A statistically selected part-based probabilistic model for object recognition

Abstract

In an object recognition task where an image is represented as a constellation of image patches, often many patches correspond to the cluttered back-ground. If such patches are used for object class recognition, they will adversely affect the recognition rate. In this paper, we present a statistical method for selecting the image patches which characterize the target object class and are capable of discriminating between the positive images containing the target objects and the complementary negative images. This statistical method select those images patches from the positive images which, when used individually, have the power of discriminating between the positive and negative images in the evaluation data. Another contribution of this paper is the part-based probabilistic method for object recognition. This Bayesian approach uses a common reference frame instead of reference patch to avoid the possible occlusion problem. We also explore different feature representation using PCA an 2D PCA. The experiment demonstrates our approach has outperformed most of the other known methods on a popular benchmark data set while approaching the best known results. © Springer-Verlag Berlin Heidelberg 2006.