矢量相关特征与边缘特征提取

提出了多光谱图像像元特征矢量之间的极小互相关系数、极大互相关系数及互相关差值和相关密度的概念。在对模拟图像进行矢量相关分析及其对边缘提取的有效性分析基础上,提出对多光谱图像采取像元特征矢量相关分析并提取边缘信息的方法,对TM图像进行试验分析的结果说明了此方法的有效性。

Analysis with the Correlation Between Grey Level Vector of Multi-spectral Image and Its Application for Edge Extraction

This paper proposes the concepts of multi_dimensional co_correlation coefficients and density of co_correlation coefficients between pixels of remotly sensed multi_spectral image.The applications of these concepts to simulative images show that the edge information can be extracted.The minimum co_correlation coefficients and the difference of them can provide evidently the information of the position and strength of edges,which implies that the analysis with co_correlation coefficients can be used for edge extraction of multi_spectral image. The further experiments have been performed using TM images.The different results with different value ranges of minimum co_correlation coefficients and the difference of them indicate that we should select the rational threshold of co_correlation coefficients for edge extractions.Finally the paper concludes that the analysis with co_correlation coefficients for edge extraction of remotely sensed multi_spectral images is quite effective.