TY - GEN
T1 - Object and scene image classification using unconventional color descriptors
AU - Banerji, Sugata
AU - Sinha, Atreyee
AU - Liu, Chengjun
N1 - Copyright:
Copyright 2013 Elsevier B.V., All rights reserved.
PY - 2012
Y1 - 2012
N2 - This paper presents novel color, texture and shape descriptors for scene and object image classification and evaluates their performance in unconventional color spaces. First, a new three dimensional Local Binary Pattern (3D-LBP) descriptor is proposed for color and texture feature extraction. Second, a novel color HWML (HOG of Wavelet of Multiplanar LBP) descriptor is derived by computing the histogram of the orientation gradients (HOG) of the Haar wavelet transformation of the original image and the 3D-LBP images. Third, these descriptors are generated in the unconventional color spaces like oRGB, I1I2I3, uncorrelated and discriminating color spaces to improve performance over conventional color spaces like RGB and HSV. Fourth, the Enhanced Fisher Model (EFM) is applied for discriminatory feature extraction and the nearest neighbor classification rule is used for image classification. Finally, the Caltech 256 object categories database and the MFT scene dataset are used to demonstrate the feasibility of the proposed new methods.
AB - This paper presents novel color, texture and shape descriptors for scene and object image classification and evaluates their performance in unconventional color spaces. First, a new three dimensional Local Binary Pattern (3D-LBP) descriptor is proposed for color and texture feature extraction. Second, a novel color HWML (HOG of Wavelet of Multiplanar LBP) descriptor is derived by computing the histogram of the orientation gradients (HOG) of the Haar wavelet transformation of the original image and the 3D-LBP images. Third, these descriptors are generated in the unconventional color spaces like oRGB, I1I2I3, uncorrelated and discriminating color spaces to improve performance over conventional color spaces like RGB and HSV. Fourth, the Enhanced Fisher Model (EFM) is applied for discriminatory feature extraction and the nearest neighbor classification rule is used for image classification. Finally, the Caltech 256 object categories database and the MFT scene dataset are used to demonstrate the feasibility of the proposed new methods.
KW - Enhanced Fisher Model (EFM)
KW - Haar wavelets
KW - Image search
KW - Scene classification
KW - The HOG of wavelet of multiplanar LBP (HWML) descriptor
KW - The fused-HWML descriptor
UR - http://www.scopus.com/inward/record.url?scp=84873299586&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84873299586&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:84873299586
SN - 9781601322258
T3 - Proceedings of the 2012 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2012
SP - 695
EP - 701
BT - Proceedings of the 2012 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2012
T2 - 2012 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2012
Y2 - 16 July 2012 through 19 July 2012
ER -