| 移动机器人闭环检测的视觉字典树金字塔TF-IDF得分匹配方法 |
| |
| 引用本文： | 李博,杨丹,邓林. 移动机器人闭环检测的视觉字典树金字塔TF-IDF得分匹配方法[J]. 自动化学报, 2011, 37(6): 665-673. DOI: 10.3724/SP.J.1004.2011.00665 |
| |
| 作者姓名： | 李博 杨丹 邓林 |
| |
| 作者单位： | 1.重庆大学计算机学院 重庆 400044; |
| |
| 基金项目： | 国家自然科学基金(60975015); 中央高校基本科研业务费专项资金(CDJXS11181162,CDJXS10181133); 教育部博士点基金(20090191110023); 重庆市重点科技攻关项目(CSTC2009AB2230)资助~~ |
| |
| 摘 要： | 针对移动机器人视觉闭环检测中,基于视觉字典本的场景外观表征性能受制于有限单词个数以及算法效率低的不足,本文对机器人视觉特征分层量化,构建视觉字典树, 计算树节点的TF-IDF熵作为对应视觉单词的权重,生成图像--单词逆向文档索引.为消除视觉字典本的单尺度量化误差,并克服基于字典树投影路径的平面匹配模式中不 区分不同层次节点的区分度对闭环检测的影响,本文融合字典树低层单词的强表征性和高层单词的强鲁棒性,提出由下而上逐层计算图像间相似性增量的金字塔得分匹 配方法.将不同时刻相似性大于阈值的图像位置提取为候选闭环,通过后验确认操作剔除误正闭环.在移动机器人视觉闭环检测实验中,本文算法提高了图像相似性计算 的效率和准确性,提高了闭环检测的准确率和召回率.
|
| 关 键 词： | 闭环检测 视觉字典树 TF-IDF得分准则 金字塔匹配 |
| 收稿时间： | 2010-09-21 |
Visual Vocabulary Tree with Pyramid TF-IDF Scoring Match Scheme for Loop Closure Detection |
| |
| LI Bo YANG Dan DENG Lin.College of Computer Science,Chongqing University,Chongqing .School of Software Engineering,Chongqing .College of Mathematics , Statistics,Chongqing. Visual Vocabulary Tree with Pyramid TF-IDF Scoring Match Scheme for Loop Closure Detection[J]. Acta Automatica Sinica, 2011, 37(6): 665-673. DOI: 10.3724/SP.J.1004.2011.00665 |
| |
| Authors: | LI Bo YANG Dan DENG Lin.College of Computer Science Chongqing University Chongqing .School of Software Engineering Chongqing .College of Mathematics Statistics Chongqing |
| |
| Affiliation: | 1.College of Computer Science, Chongqing University, Chong-qing 400044;2.School of Software Engineering, Chongqing University, Chongqing 400044;3.College of Mathematics and Statistics, Chongqing University, Chongqing 400044 |
| |
| Abstract: | The performance of visual environment modeling in appearance-based robot loop closure detection by using conventional vocabulary is restricted by limited number of visual words and high computational cost.We construct a visual vocabulary tree by clustering the visual features hierarchically captured by a mobile robot.The TF-IDF entropy for each node is computed and is treated as the weight of each visual word,and the inverted index of image-word is exploited. To avoid the quantization error of single scale ... |
| |
| Keywords: | Loop closure detection visual vocabulary tree TF-IDF scoring scheme pyramid match |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《自动化学报》浏览原始摘要信息 |
|
点击此处可从《自动化学报》下载全文 |
|
