A novel palm print and palm vein joint recognition system based video is built. First of all, a novel registration and identification approach is proposed, and we can obtain the palm motion video by proposed system instead of static image obtained by traditional collection method. The approach allows the users simply sweep their palms across the acquisition platform without having to stop, which effectively enhances the affinity of authentication. A new strategy of fusing rotating videos with sweep videos to generate the registration feature set is proposed, which ensures the abundance and integrality of the register feature and enhances the robustness of the system for various palm postures in authentication. A cascaded fusion strategy is presented to improve the recognition speed of the registered users. We construct a palm print and palm vein database containing 1200 videos with motion blur from 100 palms and carry out a series simulations. The results show that the proposed system can achieve a low equal error rate of 1.51% within the expected time consumption of 915 ms, which desmonstrates the effectiveness and practicality of the new system.