中文摘要：

针对动态的实时目标跟踪算法Camshift（continuously adaptive meanshift）在背景复杂或者存在较多与目标颜色相近的像素时,容易出现跟踪目标丢失的问题,研究并实现一套基于Camshift和SURF（speeded up robust features）算法的目标跟踪系统。使用双内核的DM3730为核心,连接网络摄像头实现图像的动态采集,在DM3730上实现SURF算法和Camshift算法的融合,并负责将采集到的视频压缩,通过网络传输到计算机,对得到的图像做进一步结果分析。实验结果表明,基于Camshift算法和SURF算法融合的目标跟踪系统在简单背景、有相似物体干扰和复杂背景等情况下都能够更准确快速地跟踪到目标,鲁棒性更强、效果更好。

英文摘要：

Camshift is a dynamic real-time target tracking algorithm based on color features.However,when similar color with the object exists in the background,it may loss the object.To solve the problem,an object tracking system based on Camshaft and SURF was proposed.The IP camera was used to capture the image,and the Camshift and SURF algorithm were accomplished on the dual-core DM3730 to solve shortages of Camshift.Meanwhile,the video was compressed and transmitted to the computer.Experimental results show that the object tracking system can work effectively in different kind of environments,such as simple or complex background,some similar objects disturbing in the environment.It is more robust and more effective.