The Dynamic Programming Based Read- Time Detection Of Dim Moving Point Targets In Three Signal

The detection technology of dim point moving target in image sequences from infrared sensor has been the research hot spot because of its wide use in military area and in civil area. The detection of dim point moving target in low Signal to Noise Ratio (SNR) environment is a difficult problem of research, since the SNR of target is low and it has no texture. In practice, the real-time demand of detecting dim point moving target in image sequences brings a great many challenges to the detection.In this thesis, firstly, the dim point moving target in image sequences from infrared sensor in low SNR environment is detected real time and effectively. With the real-time request, the remain background clutter in a single image is rejected twice and part of noise is removed, and then the gray scales of the single image are fewer, and so the number of pseudo-targets is smaller greatly. Secondly, with the assumption of unknown target's initial position, the initial positions of the suspected targets are found out, that is to say, single frame detection. Thirdly, a constraint condition of velocity is proposed, and the suspected tracks are found out. With the combination of the only energy and velocity information in the image, by decision-making optimization in stages, the energy of suspected targets is partitioned again, and the track of the target is obtained, that is to say, multiple frames connection. A novel method of decision-making optimization in stages is proposed, which has wide application, and the remain noise suppression and target initial position measurement are added, based on the improved dynamic programming algorithm.The simulation results show that this algorithm has wide application and can detect the dim point target in low SNR environment effectively. The combination process not only avoids influencing the correctness of original algorithm, but also greatly reduces computational complexity, and then improves the real time capability of the detection for practice. The simulation results also verify the efficiency of the algorithm.