Feature Extraction of Bridges over Water in High Resolution Satellite Imagery
Feature extraction plays a critical role in pattern recognition.With respect to a bridge over water in high resolution satellite imagery, the most distinct feature is a line pair. In this paper, a method for extracting line pairs of bridges over water is presented. Firstly, denoising is implemented by mean filtering to an original image and a histogram-based threshold is selected to segment the smoothed image. Then two mathematical morphology operators are used to filer the binary image, and the river direction is obtained from the centroids of two river areas in the binary image. At the same time, an edge image is obtained in which two parallel lines are detected which are most perpendicular to the river.Finally, the extracted features are verified by prior knowledge. Experimental results obtained from real high resolution satellite images show that the proposed method is effective.
Satellite Imagery Bridge Feature Extraction Parallel Pair Mathematical Morphology, Gray Histogram
Aijun Chen Jinzong Li Bing Zhu
Department of Information and Communication Engineering, Harbin Institute of Technology Harbin, Heilongjiang 150001, China
国际会议
杭州
英文
352-356
2006-10-12（万方平台首次上网日期，不代表论文的发表时间）