基于Kim与Camshift结合的运动目标跟踪算法

Moving object tracking algorithm based on Camshift and Kim algorithms

为了能够准确和快速的跟踪部分遮挡的运动目标,提出了Kim与Camshift相结合的运动目标跟踪算法。首先通过Kim算法(即帧差法与背景相减法结合)自动的提取目标,通过提取目标确定运动区域及运动区域的质心,然后通过此运动区域提取目标特征,用Camshift方法进行跟踪。此法克服了传统Camshift算法需要人为定位,对部分遮挡目标出现目标丢失和跟踪区域偏移的情况,最后通过实验验证了本算法对部分遮挡运动目标跟踪有较好的鲁棒性。

The moving object tracking algorithm based on the combination of Kim and Camshift algorithms is put forward to track partially masked moving objects quickly and exactly.Firstly,the Kim algorithm（frame difference combined with background difference between frames） is applied to extract the moving object automatically,which is then used to confirm the centroid of its moving region.After that,the feature of the moving region is extracted to track the target using the Camshift algorithm.This method gets rid of the requirement of manual positioning,and overcomes the shortcomings of traditional Camshift algorithm such as the missing of partially masked objects or the deviation of tracking regions.At last,some experiments are performed to verify the robustness of this algorithm for partially masked moving object tracking.