Moving point targets detection in infrared image sequences has been the emphasis and difficulty in the field of target detection and identification. For the detection of dim and small targets in infrared image sequences, uniform temporal profile models of the temporal behaviors of the background noise pixel, target pixel and clutter pixel, are established, in terms of the mechanism of different background temporal profiles. Further, a detection algorithm is developed based on the different temporal models in this paper.The detection algorithm is divided into two steps: background suppression and pulse target detection. Background suppression is the key and basis to the algorithm. Through investigating temporal characteristic of target, a new algorithm of background suppression is presented, according to temporal median filter. This algorithm combines temporal median filter and temporal morphological filter. After background suppression, noise pixels and most background pixels in the infrared image sequences are almost suppressed and only target pixels and a few clutter pixels can exceed the given detection threshold. The new method is compared with temporal median filter and space morphological filter. The result proves superiority of the new method. For sequence images after background suppression, the discrepancy of the temporal profile and its envelope can be modeled by Gaussian distribution according to temporal models of target pixel and background pixel. A detection measure is concluded. The method can further suppress the clutter pixels and enhance the performance of the temporal detection algorithm.The theoretical analysis and the experimental result indicate extensive adaptability, strong ability of preventing interference and good detection performance of the algorithm. Especially, it is very effective for target detection in low SNR. |