Motion Estimation Based on Eigenvalue Algorithm Matching Local Image Feature

概要

論文の詳細を見る
This paper describes a motion estimation method with eigenvalue algorithm matching local image feature. We introduce two estimation methods based on eigen value algorithm for improvement in accuracy. The first method is based on robust estimation. We apply this method to the block which includes multiple movements. The second method is applied to the block in which one typical gradient alone exists. In this method, we complement the eigenvector from the object with the same edge in the subjective block, and then motion estimation is calculated with the complementary eigenvector. Experimental results show that the proposed method improves the estimation accuracy.
社団法人電子情報通信学会の論文
2002-02-01