Improved Hand Vein Threshold Segmentation Method And Live Detection

Today,with the rapid development of science and technology,biometric identification technology has become the mainstream of identity authentication because of its high confidentiality and difficulty in stealing.The hand vein recognition technology,one of the biometric technologies,is an emerging biometric identification method.Compared with other biometric objects such as faces and fingerprints,the hand vein technology has the advantages of being able to collect the user's identity without contacting the acquisition device,high confidentiality,and large differences in the texture of the back vein of the individual hand.Therefore,this thesis conducts biometric research on the dorsal vein of the human hand.The specific research contents are as follows:(1)According to the theory of hand vein imaging,a set of hand vein collection device was built.The device helps the experimenter correct the posture of the back of the hand by adding sensors at the collection of the lying rod.The device was used to capture images of non-living hand veins,including non-living hand vein images with rubber gloves and non-living hand vein images with strokes.(2)A set of hand vein image pretreatment process is designed to remove the noise of the hand dorsal vein image by image segmentation and denoising.The effective area of the image of the dorsal vein of the hand is extracted by means of a Veno diagram.(3)The problem of image deception of non-living hand vein texture in the presence of hand vein recognition system.Design experiments for the classification of living hand vein images and non-living hand vein images.First,the collected images of the living hand vein and the non-living hand vein image are preprocessed.The vein image of the effective area is extracted by HOG,and then classified by SVM.(4)Aiming at the problem that the image edge region of the traditional Niblack threshold segmentation method can not be effectively segmented,the Niblack threshold segmentation method is improved.First,the image of the dorsal vein of the hand is divided into an edge region and a central region,and the mean and variance of the entire image are calculated to obtain a reference threshold.Then calculate the static reference threshold of the edge region of the dorsal vein image of the hand.By calculating the gray color of the back vein texture and the image in the image of the edge region,calculate the maximum variance of the two to obtain the static reference threshold of the edge region.The two types of reference thresholds are reasonably combined,the edge region corresponds to a corresponding static reference threshold,and the central region is the mean of the edge region threshold.Finally,the opponent's dorsal vein image is thresholded.(5)The algorithm for identifying the dorsal vein of the hand is implemented on the DSP.The software design and hardware design of the system were carried out separately.Among them,the hardware design mainly introduces the system selection,structure design and module design.The software design mainly introduces the software development environment,DSP/BIOS operating system and gives the design of the program framework.A program flow chart was designed for the experiment of distinguishing the image of the dorsal hand vein and the image of the non-living hand vein.The program flow chart was designed for the matching recognition experiment of the dorsal hand vein.