A detection algorithm for maneuvering IR point target and its performance evaluation

An algorithm for detecting a maneuvering IR point target in complex background is proposed in this paper. It is based on frame change detection and the Reverse Phase Feature of Neighborhood (RPFN) of a maneuvering point target in a first order frame difference image that two positions of the target are near and the gray-values of them are close to in absolute value but with inverse sign. Firstly a proper threshold is selected adaptively for a frame difference image so as to reduce a majority of noises. Secondly some pairs of points with RPFN are searched, based on which corresponding vectors can be found. Thus the IR point target detection is carried out by the distribution tendency in the vectorgraph of point pairs with RPFN. Simulation experimental results demonstrate the feasibility and effectiveness of the algorithm proposed in this paper. Finally, ROC (Receiver Operating Characteristic) curve analysis is used to evaluate the sensitivity of the threshold in the algorithm and the rationality of the selected threshold.