Palmprint Identification using Boosting Local Binary Pattern

Local binary pattern (LBP) is a powerful texture descriptor that is gray-scale and rotation invariant according to T. Ojala et al. (2002). Because texture is one of the most clearly observable features in low-resolution palmprint images, we think local binary pattern based features are very discriminative for palmprint identification. In this paper, we propose a palmprint identification approach using boosted local binary pattern based classifiers. The palmprint area is scanned with a scalable sub-window from which local binary pattern histograms are extracted to represent the local features of a palmprint image. The multi-class problem is transformed into a two-class one of intra- and extra-class by classifying every pair of palmprint images as intra-class or extra-class ones in the work of B. Moghaddam et al. (1996). We use the AdaBoost algorithm in the work of Y. Freund and R.E. Schapire (1997) to select those sub-windows that are more discriminative for classification. Weak classifiers are constructed based on the Chi square distance between two corresponding local binary pattern histograms. Experiments on the UST-HK palmprint database show competitive performance