Noncontact Multi-Modal Biological Recognition Based on Three-Dimensional Integrated Learning

Single biometric recognition has become one of the most popular and promising method for personal verification. However, further work to improve safety and reliability needs to be carried out. The work shows a novel solution to create non-contact multi-modal biological recognition because of its high security and fast matching performance, which can be used in electronic currency trading, blockchain and other scenarios. The three-dimensional image acquisition method is designed. Subsequently, multi-layer image quality evaluation, maximum ROI extraction and two image enhancement methods are proposed. Next, the algorithm based on three-dimensional CNN is designed to solve the optimization problem. Finally, “single-line parallel matching mode” is constructed. In the experiments, the proposed method shows that the leading performance on precision (99.92<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>) and processing speed (within 3ms) in comparison with other processing methods of single biometric recognition. Experimental results show that this new design is effective and feasible.