A fast probabilistic model for hypothesis rejection in SIFT-based object recognition

Abstract

This paper proposes an improvement over the traditional SIFT-based object recognition methodology proposed by Lowe [3]. This improvement corresponds to a fast probabilistic model for hypothesis rejection (affine solution verification stage), which allows a large reduction in the number of false positives. The new probabilistic model is evaluated in an object recognition task using a database of 100 pairs of images.