基于达芬奇的运动目标检测系统*

作者：孙友统，董辉*，俞立 日期：2010-06-28/span> 浏览：2949 查看PDF文档

基于达芬奇的运动目标检测系统*
孙友统，董辉*，俞立
（浙江工业大学 信息工程学院，浙江 杭州 310032）

摘要：针对视频监控领域，利用达芬奇双核(ARM+DSP)处理芯片DM6446的高处理性能，设计了运动目标检测系统。在系统硬件上围绕DM6446进行了设计和实现；在用户软件上实现了视频的录制、存储、回放和运动目标检测等功能。ARM核应用程序采用多线程设计，DSP核运行目标检测算法。算法采用改进的颜色直方图方法，利用高斯模型计算颜色的背景概率，通过比较背景和前景的概率，达到了运动目标检测的目的。测试结果表明，算法简单可行，具有实时性，且软硬件工作良好，能够满足实时采集与图像处理的要求。
关键词：达芬奇；运动目标检测；直方图；高斯模型
中图分类号：TP391.41文献标识码：A文章编号：1001-4551(2010)06-0100-04

Motion detection system based on Davinci

SUN Youtong, DONG Hui, YU Li
（College of Information Engineering, Zhejiang University of Technology, Hangzhou 310032, China）

Abstract: For video surveillance area, motion detection system was designed, taking the advantage of high processing performance of Davinci dualcore (ARM+DSP) chip DM6446. Hardware implementation were carried out with DM6446 processor. Video recording, storage, playback, and moving target detection and other functions were implemented for the application software. Multithreading program was implemented on the ARM core, and the moving target detection algorithm ran on the DSP core. An improved method of color histogram was adopted by this algorithm, using the Gauss model for calculating the background probability of color, achieving the purpose of moving target detection by comparing the probability of background and prospects. Test results show that, algorithm is simple and feasible, the hardware and software work well. And it meets the realtime acquisition and image processing requirements.
Key words: Davinci; moving target detection; histogram; Gauss model