基于Adaboost-CSHG的特定类目标跟踪识别

Specific Target Tracking and Recognition Based on Adaboost-CSHG

目标跟踪识别是计算机视觉领域的热点研究对象。首先采用基于Adaboost的目标检测算法,训练得到了特定类目标坦克模型的级联分类器,对图像中的坦克目标完成了"粗检测";通过构建类属超图(CSHG)模型,采取Adaboost与CSHG相结合的方式,有效滤除了大量虚警,实现了对坦克目标的"精检测",同时完成了对目标的跟踪;最后利用基于类属超图的目标识别原理对目标进行识别,实验结果表明该方法在简单背景和复杂背景图像条件下均具有可行性。

Target tracking and recognition is the hot spot research object of computer vision. Firstly a target detection algorithm based on Adaboost was adopted to train a specific target classifier, which only makes a rough detection on tank model. Via building CSHG model, lots of false alarm could be deleted by connecting Adaboost and CSHG,and then an accurate detection on tank model was made,in the meanwhile, tracking the target was finished . Finally a target re- cognition method based on CSHG was used to recognize the target. Experimental results show that the algorithm can work well in both simDle background and complicated background image conditions.