基于颜色特征传递的多摄像头无重叠区域运动目标跟踪

The object tracking in the multiple cameras surveillance system based on the color transferring

针对多摄像头无重叠区域目标监控时,同一目标在不同摄像头中易出现色差而导致目标颜色特征匹配失败的问题,研究了多项式转换法、基于统计特征的颜色转换法和自适应颜色转换法。在这些方法的基础上,结合视频监控的具体应用要求,提出了将基于统计特征的颜色转换法应用在多摄像头运动目标跟踪中的具体改进方法。实验结果表明,该方法对于多摄像头无重叠区域下运动目标跟踪效果良好。

When an object moves through a multiple cameras surveillance system, the objects color characteristic matching may fail due to the color characteristic differences against different cameras. For solving this problem, it proposes three approaches for transferring the color characteristic among the multiple cameras＂ color spaces： polynomial approach, statistics transferring approach and self-adapt transferring approach. Considering the ap- proaches achieving result and the specific requirement in the video processing, it presents an improved video col- or transferring approach based on the statistics transferring, the testing result proves that the methods can really improve the tracking accuracy in the multiple cameras surveillance system.