Line Extraction in Multiple Images Based on Fusion Edges
She Er · Chinese Journal of Computers · 2005
Linear feature is one of the most essential features that used in computer vision and image understanding. Effective approaches of line extraction have been paid more attentions so far. The techniques of image fusion can effectively extract useful information from a set of input images. A new line extraction algorithm in multiple images based on fusion edges is proposed,It consists of three main aspects: Firstly, edges are extracted from every image. Then, the adjacent edge points from all images are grouped into combined line support regions based on similarity of gradient orientation. Finally, the regions are fitted and the fused lines are generated. This method uses the gradient orientation of multiple images as the fusion element. More complementary information is used, and more effective lines can be extracted. The validity of the method is proved by experiments for real remote sensing images.