复杂场景下有效的遗弃物检测方法-计算机工程与应用.PDFVIP

12 2014 ，50（16） Computer Engineering and Applications 计算机工程与应用 复杂场景下有效的遗弃物检测方法 1，2 1 叶芳芳 ，许 力 YE Fangfang1，2 , XU Li1 1.浙江大学 电气工程学院，杭州 310027 2 江苏大学 电气信息工程学院，江苏 镇江 210023 1.College of Electrical Engineering, Zhejiang University, Hangzhou 310027, China 2.College of Electronic and Information Engineering, Jiangsu University, Zhenjiang, Jiangsu 210023, China YE Fangfang, XU Li. Effective abandoned object detection method in complex scene. Computer Engineering and Applications, 2014, 50 （16）：12-15. Abstract ：An effective abandoned objects detection method is proposed for the purpose to solve occlusion and other static targets interference in the detection of abandoned objects. Gaussian Mixture Model （GMM ）is updated by selectively update strategy, thus the foreground including the moving targets and the stationary targets are detected. A dual-threshold method is used to obtain static objects, then aggregate evidence and an allowed occlusion time parameter are introduced with the purpose to reduce the false alarms and handle occlusions necessarily occurred. The interference of the pedestrians and vehicles are excluded by extracting the feature of stationary target blocks, and thus the abandoned objects are detected. The experi- mental results under different conditions demonstrate that the proposed method achieves a better performance under com - plex background condition. Key words ：Gaussian Mixture Model; occlusion; static object; abandoned objects detection; selectively update strategy 摘 要：为了解决遗弃物检测过程中的遮挡以及其他静止目标干扰问题，提出了一种有效的遗弃物检测方法。采用 选择性更新策略更新混合高斯背景模型，从而得到包含运动目标以及静止目标的前景；采用双阈值的方法提取静止 目标，并通过目标静止后累积的证据及允许遮挡时间参数处理虚警和遮挡问题。通过对静止目标区块的特征提取， 排除静止行人及车辆的干扰以实现遗弃物的检测。多场景下的实验分析表明，该方法在复杂背景条件下达到了良 好的检测性能。 关键词：混合高斯模型；遮挡；静止目标；遗弃物检测；选择性更新策略 文献标志码：A 中图分类号：TP391 doi ：10.3778/j.issn. 1002-8331.1312-0248 [1-3] [4- 11] 1 引言