Object detection in a cluttered background with a noisy image

Abstract

In this paper we present a correlation filter designed with a single noisy refer- ence image in which the target is placed at unknown coordinates and its shape is unknown. The filter is designed by optimizing the peak-to-output energy criterion for a signal model that considers the presence of additive noise in the reference image and a disjoint cluttered background in the input scene. The performance of the proposed filter is compared with that of common filters by means of computer simulations.