半导体光电, 2018, 39 (2): 260, 网络出版: 2018-05-29  

一种改进的基于ViBe的运动目标检测方法

An Improved Detection Method of Moving Objects Based on ViBe
作者单位
北京航空航天大学 仪器科学与光电工程学院, 北京100191
摘要
ViBe运动检测算法消除鬼影的速度慢, 并且在分类像素点的时候由于距离阈值固定导致检测结果的准确率下降。文章对ViBe算法存在的这两个问题做出改进。样本的选取范围由8邻域扩大到24邻域, 避免选择重复的样本, 加快鬼影的消除过程; 在像素分类的时候根据背景动力学自适应地确定距离阈值, 增强算法抗背景干扰的能力, 并选用2组视频来验证改进后的算法。从检测结果可以看出, 改进后的算法去除鬼影的速度更快, 并且检测结果更加准确。
Abstract
ViBe moving detection algorithm eliminates the ghost slowly, and the detection precision will be declined when classifing the pixels due to fixed distance threshold. Thus an improved method is proposed aiming at the two problems. The selection range of samples is expanded from the 8-connected neighborhood to the 24-connected neighborhood of each pixel, avoiding the selection of repetitive samples, and thus accelerating the elimination process of the ghost. According to background dynamics, the distance threshold is determined adaptively when each pixel is classified as a foreground pixel or a background pixel. It enhances the ability of resisting background interference. The improved algorithm was tested with two datasets. Test results indicate that the improved algorithm is able to eliminate the ghost shadows more rapidly, and the detection results are more accurate.

欧阳梓标, 牛燕雄, 谢朋言. 一种改进的基于ViBe的运动目标检测方法[J]. 半导体光电, 2018, 39(2): 260. OUYANG Zibiao, NIU Yanxiong, XIE Pengyan. An Improved Detection Method of Moving Objects Based on ViBe[J]. Semiconductor Optoelectronics, 2018, 39(2): 260.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!