RANSAC类方法先入为主的缺陷和避免方法

Preconceived Error in RANSAC and an Approach to Avoid It

报告了一类模型拟合错误的现象,涉及RANSAC、MLESAC、PROSAC等包含随机采样环节的模型拟合方法。分析了该类错误为先入为主的验证策略所导致,并提出了避免产生这类错误的校验方法。实验结果表明:本方法可以快速有效地避免这类错误。

A kind of incorrect model fitting phenomena is presented, including the model fitting method of random sampling, such as RANSAC, MLESAC, PROSAC. Furthermore, it analyzes the reasons of such error, which is caused by preconceived verification strategy. In order to avoid the error, a method of checking is proposed. Experimental results demonstrate that the proposed method can avoid such errors rapidly and effectively.