基于达芬奇平台的运动目标检测研究

Moving Object Detection Algorithm Design Based on DaV inci Platform

对运动目标检测问题,提出了基于达芬奇平台的混合高斯模型背景差分运动目标检测算法,解决了达芬奇平台下硬件资源欠缺的问题。该算法使用背景差分法检测运动目标,采用混合高斯模型进行实时更新背景。实验结果表明,采用混合高斯模型背景差分法检测更新背景,较好适应外界环境的变化,能够在毫米级反应时间内准确地检测出运动目标,满足实时性要求。

This paper proposes a moving object detection algotithm under DaVinci Platform based on background substraction moving object detection. The algorithm resolves the problenl of lack of hardware resources based on DaVmci Platform, Which uses background subtraction method to detect moving objects, and updates background with Gaussian nfixture model in real-time. Experimental results show that the subtraction background based on Gaussian mixture model can better adapt to changes in the external environment, accurately detect moving object within 50ms, which satisfies the requirements of real-time systems.