Inverse nonnegative local coordinate factorization for visual tracking

Recently, nonnegative matrix factorization (NMF) with part based representation has been widely used for appearance modelling in visual tracking. Unfortunately, not all the targets can be successfully decomposed as "parts" unless some rigorous conditions are satisfied. To avoid this proble...

Full description

Bibliographic Details
Main Authors: Liu, Fanghui, Zhou, Tao, Gong, Chen, Fu, Keren, Bai, Li, Yang, Jie
Format: Article
Published: Institute of Electrical and Electronics Engineers 2017
Subjects:
Online Access:https://eprints.nottingham.ac.uk/44634/
_version_ 1848796961730199552
author Liu, Fanghui
Zhou, Tao
Gong, Chen
Fu, Keren
Bai, Li
Yang, Jie
author_facet Liu, Fanghui
Zhou, Tao
Gong, Chen
Fu, Keren
Bai, Li
Yang, Jie
author_sort Liu, Fanghui
building Nottingham Research Data Repository
collection Online Access
description Recently, nonnegative matrix factorization (NMF) with part based representation has been widely used for appearance modelling in visual tracking. Unfortunately, not all the targets can be successfully decomposed as "parts" unless some rigorous conditions are satisfied. To avoid this problem, this paper introduces NMF's variants into the visual tracking framework in the view of data clustering for appearance modelling. Firstly, an initial target appearance model based on NMF is proposed to describe the target's appearance with the incorporated local coordinate factorization constraint, orthogonality of the bases, and L1,1 norm regularized sparse residual error constraint. Secondly, an inverse NMF model is proposed, in which each learned base vector is regarded as a clustering center in a low-dimensional subspace. Potential target samples (from the foreground) will be clustered around base vectors; while the candidate samples (from the background) are very likely to spread irregularly over the entire clustering space. Such difference can be fully exploited by the inverse NMF model to produce more discriminative encoding vectors than the conventional NMF method. Further, incremental updating model is introduced into the tracking framework for online updating the initial appearance model. Experiments on Object Tracking Benchmark (OTB) suggest that our tracker is able to achieve promising performance when compared to some state-of-the-art methods in deformation, occlusion, and other challenging situations.
first_indexed 2025-11-14T19:56:18Z
format Article
id nottingham-44634
institution University of Nottingham Malaysia Campus
institution_category Local University
last_indexed 2025-11-14T19:56:18Z
publishDate 2017
publisher Institute of Electrical and Electronics Engineers
recordtype eprints
repository_type Digital Repository
spelling nottingham-446342020-05-04T18:43:17Z https://eprints.nottingham.ac.uk/44634/ Inverse nonnegative local coordinate factorization for visual tracking Liu, Fanghui Zhou, Tao Gong, Chen Fu, Keren Bai, Li Yang, Jie Recently, nonnegative matrix factorization (NMF) with part based representation has been widely used for appearance modelling in visual tracking. Unfortunately, not all the targets can be successfully decomposed as "parts" unless some rigorous conditions are satisfied. To avoid this problem, this paper introduces NMF's variants into the visual tracking framework in the view of data clustering for appearance modelling. Firstly, an initial target appearance model based on NMF is proposed to describe the target's appearance with the incorporated local coordinate factorization constraint, orthogonality of the bases, and L1,1 norm regularized sparse residual error constraint. Secondly, an inverse NMF model is proposed, in which each learned base vector is regarded as a clustering center in a low-dimensional subspace. Potential target samples (from the foreground) will be clustered around base vectors; while the candidate samples (from the background) are very likely to spread irregularly over the entire clustering space. Such difference can be fully exploited by the inverse NMF model to produce more discriminative encoding vectors than the conventional NMF method. Further, incremental updating model is introduced into the tracking framework for online updating the initial appearance model. Experiments on Object Tracking Benchmark (OTB) suggest that our tracker is able to achieve promising performance when compared to some state-of-the-art methods in deformation, occlusion, and other challenging situations. Institute of Electrical and Electronics Engineers 2017-04-28 Article PeerReviewed Liu, Fanghui, Zhou, Tao, Gong, Chen, Fu, Keren, Bai, Li and Yang, Jie (2017) Inverse nonnegative local coordinate factorization for visual tracking. IEEE Transactions on Circuits and Systems for Video Technology . ISSN 1051-8215 Local coordinate constraint inverse nonnegative matrix factorization incremental update visual tracking http://ieeexplore.ieee.org/document/7914620/ doi:10.1109/TCSVT.2017.2699676 doi:10.1109/TCSVT.2017.2699676
spellingShingle Local coordinate constraint
inverse nonnegative matrix factorization
incremental update
visual tracking
Liu, Fanghui
Zhou, Tao
Gong, Chen
Fu, Keren
Bai, Li
Yang, Jie
Inverse nonnegative local coordinate factorization for visual tracking
title Inverse nonnegative local coordinate factorization for visual tracking
title_full Inverse nonnegative local coordinate factorization for visual tracking
title_fullStr Inverse nonnegative local coordinate factorization for visual tracking
title_full_unstemmed Inverse nonnegative local coordinate factorization for visual tracking
title_short Inverse nonnegative local coordinate factorization for visual tracking
title_sort inverse nonnegative local coordinate factorization for visual tracking
topic Local coordinate constraint
inverse nonnegative matrix factorization
incremental update
visual tracking
url https://eprints.nottingham.ac.uk/44634/
https://eprints.nottingham.ac.uk/44634/
https://eprints.nottingham.ac.uk/44634/