基于多维云空间的多光谱遥感影像边缘检测研究

Study on edge detection of multi-spectrum remote sensing image based on multi-dimension cloud space

在云模型的基础之上,提出一种基于多维云空间的遥感影像边缘检测算法。该方法依据矢量角相似性准则并结合邻域关系进行图像区域生长,在此基础之上根据影像不同波段的数据特征建立多维云空间映射模型,将数据从图像空间映射到云空间。通过多维云的逻辑运算生成边界云并进行向量综合。利用云的数字特征反推隶属度,构建边缘模糊特征平面,并通过最大模糊熵原则确定最优阈值,对图像模糊边界进行提取。试验结果表明,该算法在多光谱遥感影像中能取得较好检测效果。

A new edge detection algorithm of multi-spectrum remote sensing image was proposed based on cloud model.First, image region were gotten by region growing algorithm based on vector angle similarity law and spatial-contextual information.Second, multi-dimension cloud space mapping model based on features of every band could be constructed, which mapping the image objects to multiple cloud space.Third, by Boolean calculation between two or more clouds and synthesizing multi-dimension vectors, boundary-clouds could be constructed.Forth, fuzzy property facet and optimal threshold were constructed based on conditional probabilities and maximizing the entropy of fuzzy partition.At last, by fuzzy property facet and optimal threshold, the edge of multi-spectrum image was gotten.Experiments testified that the algorithm could effectively detect edge of multi-spectrum remote sensing image.