Research Of Dorsal Hand Vein Recognition Algorithm Using Optimized Texture Features And Its Implementation On RaspberryPi System

In this paper,we present an algorithm of dorsal hand vein recognition based on optimized texture features,and implement the algorithm on RaspberryPi system.With the rapid development of scientific information technology,the traditional security authentication method has been unable to meet the needs of the current society for personal identification,and the biometric identification technology has emerged.The recognition of the dorsal vein of the hand is widely concerned because of its non-contact and high safety level.Therefore,it is very important to study it in depth.It is also urgent to explore new hardware implementation of identification system,in order to cope with the demand of miniaturization,ease of use and extensibility in the market.The Raspberry Pi is a tiny and affordable computer that you can use to learn programming through practical projects.The design of the vein recognition system using RaspberryPi can effectively reduce the cost and facilitate the product.The main research are the design of the dorsal hand vein recognition algorithm and the recognition system based on RaspberryPi,which includes the following aspects:(1)This paper builds a dorsal hand vein image acquisition device,and collects and establishes the dorsal hand vein image database for the study of the early identification system algorithm.The database now includes more than 300 volunteers' hand images.Each of them has 5 images in each left and right hand.The size of each image is 320*240 gray scale.(2)This paper presents a method based on the Voronoi of the dorsal outline of the hand to extract ROI of the dorsal hand vein.The result of ROI images has shown a better robustness.(3)This paper proposes a feature extraction algorithm based on optimized texture features for the dorsal hand vein feature extraction.The traditional Gabor filter can effectively extract the texture features of the dorsal vein,but its feature dimension is too large and the time of data processing is too long.Aiming at this problem,this paper proposes a set of methods for optimization: First,we use three-layer haar wavelet decomposition to reduce the dimension of image.Then,we use different scales and directions of gabor kernel function to extract texture features of low-frequency sub-band images.Finally,we use PCA to further reduce the features.The nearest neighbor classifier method based on European distance is used to match the final feature,and the feasibility of the algorithm is verified by experiments.(4)This paper designs a dorsal hand vein recognition system based on RaspberryPi.The hardware platform,software program structure and debugging method of the system are introduced in detail,and the algorithm is transplanted to RaspberryPi system.The recognition performance of the system is verified by experiments.In summary,this paper studies the technology of dorsal vein recognition.Its designs the acquisition device,puts forward a set of algorithms from the image preprocessing to feature extracting,and successfully applies to the vein recognition system of dorsal hand based on RaspberryPi.Better results is achieved by applying these methods.