In this paper, a novel multi-color feature selection method is proposed for person re-identification. Firstly, multi-color features, which consisting of HSV, LAB, RGB and nRnG color features, were extracted and concatenated into a whole feature vector. Then the D-optimal Partial Least Squares feature selection method was adopted to select an optimal feature subset that could minimize the variance of the regression model. Finally, an asymmetric distance model for similarity matching was utilized to observe distinctive features from a different perspective. Experimental results show that rank 1 performance of the proposed method were 48.67%, 63.12% and 65.04% respectively on the VIPeR, Prid_450s and CUHK01 databases, which have achieved state-of-art performances.
A Novel Multi-Color Feature Selection Method for Person Re-identification / Y. Zhai, L. Chen, L. Cao, W. Deng, Y. Zhi, J. Gan, Y. Xu, J. Zeng, V. Piuri, F. Scotti - In: 2018 14th IEEE International Conference on Signal Processing (ICSP)[s.l] : IEEE, 2018. - ISBN 9781538646731. - pp. 1087-1092 (( Intervento presentato al 14. convegno International Conference on Signal Processing (ICSP) tenutosi a Beijing nel 2018.
A Novel Multi-Color Feature Selection Method for Person Re-identification
V. Piuri;F. Scotti
2018
Abstract
In this paper, a novel multi-color feature selection method is proposed for person re-identification. Firstly, multi-color features, which consisting of HSV, LAB, RGB and nRnG color features, were extracted and concatenated into a whole feature vector. Then the D-optimal Partial Least Squares feature selection method was adopted to select an optimal feature subset that could minimize the variance of the regression model. Finally, an asymmetric distance model for similarity matching was utilized to observe distinctive features from a different perspective. Experimental results show that rank 1 performance of the proposed method were 48.67%, 63.12% and 65.04% respectively on the VIPeR, Prid_450s and CUHK01 databases, which have achieved state-of-art performances.File | Dimensione | Formato | |
---|---|---|---|
p1.pdf
accesso aperto
Tipologia:
Post-print, accepted manuscript ecc. (versione accettata dall'editore)
Dimensione
524.05 kB
Formato
Adobe PDF
|
524.05 kB | Adobe PDF | Visualizza/Apri |
08652340.pdf
accesso riservato
Tipologia:
Publisher's version/PDF
Dimensione
1.34 MB
Formato
Adobe PDF
|
1.34 MB | Adobe PDF | Visualizza/Apri Richiedi una copia |
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.