针对宽基线影像匹配问题,提出基于信息量和空间分布均衡性双重约束的多层次特征筛选方法,并在此基础上探讨集成该特征筛选方法、MSER(Maximally Stable Extremal Region detector)、SIFT(Scale Invariant Scale Transformation)的特征提取算法,以达到提取高质量(高信息量、空间分布均衡、高重复率)仿射不变特征的目的。实验表明,相对现有最优算法,本文方法具有更高的重复率和匹配成功率,更有利于后续的(宽基线)立体匹配。
Wide baseline stereo image matching is a challenging problem in image processing and computer vision. For the purpose of wide baseline stereo image matching, a filter strategy of affine invariant feature based on information content and spatial distribution constraints is put forward. Moreover, a new feature extraction algorithm integrating MSER, SIFT and this filter method is discussed, in order to realize high quality (high information content, good spatial distribution, high repeatability) affine invaria...