Font Size: a A A

Research On Fast Index Technology Of Vein Feature

Posted on:2019-01-28Degree:MasterType:Thesis
Country:ChinaCandidate:M ChenFull Text:PDF
GTID:2348330545958532Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
With the rapid development of communications and Internet,people's requirements for information security are constantly increasing.Traditional identity authentication technology has been unable to meet the needs of people due to its own shortcomings.Biometrics has gradually become a hot spot in the field of identity authentication,which has aroused the attention of many domestic and foreign researchers.Some technical achievements have been widely used in attendance,mobile phone unlocking,Internet finance and other fields.As an emerging biometric identity authentication technology,vein feature recognition technology has a wide range of application prospects.Compared with common fingerprint and face recognition technologies,the vein signature recognition technology is not easy to forge,difficult to copy,unique,long-term stability and so on.Although the vein feature recognition technology has been applied to products,the extensive promotion of this technology requires continuous research and exploration.In this paper,the research on identification technology of vein features mainly focuses on vein image acquisition and preprocessing,feature extraction and recognition algorithms,and feature fast retrieval.The research content and innovative work of the thesis include the following points:First,edge detection is performed on the original finger vein image,and the region of interest is extracted through the knuckle feature.The two-dimensional linear interpolation method is used to normalize the size,and the histogram is equalized to obtain a standard vein image.With median filters and band-reject filters,noise is removed in the spatial and frequency domains,respectively.Second,a new vein feature extraction and recognition operator is proposed.A variety of existing LBP operators are analyzed.The original LBP,normalized LBP,rotationally invariant normalized LBP,and CS-LBP all emphasize that the micro features ignore the local structural features,while the MB-LBP emphasizes the local macro characteristic.Therefore,there are obvious deficiencies in the above types of operators.In order to obtain a finger vein recognition operator with strong robustness and high recognition rate,this paper fuses the local macro features and micro features,and proposes an improved MB-CS-LBP operator(referred to as IMC-LBP for short),whose recognition rate reached more than 98%.Third,when retrieving the vein feature vector,the approximate nearest neighbor search strategy is used to apply the E2LSH and the random KD tree to the vein feature vector retrieval.In this way,the retrieval time is greatly shortened without substantially reducing the recognition rate.
Keywords/Search Tags:Biometrics, Vein Recognition, ICM-LBP, E~2LSH, Random KD Tree
PDF Full Text Request
Related items