融合边缘化外点处理的改进图像匹配方法

An improved image matching method combined with edge processing of outer points

针对图像匹配中SURF(speed up robust features)算法匹配效率不佳,以及RANSAC算法迭代次数不稳定和人为设置内点离散阈值所带来误差的问题,提出了一种结合改进的边缘化采样一致性算法和改进SURF的图像匹配方法。首先对输入图像进行快速引导滤波预处理,过滤图像噪声并保留边缘细节信息。然后通过BEBLID(Boosted Efficient Binary Local Image Descriptor)算法为SURF构建高效的二值描述符,结合改进的边缘化采样一致性算法边缘化外点去除误匹配。经实验对比,该方法相较于SURF准确性更高,实时性有较大提升,可满足多数复杂环境下的图像匹配。

In order to solve the problems of inefficient matching of SURF(speed up robust features)in image matching,unstable iteration times of RANSAC algorithm and error caused by artificially setting inner point discrete threshold in image matching,an image matching method combining improved marginalizing sample consensus algorithm and improved SURF is proposed.Firstly,input image is preprocessed by fast guided filtering algorithm to filter image noise and retain edge details.Then,the efficient binary descriptor for SURF is constructed by BEBLID(boosted efficient binary local image descriptor)algorithm.Combined with improved marginalizing sample consensus algorithm,incorrect matches are filtered out by marginalizing outer points.Compared with SURF,this method has higher accuracy and better real-time performance,and can meet the needs of image matching in various complex environments.