TY - GEN
T1 - Novel gabor-PHOG features for object and scene image classification
AU - Sinha, Atreyee
AU - Banerji, Sugata
AU - Liu, Chengjun
N1 - Copyright:
Copyright 2012 Elsevier B.V., All rights reserved.
PY - 2012
Y1 - 2012
N2 - A new Gabor-PHOG (GPHOG) descriptor is first introduced in this paper for image feature extraction by concatenating the Pyramid of Histograms of Oriented Gradients (PHOG) of all the local Gabor filtered images. Next, a comparative assessment of the classification performance of the GPHOG descriptor is made in six different color spaces, namely the RGB, HSV, YCbCr, oRGB, DCS and YIQ color spaces, to propose the novel YIQ-GPHOG and the YCbCr-GPHOG feature vectors that perform well on different object and scene image categories. Third, a novel Fused Color GPHOG (FC-GPHOG) feature is presented by integrating the PCA features of the six color GPHOG descriptors for object and scene image classification, with applications to image search and retrieval. Finally, the Enhanced Fisher Model (EFM) is applied for discriminatory feature extraction and the nearest neighbor classification rule is used for image classification. The effectiveness of the proposed feature vectors for image classification is evaluated using two grand challenge datasets, namely the Caltech 256 dataset and the MIT Scene dataset.
AB - A new Gabor-PHOG (GPHOG) descriptor is first introduced in this paper for image feature extraction by concatenating the Pyramid of Histograms of Oriented Gradients (PHOG) of all the local Gabor filtered images. Next, a comparative assessment of the classification performance of the GPHOG descriptor is made in six different color spaces, namely the RGB, HSV, YCbCr, oRGB, DCS and YIQ color spaces, to propose the novel YIQ-GPHOG and the YCbCr-GPHOG feature vectors that perform well on different object and scene image categories. Third, a novel Fused Color GPHOG (FC-GPHOG) feature is presented by integrating the PCA features of the six color GPHOG descriptors for object and scene image classification, with applications to image search and retrieval. Finally, the Enhanced Fisher Model (EFM) is applied for discriminatory feature extraction and the nearest neighbor classification rule is used for image classification. The effectiveness of the proposed feature vectors for image classification is evaluated using two grand challenge datasets, namely the Caltech 256 dataset and the MIT Scene dataset.
KW - EFM
KW - FC-GPHOG
KW - Gabor-PHOG (GPHOG)
KW - PCA
KW - YCbCr-GPHOG
KW - YIQ-GPHOG
KW - color spaces
KW - image search
UR - http://www.scopus.com/inward/record.url?scp=84868117020&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84868117020&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-34166-3_64
DO - 10.1007/978-3-642-34166-3_64
M3 - Conference contribution
AN - SCOPUS:84868117020
SN - 9783642341656
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 584
EP - 592
BT - Structural, Syntactic, and Statistical Pattern Recognition - Joint IAPR International Workshop, SSPR and SPR 2012, Proceedings
T2 - Joint IAPR International Workshops on Structural and Syntactic PatternRecognition, SSPR 2012 and Statistical Techniques in Pattern Recognition,SPR 2012
Y2 - 7 November 2012 through 9 November 2012
ER -