一种改进的运动目标检测与跟踪方法  

Improved Moving Object Detection and Tracking Method

作　　者：白晓方 杨卫[2] 陈佩珩[1] 

机构地区：[1]中北大学电子测试技术重点实验室,山西太原030051 [2]中北大学仪器科学与动态测试教育部重点实验室,山西太原030051

出　　处：《电视技术》2014年第1期180-182,186,共4页Video Engineering

摘　　要：为了解决传统的帧间差分法提取的运动区域不完整,且容易造成空洞等问题,将其改进为基于三帧差分的背景差分法检测运动目标。在跟踪方法上先对目标进行边界扫描,利用链码原理记录目标边界,再采用质心跟踪法。最后将算法移植到TMS320DM642的硬件系统上进行精度测试。实验表明改进后的方法使运动目标检测算法的性能得到很大的优化,可以有效、实时地完成对运动目标的检测与跟踪,实现不丢帧、跟踪精确。In order to solve the traditional inter- frame difference method to extract the motion region which is not completed and likely to cause prob- lems such as empty, it is improved based on three frames differential background difference method to detect moving targets. In the tracking method first- ly scan the target＇ s boundary,object boundaries is recorded using the theory of chain code, then the centroid tracking method is adopted. Finally,the al- gorithm is ported to the TMS320DM642 hardware system to test for the accuracy. The experiments show that the improved method makes the performance of the moving target detection algorithm to get a lot of optimization ,and can real- time complete the moving target detection and tracking with no frame loss and tracking precision.

关 键 词：三帧差分法 背景差分法 目标边界 质心跟踪法 TMS320DM642 

分 类 号：TN911.73[电子电信—通信与信息系统]