摘要
介绍一种基于MATLAB的运动行人检测算法,可用于基于FPGA的智能监控系统。由于目前最常用的行人检测的基于统计学习的方法算法的复杂性等原因很难在FPGA上实现,在传统的运动检测算法帧差分法的基础上,提出了帧间相对模板匹配差分法,首先利用帧间差分法检测出运动目标区域,然后将该运动区域截取,并归一化,将截取到的运动区域分成若干个20*20矩形区域,分析车辆和行人运动的区别以及每个20*20矩形区域的水平投影和垂直投影的区别,将每个帧间20*20矩形区域进行相对模板匹配,达到运动行人检测的目的。实验表明该算法对运动行人进行检测有较好的效果,并且算法简单,适合基于FPGA的智能监控系统。
An algorithm of pedestrian detection on MATLAB is introduced,which can be used for intelligent monitoring system on FPGA.The complexity of algorithm,which is pedestrian detection on statistical learning,brings the difficulty of realization on FPGA.Based on the traditional motion detection,which is inter-frame difference,the inter-frame relative template matching difference algorithm is proposed in thsi paper.Firstly,detecting moving target region with the inter-frame difference, intercepting the normalized movement area,which is divided into several 20 ＊ 20 rectangles.
出处
《工业控制计算机》
2016年第6期88-89,共2页
Industrial Control Computer
