Video object matching across multiple non-overlapping camera views based on multi-feature fusion and incremental learning
作者：
Highlights：
• SIFT-based vocabulary tree vector and colour features for object representation.
• A modified kernel-based feature fusion for fast and accurate appearance modelling.
• An incremental general multicategory support vector machine for accurate real-time object matching.
• Only a small amount of samples for building classification model.
• Our method is superior to state-of-the-art classification-based matching approaches.
摘要
Highlights•SIFT-based vocabulary tree vector and colour features for object representation.•A modified kernel-based feature fusion for fast and accurate appearance modelling.•An incremental general multicategory support vector machine for accurate real-time object matching.•Only a small amount of samples for building classification model.•Our method is superior to state-of-the-art classification-based matching approaches.
论文关键词：Video object matching,Non-overlapping multi-camera views,CMFH,Incremental learning,Video surveillance system
论文评审过程：Received 28 January 2014, Revised 19 May 2014, Accepted 19 June 2014, Available online 28 June 2014.
论文官网地址：https://doi.org/10.1016/j.patcog.2014.06.019
