Research On Directional Local Descriptor Based Palmprint And Palm Vein Recognition

In recent years,biometrics has been recognized as a safe and effective authentication solution,which has become a key technology and industry in developed countries.Among all the biological characteristics,the most widely used ones are fingerprints,faces and irises.However,there are some limitations of these biological features.In the recent ten years,palmprint and palm vein recognition have drawn much attention from academia and industry.Compared with other biological features,palmprint and palm vein have many unique advantages.These advantages will make them widely used biometrics.However,how to extract effective features from palmprint and palm vein has become an issue that both academia and industry are exploring.Among all the feature extraction methods,the method based on image local descriptor can extract more detailed information and be easily implemented.It is also robust to light changes and has no training cost.It has been successfully applied to face recognition.Recently,researchers have introduced local descriptors into palmprint and palm vein recognition and good performance has been achieved.Therefore,aiming at the rich line direction features in palmprint and palm vein,this thesis has proposed discriminant local descriptors based on directional features,which can improve the recognition performance greatly.The results are of great significance to the scientific development and market application.The main works are as follows:1)The WLD-based local descriptors are summarized,their characteristics and performance in mainly used fields are analyzed.2)Generalized WLD(GWLD)in multi-scale and multi-orientation maps is proposed to improve the accuracy of palmprint and palm vein recognition.3)Enhanced HOL(EHOL)in multi-scale and multi-orientation maps is proposed for palmprint and palm vein recognition.4)A palmprint recognition system is designed.