Object recognition using multi-view imaging

Difficult situations such as high noise or low resolution can seriously degrade the performance of object recognition algorithms that operate on isolated images. We show that recognition performance may be improved substantially in such cases by fusing the information available from a sequence of multi-view images. In this paper we present two algorithms for object recognition based on SIFT feature points. The first operates on single images and uses chirality constraints to reduce the recognition errors that arise when only a small number of feature points are matched. The procedure is extended in the second algorithm which operates on a multi-view image sequence and, by tracking feature points in the plenoptic domain, is able to fuse feature point matches from all the available images resulting in more robust recognition.