| As society progresses,traditional identification is unable to meet people’s needs due to the need to wear foreign objects or remember passwords at all times,while biometric identification technology completes the authentication process through only some of the human body’s characteristics,and the advantage of convenience and security makes it a popular research direction.The epidemic era has further highlighted the security of noncontact identification methods in biometric technology,so palm vein recognition technology,which has that advantage,has become a focus of research.Palm vein recognition technology has the advantage of not being directly observable and not be easily replicated,as it uses the vein texture as proof of identity.However,numerous studies have shown that this recognition technology currently suffers from poor quality of the original palm vein image and low security in the recognition process.This paper focuses on optimizing the palm vein identification technique to alleviate these difficulties.To address the issue of the quality of the original image affecting the recognition performance,this paper proposes a palm vein reconstruction recognition method based on compressed sensing.The method first performs compression-aware image reconstruction on the original palm vein image,followed by ROI interception,CLAHE enhancement and SURF feature extraction processes,and finally matching and comparing the output results.The original low-quality image is reconstructed to make the vein texture clearer,thus allowing more information to be obtained about the palm vein image features.Experimental results show that the method enhances the clarity of the original palm vein image and reduces the error rate in performance parameters.To address the security issues in the recognition process,an encrypted palm vein recognition method based on ResNet FFT is proposed in this paper.The method takes the palm vein image features extracted by the ResNet network,encrypts them with Fast Fourier Transform and Logistic mapping and then performs comparison verification.The scheme can effectively prevent the occurrence of information being stolen and then restored to the real vein features.Experiments have been conducted to test the recognition performance parameters such as error rate and resistance to attack,and the results show that the recognition method has good recognition performance under normal conditions and is still accurate when the attack parameters are set within a certain range,so the method is also robust in terms of resistance to attack.In addition,this paper deploys palm vein recognition technology in smart hardware,using a Raspberry Pi as the processor,an external LCD touch screen as the user interface,a near-infrared camera as the input,and a near-infrared fill light to enhance the image texture,enabling the registration and verification function of palm vein images.However,there is still some way to go before the device can be used commercially,and further iterations will be required to achieve higher recognition rates and security. |
