Research On Recognition Of Infrared Ship Target With Complicated Background

With the development of navy in modern war and many advanced naval ships armed in navy, the technique for automatic target recognition of naval ships become more important. Therefore the technique of recognition of infrared naval ships becomes important research task in automatic recognition of infrared target with highly academic and practical value. However, because war field and the background of the sea-sky ground are usually comparatively complex, and the target recognition characteristics are greatly influenced by the weather, it is not only very difficult but also very challenging to recognize the targets of infrared naval ships.In this paper, based on the background of study task, we analyze the characteristics of naval ships in infrared image. We made a comparison and integration on various image pre-process algorithms, then we propose a method of Image Enhancement based on adaptive double equalization on plateau equalization; and we present the method of adaptive threshold segmentation at IR ships target based on the mean of row and line after we study the methods of image segmentation, compare and experiment different threshold segmentation algorithms. The algorithm is simple, fast and easy to be realized with less calculation, so it plays a roll in segmenting. To decrease the calculation of target extraction, segmented images are identified by the algorithm of Run-long code region identification.This paper is focused on the methods of infrared ship recognition in complex background. Geometry character and invariant character of ship target is analyzed and a united relative invariant moment is used to calculate the feature to meet real-time requirements. A phase recognition is put forward: first the method of Track Before Recognition is used to detect and recognize ship target based on centroid track, then the target is recognized by feature parameters match. The experiment results verify the efficiency of our algorithm.