Research of Feature Extraction Based on Improved LBP
2014 ◽
Vol 644-650
◽
pp. 1573-1576
Keyword(s):
This paper presents an improved Local Binary Pattern (LBP) operator for feature extraction f which considers both sign and magnitude information of the local difference of neighborhood and center pixels. The image is first divided into small blocks from which improved LBP histograms are extracted and concatenated into a single feature histogram. Then, the Principal Component Analysis (PCA) method is used to reduce feature dimensions. Finally, the recognition is performed by a nearest-neighbor classifier with Chi square statistic as the dissimilarity measurement. Experiments on AR face image databases by the leave-one-out (LOO) procedure illustrate that this method has higher recognition rate and more robust than the original LBP.
2019 ◽
Vol 8
(2)
◽
pp. 1-25
2014 ◽
Vol 568-570
◽
pp. 668-671
2021 ◽
pp. 517-544
2017 ◽
Vol 2017
◽
pp. 1-9
◽
2018 ◽
Vol 5
(1)
◽
pp. 8
◽