江汉大学学报（自然科学版）

在人体运动目标检测部分,根据室内监控的特点,提出了先采用统计平均法获取室内背景,然后选用背景减除法对运动目标进行检测,接着利用改进的区域增长法对属于同一目标的像素区域进行合并,判断并提取人体目标。在人体跟踪部分,使用卡尔曼滤波器预测目标参数,再根据预测参数跟踪人体目标,得到行人的运动轨迹。利用Matlab/Simulink对整个系统进行建模仿真,并不断修正算法的参数。通过室内实际采集的几个视频序列进行试验,该算法能够正确地检测并且跟踪运动的人体。

Detecting and tracking moving object is one of the key technologies in analyzing human movement,and is also an important part of intelligent monitoring system.In the part of detecting,according to characteristics of indoor monitoring,this paper puts forward the method obtaining in indoor background by statistical averaging method,and then uses background subtraction method to detect indoor moving object,and uses an improved regional growth method to connect the areas which belong to the same moving object region.In the part of tracking,this paper uses Kalman filter to predict the location′s parameters of the object.According to the predicted parameters,the system tracks human target and obtains human moving track.The system is modeled and simulated through Matlab/Simulink,and the algorithm parameters are revised accordingly.Through testing several indoor video sequences,it is shown that the algorithm can detect and track human movement correctly.