全视角三维指静脉识别关键问题研究

指静脉识别特有的活体检测能力使其成为了高安全场合一种有效的身份识别方式，但目前的单视角平面成像模式使得指静脉识别存在一些自身难以克服的问题，如手指轴向旋转引起的指静脉纹理偏移和形变，单视角成像区域受限造成手指中部分静脉纹理的缺失，以及单一平面成像模式造成三维几何信息的损失，这些影响了指静脉识别的准确率及其仿冒攻击检测能力。为了从根本上解决这些问题，本项目提出了全视角三维指静脉识别的新模式，并从二维图像识别，三维点云匹配，以及多视角学习等三个维度对三维指静脉识别展开全路径的创新性研究，重点探索全视角三维指静脉的重建思路和抗旋转机理，研究充分挖掘和全面利用三维指静脉特征信息的有效策略，发掘提高指静脉识别准确率及仿冒攻击检测能力的有效途径。本项目的顺利开展将对三维指静脉识别核心技术的发展起到积极的促进作用，丰富了现有的生物特征识别模态，并为生物特征识别研究提供一些有益的新思路。

With unique capabilities of liveness detection, finger vein recognition becomes an effective form of identification in high-security occasions. However, under the current single view based plane imaging mode, there exist some insurmountable problems in finger vein recognition, for instance the finger vein texture displacement and deformation resulted from rotation around the finger, the finger vein texture missing caused by limited imaging area of single-view, and the 3D geometric information loss brought by plane imaging mode, which affects the accuracy of finger vein recognition and the ability of representation attack detection. In order to fundamentally solve these problems, this project proposes a novel mode of full-view 3D finger vein recognition. An innovative study on 3D vein recognition is carried out from three dimensions including 2D image recognition, 3D point cloud matching, and multi-view learning. This project aims at exploring the reconstruction strategies and the anti-rotation property of full-view 3D finger vein, and effective strategies of fully utilizing feature information of finger vein so as to improve the performance on finger vein recognition and its representation attack detection ability. The successful implementation of this project will play a significantly positive role in promoting the development of 3D finger vein recognition technology, enriching the existing biometric modality, and providing some beneficial ideas for the research on biometrics.