中文摘要：

随着智能技术的快速发展,广大学者对于视频监控领域的运动目标检测与跟踪产生了极大的兴趣,纷纷投身其研究行列之中。另外,运动目标检测与跟踪在工业控制、智能交通等方面也得到了一定的应用。随着社会经济的快速进步,交通的监管变得越发复杂,如何对道路上的车辆进行实时检测、对违章车辆进行监控、对违法车辆进行有效布控等成为了大家争相研究的热点问题。针对各种交通视频中的运动车辆的复杂情况,提出了对运动车辆进行有效的检测与跟踪的改进算法,即简单快速的自适应k均值聚类算法,实验表明：这种改进算法较其他聚类算法在特征光流中的应用具有一定的优势,能够提高聚类精度,给出更加精确的目标数,对运动车辆能够进行有效跟踪,算法的实验效果比较理想。

英文摘要：

With the rapid development of intelligent technology, many scholars for moving target detection and tracking in the field of video surveillance had great interest and been devoted to the study. In addition, the moving target detection and tracking in the field of industrial control and intelligent transportation which got a certain application. With the rapid progress of social economy, the traffic regulation becomes more and more complex. How to detect and monitor the vehicles on the road in real time, how to effectively control the illegal vehicles have become hot issues that everybody is studying. In this paper, under the complexity of motion range of vehicles in traffic video condition, an algorithm that effectively improves the moving vehicle detection and tracking is proposed, namely simple and fast adaptive k - means clustering algorithm. The experiments show that the improved algorithm, compared with other clustering algorithm in the application of feature optical flow, has a certain advantage. It can improve the clustering accuracy, give a more accurate target number, effectively track the sport vehicles. The experimental result of algorithm is ideal.