摘要
线特征匹配是实现图像匹配的常用方法。本文提出了一种基于图匹配的直线段匹配方法,该方法利用图来表示图像中提取的直线段集合,图的节点表示直线段,图的边表示直线段之间的关系,并构造了描述向量来表示图的节点和边的属性信息,从而在匹配时充分利用了图像的结构信息。采用仿真图像和实际图像对算法进行了验证,实验结果表明,该方法取得了较好的直线段匹配性能,且具有较强的抗干扰能力。
Line matching is a popular method in image matching. An algorithm for matching lines based on graph matching is presented in this paper. The set of lines extracted from images is represented by a graph, where vertices stand for straight lines and edges depict the relationship between them. Descriptors are constructed and taken as the attributes of vertices and edges. Therefore, the matching of straight lines can better utilize the structure of images. Simulated and real images are used to test the proposed method. Results demonstrate that the proposed method achieves satisfactory matching performance and behaves robust to interferences.
出处
《图像与信号处理》
2017年第1期44-51,共8页
Journal of Image and Signal Processing
基金
国家自然科学基金(编号61401504)
博士后资助面上项目一等资助(2014M562562)。
