假高帽变换的高分影像居民地信息提取

Residential area extraction of high resolution RS imagery based on multiple contour structuring elements with pseudo top-hat transformation

针对高分辨率遥感影像暗背景下亮居民地的提取问题,该文提出了一种基于形态学多轮廓结构元素假高帽变换的提取方法。该方法首先采用腐蚀型边缘检测算子对影像进行预处理,接着利用多轮廓结构元素分别对影像进行假高帽变换,然后对各个处理结果进行加权融合,最后对加权融合后图像区域生长完成轮廓跟踪。实验结果表明,该方法能够提取居民地,特别适合于暗背景下较亮区域居民地的提取,并且提取效率较高,对于遥感影像居民地提取的研究与应用具有一定的参考价值。

In order to solve the problem in the light residential area extraction from dark background of high resolution RS imagery, the paper proposed a extraction method based on morphological multiple con- tour structuring elements with pseudo top-hat transformation： firstly, corrosion edge detection was used to preprocess the RS images; secondly, the images were processed with pseudo top-hat transformation by multiple contour structuring elements; then each processing result was weighted fused; finally, the fused images were processed with region growing and tracking contours. Experimental results showed that the proposed method could extract residential area with high accuracy, particularly suitable for light residen- tial area extraction in dark background, which would provide a reference for residential area extraction of RS imagery.