
Determining the Probability of a False Positive When Matching Chains of Oriented Pixels
Clark F. Olson and Daniel P. Huttenlocher In Proceedings of the ARPA Image Understanding Workshop, pages 11751180, 1996. This paper gives a method for determining the probability of finding a false positive instance of an object in an image when matching is performed using chains of pixels with associated local information (such as orientation). We model the matching process between a chain of object pixels and the image as a Markov process, where each state represents either the presence or absence of a match for an object pixel in the image. This model yields the probability that particular position of an object model will result in a false positive. We use the probabilities from discrete positions of the model to estimate the probability of a false positive over the space of allowable transformations of the model. These techniques have been used to implement an adaptive recognition system, where the matching threshold is set such that the probability of a false positive is low. 