Research On Biological Image Verification For Mobile Devices

The 21 st century is an era of rapid development of science and technology and every walks of life need a efficient security authentication technology urgently, such as online banking, online shopping, medical, forensic and other industries. Compared with the traditional identity verification/recognition technology, biometric verification/identification technology is more secure and convenient by using the biological characteristic of the human body to identify someone's identity, more secure and convenient. Along with the rapid development of mobile Internet technology and the constant updating and popularization of mobile equipment, more and more researchers put forward combining biometrics and mobile devices. At present, the biological image recognition research for mobile devices also has many problems. Therefore, this paper proposes the identification for mobile equipment, mainly for the palm image recognition and multiply biological image recognition.When using the camera on a mobile device for image acquisition, it is easy to generate the image exposure intensity too strong or too weak, and the image angle inconsistent or other problems. After deep research of biometric identification algorithm for mobile devices, this paper will study the following three aspects. The first aspect is the image preprocessing stage. In this stage, we try to eliminate the effects of light and angle problems by the histogram equalization method and a variety of filters. The second stage is the feature extraction, we compare the experimental results of the PCA, LDA, SIFT, LBP feature extraction algorithm, and then we can see SIFT feature extraction algorithm has the highest recognition rate which can reach 90.65 %. We conduct deep study of SIFT algorithm and put forward the improvement scheme for removing the matching error points, the experimental results show that the recognition rate of the improved SIFT algorithm can reach 94.83%. The third problem is multiple biological image recognition method for mobile devices. In this paper, we proposed a multiple biological image features fusion method based on SIFT algorithm for fusing the matching result of the each feature image and the experimental results showed that its accuracy reaches as high as 99.39%.There is no database comform to the demands in this paper on the internet, we manually collected different database respectively specific to different problems of this subject. The recognition rate of improvement scheme of SIFI is more than standard SIFI algorithm of 4% in the stage of feature extraction in the experiment. In terms of time efficiency, the time-consuming of the improved scheme of SIFI is close to the time consuming of the original SIFT algorithm; Although the time efficiency of the multiple biological image features fusion method that we proposed decreased, the final recognition rate of more than 15% comparing with other methods. The experiment proof that the improvement scheme we proposed in this paper has important value to the following related biological recognition study for mobile devices. At the same time, after the algorithm research achieving certain results, the paper also implements the palm verfication system on mobile devices, the system has high practical value.