Research On Identification Method Based On Palmprint And Hand Shape Biometric

In today's society personal identity authentication demand is increasing. Biometric technology can overcome many shortcomings of traditional identity authentication technology, so it's already became a important means for the of security information and network, these techniques used to prove their identity is just the important biological information carried by individual, and this kind of information is not easily lost, it is difficult to counterfeit and sham, which is a convenient and safe method to identity authentication.Single biometric technology has its disadvantages and limitations it is hard to overcome this disadvantages and limitations especially in the actual application conditions, which need a variety of biological characteristics to be fused. For hand shape identification, the characteristic vector constitute a simple but also have scalability, however it will loss some information of palm by a single geometric vector structure. For palmprint identification, palmprint image has very good degree of distinction by making full use of the information of palm texture, so that it can get a high recognition rate, but the time consumption is larger and the speed of recognition is affected.This paper proposed two kinds of hand and palm prints fusion method taking advantage of fast hand identification matching recognition and high palm print recognition rate, under hands spontaneously open and non-contact collection condition. Overcame low single hand Recognition rate and slow palm print identification matching problem. Taking the relative length of fingers as a feature vector for hands, and using 2D-Gabor to filter the region of interest (ROI), extracting palm prints ridge direction information as feature. One method filtered out unqualified samples using hand recognition in the matching layer, then took the candidate sample as a palm print recognition object, and obtain the final recognition results. Experiment proved a 98.57% recognition effect. This method improved the system recognition rate effectively and reduced matching time. The other method combined k nearest neighbor classifiers and supporting vector machine to classify the hand shape recognition in the decision-making layer, then used palm prints identification to certify hand shape classification results. The method improved the system practical application security and stability effectively.For further research the performance of hand and palm prints fusion method under the spontaneously open and non-contact collection condition, this paper designed an non-contact online simulation identification system using matlab Image Acquisition Toolbox functions and GUI, achieved online image acquisition, preprocessing, feature extraction and matching, etc.