OBJECT RECOGNITION BASED ON IMPULSE RESTORATION WITH USE OF THE EXPECTATION-MAXIMIZATION ALGORITHM

It has recently been demonstrated that object recognition can be formu lated as an image-restoration problem. In this approach, which we term impulse restoration, the objective is to restore a delta function tha t indicates the detected object's location. We develop solutions based on impulse restoration for the Gaussian-noise case. We propose a new iterative approach, based on the expectation-maximization (EM) algorit hm, that simultaneously estimates the background statistics and restor es a delta function at the location of the template. We use a Monte Ca rlo study and localization-receiver-operating-characteristics curves t o evaluate the performance of this approach quantitatively and compare it with existing methods. We present experimental results that demons trate that impulse restoration is a powerful approach for detecting kn own objects in images severely degraded by noise. Our numerical experi ments point out that the proposed EM-based approach is superior to all tested variants of the matched filter. This result demonstrates that accurate modeling and estimation of the background and noise statistic s are crucial for realizing the full potential of impulse restoration- based template matching. (C) 1998 Optical Society of America.