摘要
基础矩阵描述了单个场景两幅视图间的对极几何关系,在计算机视觉领域占有重要地位。该文基于RANSAC算法的随机抽样思想,通过记录抽样过程中每对匹配点被判定为内点的次数,选择次数较多的匹配点作为初始集合。采用鲁棒扩充算法对初始集合进行扩充寻求内点集,并利用该内点集计算基础矩阵。以大量的模拟数据和真实图像进行了实验,结果表明,该算法能进一步降低噪声和错误匹配点对基础矩阵估算精度的影响,可以有效检测和删除错误匹配点。
The fundamental matrix describes the geometric relation that exists between two images of the same scene,and plays an important role in computer vision.This paper presents a new robust method to estimate the fundamental matrix based on the RANSAC algorithm.The proposed algorithm records the number of being an inlier for each point from each sampling process,then choose the points which were have larger value in the array to construct an initial inliers set from the matching points.It takes the greedy strategy to expand the initial set to obtain a better subset.Finally,the fundamental matrix is calculated with the obtained inliers set.Through a mass of experiments on simulated data and real images in the case of mismatching and Gaussian noise,the comparing results indicate that the algorithm not only reduce the noise and mismatching of the impact on the fundamental estimating precision,but also could detection and deletion the mismatching point.
出处
《地理与地理信息科学》
CSCD
北大核心
2013年第2期26-30,共5页
Geography and Geo-Information Science
基金
江苏省高校自然科学研究重大项目(10KJA420025)
国家支撑计划项目(2012BAH35B02)
江苏省高校优势学科建设工程资助项目
福建省自然基金青年创新项目(2011J05104)
