基于概率抽样一致性的基础矩阵估计算法  

Fundamental matrix estimation based on probability analysis and sampling consensus

作　　者：鲁珊[1] 雷英杰[1] 孔韦韦[1] 雷阳[1] 

机构地区：[1]空军工程大学导弹学院,陕西三原713800

出　　处：《控制与决策》2012年第3期425-430,共6页Control and Decision

基　　金：国家自然科学基金项目(60773209)

摘　　要：随机抽样一致性算法是应用最广泛的鲁棒性基础矩阵估计算法之一.针对随机抽样一致性算法效率低的问题,提出一种基于概率分析的随机抽样一致性算法.该算法减少首次抽样的次数,利用预检验技术确定一个较优的模型子集,通过定义样本属于该较优模型的概率来选择内点概率大的样本子集,并通过多次迭代得到一个只包含内点样本的子集.模拟数据和真实数据的基础矩阵估计实验表明,算法在计算效率和精度上均优于随机抽样一致性算法.Random sample consensus（RANSAC） algorithm is one of the most widely used for robust fundamental matrix estimator. Considering low efficiency of RANSAC, a random sample consensus algorithm based on probability analysis is proposed. The algorithm reduces the quantity of fist random sampling, and incorporates a better subset of model using preemption scheme. The probability that the sample belongs to the better model is defined in order to be incorporated into the sample subset. After several iterations, the sample subset contains only inliers. Experimental results of the fundamental matrix computation on both simulated and real data show the superiority of the proposed algorithm in precision and efficiency.

关 键 词：随机抽样一致性算法 基础矩阵 鲁棒性 模糊集 特征匹配 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]