【摘要】 针对传统Vibe运动目标检测算法提取的目标存在Ghost区域、并且在目标有阴影时检测出的前景区域存在阴影的问题,本文提出了一种结合快速初始化背景建模和阴影去除的Vibe运动目标检测算法。首先采用改进的帧差背景建模方法快速初始化背景模型;然后利用Vibe算法找出当前帧内所有可能目标的前景区域;最后引入HSV模型去除前景目标区域中存在的阴影。实验结果表明,本算法可以有效消除Ghost区域,并且很好地去除前景目标区域里存在的阴影。更多还原

【Abstract】 To eliminate Ghost regions and shadows of targets which are extracted by traditional Vibe moving objects detection algorithm, an improved Vibe algorithm is proposed based on fast background initialization modeling and shadow removing.Firstly, improved frames difference background modeling method is applied to develop initialization background model rapidly. Secondly, all possible foreground target regions in current frame are located by using Vibe. Finally, all existed shadows of detected targets are removed according to the HSV model. Experimental result shows that, the improved algorithm can effectively eliminate Ghost regions in the process of target detection, and remove all shadows in foreground target regions.更多还原