Robust Real-Time Pattern Matching Using Bayesian Sequential Hypothesis Testing

This paper describes a method for robust real-time pattern matching. We first introduce a family of image distance measures, the Image Hamming Distance Family. Members of this family are robust to occlusion, small geometrical transforms, light changes, and nonrigid deformations. We then present a novel Bayesian framework for sequential hypothesis testing on finite populations. Based on this framework, we design an optimal rejection/acceptance sampling algorithm. This algorithm quickly determines whether two images are similar with respect to a member of the Image Hamming Distance Family. We also present a fast framework that designs a near- optimal sampling algorithm. Extensive experimental results show that the sequential sampling algorithm´s performance is excellent. Implemented on a Pentium IV 3 GHz processor, the detection of a pattern with 2,197 pixels in 640times480 pixel frames, where in each frame the pattern rotated and was highly occluded, proceeds at only 0.022 seconds per frame.