计算机应用 ›› 2018, Vol. 38 ›› Issue (5): 1432-1435.DOI: 10.11772/j.issn.1001-9081.2017102587

• 虚拟现实与多媒体计算 • 上一篇    下一篇

基于改进LBE特征的RGB-D显著性检测

袁泉1,2, 张建峰1, 伍立志1   

  1. 1. 重庆邮电大学 通信新技术应用研究中心, 重庆 400065;
    2. 重庆信科设计有限公司, 重庆 400065
  • 收稿日期:2017-10-31 修回日期:2017-12-20 出版日期:2018-05-10 发布日期:2018-05-24
  • 通讯作者: 张建峰
  • 作者简介:袁泉(1976-),男,湖南绥宁人,高级工程师,硕士,主要研究方向:数字图像处理、通信网络;张建峰(1992-),男,安徽宣城人,硕士研究生,主要研究方向:图像显著性检测;伍立志(1992-),男,湖北潜江人,硕士研究生,主要研究方向:视频目标跟踪。
  • 基金资助:
    重庆市研究生科研创新基金资助项目(CYS15166)。

RGB-D saliency detection based on improved local background enclosure feature

YUAN Quan1,2, ZHANG Jianfeng1, WU Lizhi1   

  1. 1. New Communication Technology Application Research Center, Chongqing University of Posts and Telecommunications, Chongqing 400065, China;
    2. Chongqing Information Technology Designing Company Limited, Chongqing 400065, China
  • Received:2017-10-31 Revised:2017-12-20 Online:2018-05-10 Published:2018-05-24
  • Contact: 张建峰
  • Supported by:
    This work is partially supported by the Chongqing Postgraduate Research Innovation Fund (CYS15166).

摘要: 针对LBE算法难以完整检测出结构复杂的目标和过度依赖深度信息的问题,提出一种基于改进LBE特征的RGB-D显著性检测算法。首先,对输入图像进行多级分割;然后,在各级分割图上计算LBE特征并融合,得到深度显著图;最后,利用色彩信息和先验信息对深度显著图进行矫正得到最终显著图。实验结果表明,改进算法与原始LBE算法相比在准确率上略有降低,在召回率上明显提升,得到的显著图更接近真实值。

关键词: 显著性检测, RGB-D图像, 多级分割, 深度信息

Abstract: Focusing on the issue that the LBE (Local Background Enclosure) algorithm is over dependent on depth information and difficult to fully detect the object with complex structure, a RGB-D saliency detection algorithm based on the improved LBE features was proposed. Firstly, a set of segmentations was obtained by multi-level segmentation. Then, the depth saliency map was obtained by computing and merging the LBE features on each level segmentation map. Finally, a saliency map was obtained by adjusting the depth saliency map with color information and prior information. The experimental results show that compared with LBE algorithm, the precision of the proposed algorithm is slightly decreased and the recall is significantly improved, and the obtained saliency maps are much more close to the true values.

Key words: saliency detection, RGB-D image, multi-level segmentation, depth information

中图分类号: