«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

28

期数:

2018年06期

页码:

184-187

栏目:

应用开发研究

出版日期:

2018-06-10

文章信息/Info

Title:

Research on Removed Object Detection Technique Based on Binocular Stereo Vision

文章编号:

1673-629X(2018)06-0184-04

作者:

张爱升; 张艳彬

南京邮电大学 通信与信息工程学院,江苏 南京 210003

Author(s):

ZHANG Ai-sheng; ZHANG Yan-bin

 School of Telecommunications &Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210003,China

关键词:

取走物检测; 双目立体视觉; 运动前景; 彩色图像; 深度图像

Keywords:

removed object detection; binocular stereo vision; motion foreground; color image; depth image

分类号:

TP391

DOI:

10.3969/ j. issn.1673-629X.2018.06.041

文献标志码:

A

摘要:

双目立体视觉技术应用在视频监控领域能使其更加智能化。 对此,提出了一种借助双目立体视觉技术进行取走物检测的方法。 一方面,在深度图像中获取深度变化信息,对深度变化情况进行标记,分为三种区域:深度明显增加的区域、深度明显减少的区域和深度基本不变的区域。 另一方面,在彩色图像中利用改进的 surendra 算法提取运动前景,即对深度增加或基本不变的区域进行背景更新,再利用背景减除法得到运动前景,在前景区域中,符合深度增加或深度基本不变的前景区域即为取走物区域。 最后在光线合适的室内环境中,用提出的取走物检测方法进行了实验,结果表明,该方法能有效地检测出取走物且能显露出被取走物品的大体形状,场景适应性良好。

Abstract:

Binocular stereo vision technology is used in video surveillance and can make it more intellectualized. For this,we propose a object removal detection method based on binocular stereo vision. For one thing,the depth variation information is obtained in the depth image and marked into three areas including the areas with significantly increased depth,the areas with significantly reduced depth and that with little depth variance. For another,the motion foreground is extracted by improved surendra algorithm in the color image,that is,to update the background in the area of depth increase or the little variance,and then using the background subtraction method to obtain the foreground of the motion. At last,the foreground region with depth increase or depth little variance is the removed object region. The experiment is carried out with the method of removed object detection in the indoor environment with appropriate light,which shows that it can effectively detect the removal with general shape and the scene adaptability is better.

常用功能

更新日期/Last Update: 2018-08-22