Discrete-time point process filter for image-based target mode estimation

The tracking and prediction of a maneuvering target can be improved by using additional (and unconventional) measurements of its apparent features (target modes), typically provided by an image sensor-processor. A model for the image-based observation channel in discrete time is presented, and a multidimensional point process filter for target mode estimation is obtained by making use of discrete-time point process theory. The filter's utilization is illustrated by simulation examples.<>