Research On New Methods Of Detection Of Infrared Dim Small Target In Cluttered Sky Background

The following studies are carried out on detection of infrared dim small target in cluttered sky background in this paper.The separable features of dim small target and background in temporal domain and spacial domain were studied based on the analysis of image characteristic of background and targets. The separable characteristics were analyzed and compared in performance. Taking feature of gray distribution of fluctuant background strongly correlated in space, feature of target gray strongly correlated between consecutive frames and feature of target location strongly correlated between consecutive frames as the separable characteristics adopted by algorithms in this paper.Based on the analysis of the model of background prediction, the definite weight model and the adaptive model were compared in perfromance. Taking image feature of gray distribution of fluctuant background in strong correlation space as the reliable space separable characteristic between background and dim small targets, a local homogeneous background prediction methods was presented to overcome the influence on the prediction accuracy from heterogeneous background region. As a result, the unnecessary prediction operations and derived operation errors were sharply decreased, signal clutter ratio in the residual image was largely increased, and the detection performance of dim small target was improved.A temporal profile algorithm was proposes based on comparison filtering as a responding method to the fake-alarm occurrence existing in the traditional detection algorithm when the dim small target has equivalent velocity with that of cloud edge clutters. Based on the analysis on the time domain characteristics of the dim small target, cloud edge clutters as well as the stationary background, the characteristic of the temporal profile is adopted to restrain the stationary background, then the spatial domain comparison filter is structured based on the fact that the pixels of the cloud edge clutters are continuous in spatial domain while the pixels of dim small target are discrete, and the images after removal of the static background are filtered with comparison filter; lastly, connecting line of the stagnation points based filtering is used to realize the detection of dim small target. Simulation data show that this algorithm can significantly eliminate the fake-alarm caused by the cloud edge clutters with equivalent velocity of the target, thus further improve the detection probability of dim small target.A pipeline filter algorithm based on motion direction estimation was suggested as a method to improve the flaw of detection probability deduction due to the strong interferential noise within the pipeline and low signal noise ratio. The method analyzed the motion characteristics of infrared dim small target and establishes the motion direction estimation model according to the continuity characteristic of the targets between consecutive frames. Through the model, the prior position information of the targets is detected frame by frame and analyzed in order to estimate the motion direction and trajectory of the targets. The estimation results are made use to eliminate the interference on the targets caused by the pipeline inner noises. Experiments and simulation results show that the algorithm can suppress noises in the pipeline effectively, increase the detection probability of the targets, and strengthen the resistance characteristic of the targets against the noises within the pipe.