Image Tracking in Multitarget Enviornment

Abstract

Pattern recognition techniques are applied to the problem of multiple target identification and tracking of (forward looking infrared) FLIR image sequences of missiles. The image is first thersholded, using an automatic thresholding technique, to yield a binary image. Clustering analysis is then performed and small clusters that represent noise are removed. Geometrical properties of each cluster are calclulated using the method of moments. Overlapping targets are identified using a newly derived moment that is sensitive to distortions in ellipse-like sahpes (missile shape). The number of targets is estimated as the number of nonoverlapping clusters, and their centroid is calculated in each frame. A record of each target is kept till it leaves the field of view. Simulation studies are presented and the results indicate that the proposed approach has good tracking performance.