The Study Of Key Techniques Of Palmprint Recognition

In information and network society,there are many occasions in which the personal authentication is required.There is no doubt that biometrics is one of the most important and effective solutions for this task.Generally speaking,biometrics is a field of technology that uses automated methods for identifying or verifying a person based on physiological or behavioral traits such as face,fingerprints,iris and palmprint etc.Due to its importance and promising market prospect,biometrics has been receiving wide attention from researchers.And,the research related to this field is becoming more and more active and hot.Palmprint recognition,developed in recent years,is a new biometrics technology. Compared with fingerprint or iris based personal biometrics system,palmprint based biometrics system has several unique advantages such as stable line features, low-resolution imaging,and easy self positioning etc.More importantly,it can obtain high accurate recognition rate with fast processing speed.Thus,it is very necessary and impending for ones to make a deep study on palmprint recognition.However, palmprint recognition is still staying at the developing stage,and the investigation of its theory and application should be enhanced and improved further.In this thesis, some key techniques of palmprint recognition have been investigated deeply. Particularly,several novel and effective algorithms have been proposed such as recognition methods based on principal lines,orientation code,and appearance etc.The main work in this thesis can be summarized as follows:(1) Line is the basic feature of palmprint.We proposed a novel palmprint verification approach based on principal lines.In feature extraction stage,the modified finite radon transform was proposed,which can extract principal lines effectively and efficiently even in the case that the palmprint images contain many long and strong wrinkles.In matching stage,a novel matching algorithrn was devised to calculate the similarity between principal lines,which has shown good robustness for slight rotations and translations of palmprints.The experimental results for the verification showed that the discriminability of principal lines is also strong.(2) In palmprint recognition field,orientation codes were deemed to be the most promising methods,since the orientation feature contains more discriminative power than other features,and is more robust for the change of illumination.In this thesis, we proposed a novel robust line orientation code for palmprint verification,whose performance is improved by using three strategies.Firstly,the proposed modified finite Radon transform can extract the orientation feature of palmprint more accurately and can solve the problem of sub-sampling better.Secondly,we constructed an enlarged training set to solve the problem of large rotations caused by imperfect preprocessing.Finally,the proposed matching algorithm based on pixel to area comparison had better fault tolerant ability.The experimental results showed that the proposed approach has high recognition rate and fast feature extraction speed.(3) For fast and robust palmprint identification,we developed a new approach, which contains two interesting components.Firstly,the directional representation was proposed for appearance based applications,which is robust to drastic illumination changes and preserves important discriminative information for classification.And then virtual samples were generated to enlarge the training set to compensate matching errors caused by large rotations and translations.Based on these two strategies,the recognition performance of representative appearance based approaches can be improved significantly.Secondly,in order to improve the robustness of palmprint identification,we proposed a fusion method named as Fusion_R combining the proposed appearance based approaches and orientation based approaches,which can obtain very low EER More importantly,we proposed a fast palmprint retrieval scheme named as Fusion_FR,which can obtain very high accuracy recognition rate. Particularly,its matching speed is about 4 times faster than that of Competitive Code, which is one of the approaches with fastest matching speed until to now.(4) A fast palmprint identification approach using a hierarchical retrieval scheme was developed.In the first layer,a fast retrieval method based on the matching of key points of principal lines was proposed for coarse recognition.In this stage,a small database,which contains a lot of training samples with similar structures to test sample,was constructed for subsequent free recognition.In the second layer,Tensor Marginal Fisher Analysis method combining directional representation and enlarged training set was adopted for performing the second retrieval.In the last layer,three approaches i.e.Tensor Marginal Fisher Analysis,Competitive Code and Phase-Only Correlation were fused to form a very robust recognition system.It has been validated that the proposed scheme is effective and efficient by the experiments.(5) An infant biometrics system based on footprint recognition has been developed.Firstly,an infant footprint image acquisition device using digital camera was designed.Secondly,a preprocessing method exploiting the contour of footprint was developed to crop the ROI sub-image.Finally,the footprints can be recognized by means of line feature.The primal experimental results showed that the proposed system is feasible.