A Method of Color Palmprint recognition Based on DenseNet Integrate Spatial and Channel Features

Palmprint recognition is a novel biometric recognition technology. The palmprint image has rich texture features, but traditional methods are difficult to accurately characterize them, and most of the existing palmprint image recognition is gray.This paper proposes a color palmprint recognition method based on improved convolutional neural network, and we design the PRSCNet(Palmprint recognition fusing spatial and channel features network) model of this network according to the insertion position and number of SE modules.We conducted experiments on the palmprint dataset provided by the Hong Kong Polytechnic University.And the palmprint database is preprocessed to transform into color palmprint database. The accuracy of recognition reached 99.2%. Therefore, it is proved that PRSCNet can recognize color palmprint accurately and has great application prospect.