Color Based Pre-rank Categorization for Person Re-identification

DOI

10.2991/icca-16.2016.69How to use a DOI?

Keywords

Pre-rank categorizatio, Dense color features, LDA, mRMR

Abstract

The most addressed issue in person re-identification systems is to manage the performance under large gallery set and invariant appearance. In this study, considering these issues we proposed a color based pre-rank categorization (color basket) for person re-identification. In addition, coarse saliency dense color features were extracted to generate signature. For this purpose, we remove background and apply scale-invariant feature transform (SIFT) to get convex hull boundary to find the coarse saliency region. Furthermore, an image is divided into three horizontal strips and the dense color features are obtained. For training, incremental LDA (linear discriminant analysis) is utilized and extended with mRMR (minimum Redundancy Maximum Relevance) to handle the high computational cost. The proposed method is evaluated on three publically available datasets, namely i-LIDS, VIPeR and GRID. Moreover, a cumulative matching characteristic (CMC) curve is generated. The curve shows the propose strategy is good in dealing with the above-mentioned problems.

Copyright

© 2016, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Shah Jamal Hussain
AU  - Zonghai Chen
AU  - Rehman Saeed ur
AU  - Raza Mudassar
AU  - Mingqiang Lin
PY  - 2016/01
DA  - 2016/01
TI  - Color Based Pre-rank Categorization for Person Re-identification
BT  - Proceedings of the 2016 International Conference on Intelligent Control and Computer Application
PB  - Atlantis Press
SP  - 293
EP  - 296
SN  - 2352-538X
UR  - https://doi.org/10.2991/icca-16.2016.69
DO  - 10.2991/icca-16.2016.69
ID  - JamalHussain2016/01
ER  -