Image Nonlinear Processing, And Model Electro-optical Tracking Applications

Automatic target recognition (ATR) is the key problem in EO (electric and optical) tracking, however, how well the image preprocessing performs largely affects the performance of the target detection and recognition algorithms. On the demand of real weapon tasks, this paper focuses on the detection and recognition of small target in EO tracking, and the major subjects studied in this paper include: Nonlinear image processing techniquesPerfect solutions for background leakage and target polarity reversionIn this paper, we first study systemically some typical nonlinear image preprocessing techniques in current EO tracking applications, i.e., stack filter, median filter and GOS filter, and analyze their merits and defects respectively according to the experimental simulations with real infrared or TV images.Secondly, the theory of mathematical morphology is studied in depth, and with consider of image properties and target characteristics in EO tracking, a new approach based on the combination of horizontal direction processing with vertical direction processing, and the combination of opening filtering with closing filtering is put forward. This approach could solve the problem of background leakage and target polarity reversion, and has been successfully applied to real information processing system in EO tracking, which makes outstanding progress in the practicability of the algorithm and the real-time implementation of the processor.