Infrared Imaging Detection And Recognition Technology For Ground Scatterable Mine

With the rapid development of infrared imaging technology, it is widely applied in military target detection and recognition. Mine-detection technology based on forward-looking infrared imaging has become the world's research hotspot, thanks to high imaging resolution, wide field of view, high efficiency and high precision. In this paper, aimed at the detection of ground scatterable mine, infrared imaging technology applied in mine-detection system is analyzed according to the design of system and the implementation of process, mainly to expand research work from the following aspects:(1) To analyze the feasibility, the research starts from the basic principles of mine-detection system to which applied the infrared imaging technology. Then, combined the characteristics of ground scatterable mine, different environmental conditions influencing factors are considered by infrared radiation mathematical model. Moreover, several special factors are provided.(2) Mine-detection system using infrared imaging technology based on vehicle is set up, and is discussed from the system components, the basic working principle and system workflow. In order to achieve the detection, identification and location of mines objectives, the detailed algorithm processing is designed.(3) An infrared focal plane array imaging may exist non-uniform noise, this paper select the column equalization non-uniformity correction algorithm, to recover the degraded infrared image. Then an adaptive piecewise linear transformation algorithm is used, to achieve the purpose of mine target inhibition' enhancement and complex background' inhibition, and meet real-time requirement on processing efficiency.(4) After the infrared mine image of preprocessing, starting from the segmentation method based on threshold, according to the characteristics of infrared mines scene' distribution, a threshold segmentation method constrained by cross entropy is introduced. Compared with the traditional inter-class variance or maximum entropy method, the selected algorithm can divide landmine more accurately. To identify mine targets accurately in the late, the image after segmentation by threshold should also be eliminated background noise and be filled the hole of landmines target.(5) After the segmentation of infrared mine scene, the connected domain of different mine targets waiting to be recognized can be counted by the obtained binary image. Based on the infrared image' object recognition principle by features invariant, the model of similar circular mine targets' convex hull invariant can be established for solving the target area' matching and recognized. Finally, the gray center of gravity positioning algorithm is employed to mark the target location coordinates of mines.Experimental results show that, the selected processing algorithm can not only achieve high-precision detection of mine targets in infrared mine scene, but also meet the real-time processing requirements of infrared imaging system for mine detection.