基于灰度特征和自适应阈值的虚拟背景提取研究-电子与信息学报.pdfVIP

第 37 卷第 2 期 电 子 与 信 息 学 报 Vol.37 No.2 2015 年 2 月 Journal of Electronics Information Technology Feb. 2015 基于灰度特征和自适应阈值的虚拟背景提取研究 庄哲民① 章聪友*① 杨金耀② 李芬兰① ①(汕头大学电子工程系 汕头 515063) ②(汕头市超声仪器研究所有限公司 汕头 515000) 摘 要：针对虚拟背景提取(Visual Background extractor,ViBe)算法在目标检测时容易出现鬼影和运动目标阴影的 缺点，该文提出了一种基于灰度特征和自适应阈值的 ViBe 背景建模改进方法。该算法首先利用 ViBe 算法进行背 景建模，得到前景目标，然后对前景目标进行灰度特征判断和自适应阈值比较，得到没有鬼影和运动目标阴影的运 动目标。实验结果表明，改进后的算法可以很好地弥补 ViBe 算法的不足，提高 ViBe 算法的识别准确率。 关键词：计算机视觉；运动目标检测；背景建模；虚拟背景提取(ViBe)；自适应阈值；灰度化特征 中图分类号： TP391.4 文献标识码： A 文章编号：1009-5896(2015)02-0346-07 DOI: 10.11999/JEIT140317 Investigation on Visual Background Extractor Based on Gray Feature and Adaptive Threshold ① ① ② ① Zhuang Zhe-min Zhang Cong-you Yang Jin-yao Li Fen-lan ① (Department of Electronic Engineering, Shantou University, Shantou 515063, China) ② (Shantou Institute of Ultrasonic Instruments Limited Corporation, Shantou 515000, China) Abstract: In order to solve the problem of the ghost and the shadow of moving object, an improved Visual Background extractor (ViBe) algorithm is proposed based on gray feature and adaptive threshold. The new method firstly applies the ViBe algorithm to obtain the foreground object, and then uses the gray feature judgment, as well as the adaptive threshold comparison in the foreground object to get the moving object without the ghost and the shadow. Experiments show that the improved algorithm results in better recognition accuracy. Key words: Computer vision; Moving object detection; Background