Experiments with an extended tangent distance

Abstract
Invariance is an important aspect in image object recog- nition. We present results obtained with an extended tangen t distance incorporated in a kernel density based Bayesian classifier to compensate for affine image variations. An im- age distortion model for local variations is introduced and its relationship to tangent distance is considered. The pro - posed classification algorithms are evaluated on databases of different domains. An excellent result of 2.2% error rate on the original USPS handwritten digits recognition task is obtained. On a database of radiographs from daily routine, best results are obtained by combining tangent distance and the proposed distortion model.

This publication has 5 references indexed in Scilit: