Finger vein counterfeit attack detection method based on capsule network

• Main Authors: HUANG RONG, JIANG YIFAN, GUO JIAN, CHEN RUOXING, TAI YIWEN, ZHU ZHA, WANG JUAN, HAN CHONG, ZHANG QIN, MU HENGYU
• Format: Patent
• Language: Chinese; English
• Published: 09.06.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

According to the finger vein counterfeit attack detection method based on the capsule network, the capsule network is introduced, the capsule network is not only suitable for a finger vein data set of a small sample, but also neurons are replaced by capsules represented by vectors on the basis of the CNN, so that spatial information such as relative positions and angles can be better processed, and the detection accuracy is improved. The adaptability of the network to finger offset and rotation scenes is enhanced; a Bayesian routing algorithm is provided, the differential entropy of the capsule is added as a consideration for calculating an activation value, a feature capsule with a high activation probability and a large concentration degree is selected during final classification, and the accuracy of true and false vein classification can be improved; by simulating the uncertainty of the capsule parameters, the training error can be reduced, and the recognition accuracy can be improved. 基于胶囊网络的指静脉仿冒攻击检测方法，引