Target image tracking using a nonlinear filter based on Pontryagin minimum principle

In this report, the focus is on the problem of target tracking in forward looking infared (FLIR) image sequences. In this problem, the output of FLIR sensors is used as measurements to track highly maneuverable targets, such as air-to-air missles, with spatially distributed and changing image intensity profiles. The target intensity profile is assumed to be unimodal and well described by a bivariate geussian function. A new nonlinear filter based on Pontryagin minimum principle is introduced. This filter is used to improve the signal to noise ratio (SNR) of FLIR image. By horizontally and vertically scanning the image after filtering, the target center is estimated. Monte Carlo simulation is used to compare the performance of the proposed method and that of the conventional correlation method for a highly maneuvering target environment and at different SNR values (from -65 db. to + 45 db). It is shown that the proposed method improves the accuracy of the estimate of the target center by 40% on the average, with about 10% of the time.