期刊导航

论文摘要

一种鬼影及边缘传播抑制的ViBe目标检测算法

A ViBe target detection algorithm with ghost and edge propagation suppress

作者:陈金戈(四川大学机械工程学院);龙伟(四川大学机械工程学院);李炎炎(四川大学机械工程学院);俞晓红(四川大学机械工程学院);杜玉龙(四川大学机械工程学院)

Author:Chenjinge(School of mechanical engineering sichuan university);Long Wei(School of mechanical engineering sichuan university);Li Yanyan(School of mechanical engineering sichuan university);Yu Xiaohong(School of mechanical engineering sichuan university);Du Yulong(School of mechanical engineering sichuan university)

收稿日期:2019-10-11          年卷(期)页码:2020,57(2):284-288

期刊名称:四川大学学报: 自然科学版

Journal Name:Journal of Sichuan University (Natural Science Edition)

关键字:目标提取; ViBe; 阈值分割; 边缘检测

Key words:Object extraction; ViBe; Threshold segmentation; Edge extraction

基金项目:中国博士后科学基金(198606); 四川大学博士后中央财政专项研究基金(2018SCU12065)

中文摘要

针对在目标运动检测中ViBe算法容易产生鬼影、缓慢移动的目标加速融入背景样本模型等问题,提出一种改进的ViBe算法,首先在使用ViBe算法检测到运动目标后,利用迭代自组织分析算法的阈值分割来进行鬼影判别;其次利用改进的Canny算子获得图像的边缘特征,当ViBe算法在更新目标边缘的背景样本模型时不对其邻域的样本模型进行更新,从而使得缓慢移动的目标融入背景样本模型的时间得以延长;最后通过形态学处理获得完整的运动目标。实验结果表明,与传统ViBe算法相比,本文算法在有鬼影的情况下能更快获得较高的检测准确率,在有缓慢移动目标的情况下准确率的下降时间更晚。

英文摘要

Vibe algorithm is easy to produce ghost and slow moving objects are easy to integrate into the background sample model. A new algorithm was proposed,first of all, using an iterative selection method to distinguish ghost shadow after detecting moving target with ViBe algorithm. Secondly, the edge feature of the image is obtained by using the improved Canny operator, and when the ViBe algorithm updates the background pixel model at the edge of the target, it does not update the pixel model of its neighborhood, so that the time of the slow moving target into the background pixel model is extended, and finally the complete moving target is obtained by morphological processing. Experimental results show that compared with the traditional ViBe algorithm, this algorithm can obtain higher detection accuracy more quickly in the case of ghosts, and the accuracy decreases later in the case of slow moving targets.

关闭

Copyright © 2020四川大学期刊社 版权所有.

地址:成都市一环路南一段24号

邮编:610065