Abstract:

The automatic detection and tracking of ground moving targets are main tasks in the photoelectric imaging guidance. For the detection of ground motion, a coarsetofine iterative global motion compensation method is proposed to compensate the displacement of image pixels due to the motion of sensor, and a moving target detection algorithm based on the change energy measure is presented to extract the region of moving targets on each level of image pyramid and difference image pyramid. The moving targets are located based on two change energy images evaluated from the three sequential images. For the ground motion tracking, particle filters fused motion feature and gray histogram are proposed to reliably track the moving targets on the ground. In the practical aerial video, the iterative global motion compensation method is compared with Fourier-Mellin transform (FMT), Kanade-Lucas-Tomasi (KLT) feature tracker and Harris corner, and the detection algorithm with change energy measure and the tracking algorithm with particle filters are validated.