Feature based object recognition using statistical occlusion models with one-to-one correspondence

TL;DR: A new Bayesian framework for partially occluded object recognition with one-to-one correspondence is presented, and fast algorithms for finding the optimal one-to-one correspondence between scene features and object model features to compute the generalized likelihood are developed.

Abstract: In this paper we present a new Bayesian framework for partially occluded object recognition with one-to-one correspondence. We introduce two different statistical models for occlusion: One model assumes that each feature in the model can be occluded independent of whether any other features are occluded, whereas the second model uses spatially correlated occlusion to represent the extent of occlusion. Using these models, the object recognition problem reduces to finding the object hypothesis with largest generalized likelihood We develop fast algorithms for finding the optimal one-to-one correspondence between scene features and object model features to compute the generalized likelihood. We evaluate our algorithms using examples extracted from synthetic aperture radar imagery, and illustrate the performance advantages of our approach over alternative algorithms proposed by others.