微型电脑应用2025，Vol.41Issue(3)：215-219,5.
基于多层级局部词袋模型的闭环检测算法
A Closed-loop Detection Algorithm Based on Multi-level Local Bag of Words Model
王晓卓 1王楷 1曹澍1
作者信息
- 1. 国网新疆电力有限公司信息通信公司,新疆,乌鲁木齐 830000
- 折叠
摘要
Abstract
The current closed-loop detection and analysis mainly rely on the global descriptor algorithm to get detection results,ignoring the local spatial characteristics of the image,which results in a lower F1 value in the closed-loop detection result.Therefore,a closed-loop detection algorithm based on multi-level local bag of words model is proposed.Scale invariant feature transformation algorithm is used to extract features contained in scene images,and improved fast orientation and rotation brie-fing(ORB)algorithm is used to construct feature descriptors.The multi-level local bag of words model that includes both low-level and high-level bag of words is established to treat each feature descriptor as a vocabulary representing local features within the bag of words model.After obtaining the frequency-inverse document frequency,the image similarity is calculated and closed-loop alternative frames are filtered out.Starting from the aspects of depth distance and color information,the salient fea-tures of closed-loop alternative frame images are extracted,and the salient feature maps are segmented.Alternative frames that do not meet the requirements are removed based on mutual information entropy to obtain the final closed-loop detection results.The experimental results show that the F1 value in the closed-loop detection result of the proposed algorithm is 0.96,which meets the expected design requirements of the closed-loop detection algorithm.关键词
词袋模型/特征描述子/聚类/候选帧/闭环检测Key words
bag of words model/feature descriptor/clustering/alternative frame/closed-loop detection分类
信息技术与安全科学引用本文复制引用
王晓卓,王楷,曹澍..基于多层级局部词袋模型的闭环检测算法[J].微型电脑应用,2025,41(3):215-219,5.
