摘要
给出了一种基于DM6446处理器的视频运动车辆检测系统的技术实现方法,介绍了系统的硬件结构及软件设计思路。结合DM6446双核处理器的特点,给出了处理器ARM端与DSP端通信及双缓冲区切换的具体方法。在视频运动车辆检测算法上,采用差异积累法对DM6446实时获取的交通视频数据进行背景建模,用背景差法检测车辆运动区域,再结合Otsu阈值化、形态学滤波及区域生长等算法,最终在DM6446硬件平台上实现视频运动车辆的实时检测。实验结果表明,该系统具有良好检测效果。
This paper gives a method of realizing a video moving vehicle detecting system based on DM6446 processor, and describes the system's hardware architecture and software design ideas. The methods of ARM-side and DSP-side communication and double-buffer switch are given combining the characteristic of dual-core of DM6446 processor. Background modeling is made on the traffic video data from DM6446 via difference accumulation. Background subtraction is used to detect vehicle movement areas. With the adoption of other algorithms such as Otsu, morphological filtering and region growing, the system is finally completed on this hardware platform. Experimental results show that the system is good in moving vehicle detection.
出处
《计算机系统应用》
2010年第7期10-14,共5页
Computer Systems & Applications
基金
浙江省重大科技专项(2008C13076)
浙江省自然科学基金(Y1080883)
杭州电子科技大学校立科研项目(KYS055609080)