Discriminant Filters For Object Recognition

This paper presents a technique for using training data to design image filter s for appearance-based object recogni- tion. Rather than scanning the image with a single set of fil- ters and using the results to test for the existence of objects, we use many sets of filter s and take linear combinations of their outputs. The combining coefficients are optimized in a training phase to encourage discriminability between the filter responses for distinct parts of the object and clutter. Our experiments on three popular filter types show that by using this approach to combine sets of filter s whose design parameters vary over a wide range, we can achieve detec- tion performancecompetitive with that of any individual fil- ter set. This in turn can ease the task of fine-tuning the settings for both the filter s and the mechanisms that analyze their outputs.1