Research On Infrared Small Target Detection Method And Its Key Technology Under Strong Ground Clutter

As the key techniques of infrared surveillance system,infrared small target detection,including background suppression,false alarm rate suppression and track correlation,was widely used in military and civilian fields.The infrared small target detection method under the strong clutter was in-depth studied in this paper.The specific contents are as follows:To cope with the background suppression under strong clutter,two spatial background suppression algorithms were proposed which is based on the establishment of infrared target spatial distribution model in this paper.The differences between target and background in the shape of the surface was analyzed which was represented by curvature in first algorithm.And the Facet model was used to calculate the image curvature.The experimental results show that the signal-to-clutter and background suppression factor are higher than others in this paper.The difference between target and background in the local area of gradient was analyzed in the other algorithm.The histogram of oriented gradients difference measure method was constructed which was based on human vision attention mechanism.This method could process the bright target and the dark target simultaneously,and the experimental results show the algorithm could double the local signal to background ratio gain,thus effectively suppressing background clutter.For the problem of false alarm rate suppression under strong clutter scenarios,a false alarm rate suppression algorithm based on background model was proposed.The differences of target and background in the time domain was analyzed,and the background model was established used by Gaussian distribution and the suppression method based on background model was designed.In order to further suppress the false alarm rate and improve the stability of the system,a multi-feature joint track correlation algorithm was proposed,which designed the calculation method for the multi-feature joint observation probability of image points based on Bayesian framework.And the method of track correlation was constructed based on the multi-feature joint observation probability of image points.The experimental results show the effectiveness and robustness of the above algorithm.Finally,the signal processing system based on FPGA and ARM was designed to implement the detection method which was proposed above,and the result was displayed in real time using the host computer software.A large number of field experiments proved that the system designed in this paper is feasible and stable.