Extraction for finger vein pattern based on PDE

In this paper we discussed how to apply partial differential equation (PDE) method to image denoising and vein extraction for the finger vein image, and proposed two new algorithms based on PDE. One is a new denoising model, which used the new diffusion function based on the P-M model and combined with fourth-order PDE model to transform the original model structure. Another is a new segmentation model, which introduced the image gradient information to the local binary fitting item, and made the new model have a good performance of both global optimization and local regulation. The performance of new models was verified by real low-quality finger vein image and the experiment results showed that the new denoising model could increase the Signal-to-Noise Ratio (SNR) up 5 dB and maintain the features of finger vein better, the new segmentation model can well segment the vein pattern.