Occlusion-resistant target tracking by the particle filter track-before-detect algorithm
Osman Topçu, Hakan Büyük · 2016
Targets with low signal to noise ratio are difficult to detect. Track-before-detect algorithm detects weak targets by sufficiently tracking them. Particle filter track-before-detect algorithm is flexible enough to be modified for occlusion handling. The first novelty of this work, is the modification of the particle filter track-before-detect algorithm to handle occlusions. The second one is the modeling of the image of a point target on an electro-optic sensor. For this purpose, an optical system design is made and point spread function of the system is modeled by a Gaussian function, that changes as the angle to the optical center of the sensor changes. This Gaussian model includes amplitude, optical distortion, X and Y axis standard deviations. The proposed algorithm is tested with varying signal to noise ratios and successful tracking and detection is achieved during occlusion and maneuvering of the target.