Palm vein recognition based on competitive code and DPL

Abstract

Dictionary learning is essentially a dimensionality reduction representation of large data sets and always attempts to learn the most pristine features behind the sample. In this paper, the palm vein feature is extracted with the Competitive code, then the Competition code feature of the palm vein is learned by a novel projective Dictionary Pair Learning (DPL) model for pattern classification tasks. Different from conventional Dictionary Learning (DL) methods, which learn a single synthesis dictionary, DPL learns jointly a synthesis dictionary and an analysis dictionary. Such a pair of dictionaries work together to perform representation and discrimination simultaneously. Compared with previous DL methods, DPL employs projective coding, which largely reduces the computational burden in learning and testing. The Competition code is used to extract the direction information of the palm vein. The above two methods are comprehensively applied to the palm veins. This paper uses the multispectral near-infrared palm vein image database of Hong Kong Polytech University for testing. Compared with the palm vein Competition code classification, the recognition rate of the palm vein Competition code feature learned by DPL is improved in some degree. When the palm vein features are extracted with filters in six different directions, the recognition rate of the palm vein Competitive code feature after learning with DPL is increased to 98.96%.