计算机应用 ›› 2014, Vol. 34 ›› Issue (5): 1463-1466.DOI: 10.11772/j.issn.1001-9081.2014.05.1463

• 虚拟现实与数字媒体 • 上一篇    下一篇

基于前景分割的目标实时检测方法

牛杰1,2,卜雄洙2,钱堃3   

  1. 1. 常州信息职业技术学院 电子与电气工程学院,江苏 常州 213164;
    2. 南京理工大学 机械工程学院,南京 210094
    3. 东南大学 自动化学院,南京 210096
  • 收稿日期:2013-11-20 修回日期:2013-12-29 出版日期:2014-05-01 发布日期:2014-05-30
  • 通讯作者: 牛杰
  • 作者简介:牛杰(1983-),男,江苏淮安人,讲师,博士研究生,主要研究方向:机器视觉、智能机器人;卜雄洙(1966-),男,吉林延吉人,教授,博士生导师,博士,主要研究方向:智能测控系统、机器视觉;钱堃(1982-),男,江苏南京人,讲师,博士,主要研究方向:服务机器人导航、人机交互。
  • 基金资助:

    国家自然科学基金资助项目;江苏省大学生实践创新训练计划项目;常州市应用基础研究计划资助项目

Real-time object detection method based on foreground segmentation

NIU Jie1,2,BU Xiongzhu2,QIAN Kun3   

  1. 1. School of Electrical and Electronic Engineering, Changzhou College of Information Technology, Changzhou Jiangsu 213164, China;
    2. School of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing Jiangsu 210094, China;
    3. School of Automation, Southeast University, Nanjing Jiangsu 210096, China
  • Received:2013-11-20 Revised:2013-12-29 Online:2014-05-01 Published:2014-05-30
  • Contact: NIU Jie

摘要:

针对基于单一颜色信息的目标分割算法易受光线因素影响的问题,提出一种颜色及深度信息融合进行前景分割的目标实时检测方法。采用Kinect传感器采集低成本深度(RGB-D)图像,利用改进的ViBe算法及多帧差分法分别对于RGB以及深度图像进行建模。前景分割后,利用选取基准(SC)融合策略优化目标结果,然后通过rg Chromaticity颜色模型计算前景区域直方图信息并与模板匹配完成目标标记。实验结果表明,该方法对于环境光线及噪声干扰具有一定的鲁棒性,对于ViBe算法中背景前景同色误检及“鬼影”现象,对于深度图像分割中前景背景距离过近而造成误检现象都有很好的识别效果。

Abstract:

Aiming at the problem that object segmentation algorithms based on single color information are very sensitive to the changes on lighting, a novel approach to detect target based on the fusion of color and depth information was proposed. Firstly, improved Visual Background Extractor (ViBe) and multiple-frame subtraction algorithm were used to establish models for RGB and depth images which captured by Kinect senor respectively. Then, strategy of Selection Criterion (SC) was used to optimize segmentation results. Lastly, most likely target was labeled by calculating similar degree between foreground and template in the rg chromaticity space. The experimental results demonstrate that the proposed method exhibit a certain degree of resilience to light disturbance and noise, and it can overcome the disadvantages of single RGB based algorithms effectively.

中图分类号: