An Algorithm of Target Tracking and Detecting Synchronously Based on Inter-Frame Dual Match

Abstract

Aiming at the detecting and tracking of dim point targets in complicated sea and sky background infrared images, an algorithm of tracking while detecting is proposed based on inter-frame dual match in image sequences. A mathematical model is established, using the marking image sequences M to memorize the interest region of input image sequences' adjacent frame matching result for the first time in frame-to-frame manner, using the mark image T to memorize the interest region of M's adjacent frame matching result for the second time in point-to-point manner, using the statistical frame S to memorize the successful match times of each pixel in point-to-point manner, outputting matrix T and S to show the target trajectories with the input single frame image synchronously. Algorithm's whole matching process doesn't change as targets' numbers or targets' movement state change; it takes only three adjacent images to complete a single matching process; there's no need to judge the target's suspected location; tracking problem such as a sudden loss or appearance of target in image sequences is effectively solved. Especially, the algorithm solves the difficult head-on targets detecting and tracking problems under the highly noisy condition. According to the simulation and experiment with practical images under the different condition time after time, it has been confirmed that the merits mentioned above with high application value is different from the traditional algorithms and the algorithm's effectiveness is validated with high reference value in the military applications.