Abstract

Processing of quantum limited images requires optimal utilization of all the available information: all the registered photons, prior information such as geometrical and intensity model of imaged objects formulated in terms of relatively small number of parameters, and real time and a posteriori information that could be available from experts or from other sensors. Such an approach to adaptive information fusion image processing is presented here based on maximum likelihood neural network developed previously by the author for adaptive classification and sensor fusion^[1,2,3].

© 1989 Optical Society of America