Research On Dim Target Detection Methods For Infrared Oversampling Sensors

Infrared dim small target detection is a hot topic in the field of target detection and tracking. The signal-to-noise of dim and small target is low, easy to be background and noise, brought difficulty to detect. This paper focuses on the research of the over sampling infrared dim small target detection algorithm in complex cloud background. first of all to establish the model of infrared image, and the infrared dim small target image characteristics analysis. Second, this paper introduces several typical background suppression algorithm, and on the spatial high pass filter, mean filter, Max median filter are introduced in detail. Third, through the analysis of background suppression with a rectangular structure element and linear multijunction structural element morphology,linear multi structure elements can inhibit the cloud background edge, but spot noise cannot eliminate less than the target, and the rectangular structure element on the cloud background edge inhibition effect is very poor, spot noise but can eliminate less than a target. On the basis of combining the advantages of two kinds of structure elements, puts forward the infrared dim small target detection algorithm based on morphology of multilevel filter. Introduces the structure and principle of the algorithm, and the algorithm has carried on the simulation experiment. The simulation experiment show that this algorithm can largely improve the over sampling image signal-to-noise ratio, the candidate target points of single frame detection are also relatively small, this algorithm effectively improves the ability to detect dim target. In multiframe detection, this paper improve the traditional neighborhood decision method and puts forward the detection algorithm of the over sampled target sequence based on united state estimation, then candidate points with the field of decision method on single frame detection results for tracking, draw the trajectory of moving object, which verified the validity of the over sampling infrared dim small target dectection algorithm based on morphology of multilevel filter.