摘要：针对基于局部特征的随机抽样一致性算法在抽样策略上的“盲目性”及其应用在低 内点率场景下出现的不稳定性，本文提出了一种基于顺序抽样评估一致性算法（Sequential Evaluation on Sample Consensus，SESAC）及其改进的基于似内点标记的顺序抽样评估一 致性框架（Sequential Evaluation on Sample Consensus with Likelihood Inlier-Point Marker， LIMP-SESAC）。通过对每个特征点置以似内点标记及依据假设模型的一致性验证结果对 似内点标记的重新评估，LIMP-SESAC 在特定的时机对特征点进行重排序以达到将内点 和噪点分层的目的，以便于顺序抽样能够并且尽快地抽到较优的无噪点样本。实验结果 表明，LIMP-SESAC 在内点率、时间和鲁棒性上均比 USAC 要好。
关键词 特征匹配 随机抽样一致性算法 渐进抽样一致性算法 随机抽样一致性通用 框架 基于似内点标记的顺序抽样评估一致性框架
Title Evaluation Method on Sample Consensus of Local Feature Matching
Abstract：To solve problems of the “blindness” about sampling strategy of Random Sample Consensus based on local feature and the instability of the algorithm’s applications under low-inlier-point-rate scenes, we proposed a Sequential Evaluation on Sample Consensus(SESAC) and its improved framework named Sequential Evaluation on Sample Consensus with Likelihood Inlier-point Marker(LIMP-SESAC). By assigning the likelihood inlier-point marker to each point and reassessing the markers according to the consensus verification results of hypotheses, LIMP-SESAC reorders the feature points at specific condition to separate inlier-points and noise-points for making it possible to get an uncontaminated sample and get a better one as soon as possible through sampling sequentially. Experiment results demonstrate that the performance of LIMP-SESAC is better than USAC in inlier-point rate, time and robustness.
Keywords Feature matching RANSAC PROSAC USAC LIMP-SESAC
目 次
1 引言 1
1.1 课题研究背景及意义 1
1.3 总体技术方案及其社会影响 5
1.4 技术方案的经济因素分析 6
1.5 论文主要内容 6
2 随机抽样一致性类型算法 7
2.1 随机抽样一致性算法 7
2.2 渐进抽样一致性算法 10
2.3 本章小结 14
3 顺序抽样一致性算法 15
3.1 基于顺序抽样评估一致性算法 15
3.2 实验比较 局部特征匹配的一致性评估方法:http://www.chuibin.com/jisuanji/lunwen_205619.html