Font Size: a A A

Kernel Function-based On Handmetric Recognition

Posted on:2010-11-17Degree:MasterType:Thesis
Country:ChinaCandidate:Y LongFull Text:PDF
GTID:2178360278452344Subject:Human-computer interaction projects
Abstract/Summary:PDF Full Text Request
Recently, the method of feature extraction based on kernel function method has been widely applied in the community of biometric. The kernel function plays an important role on the performance of the system. Hence, people pay much more attention on recognition of the kernel function gradually. This paper analyses the kernel function, and puts forward conditional positive definite function according to its own characteristics of hand image. Finally, this paper presents the kernel matrix fusion methods under relational measure framework by using multiple modality algorithm and feature-level fusion.This paper mainly presents three parts:1. This paper puts forward rough classification algorithm based on box fractal dimension of the dynamic palmprint, and compare the time of several rough classification algorithm in the circumstance of ensuring high recognition rate, by the analysesing experimental results. A conclusion that the rough classification algorithm based on box fractal dimension shorten the search time, and reduces the computational complexity, is given.2. This paper proposes conditional positive definite kernel function (CPD) according to the existing problems of different kernel functions, and palmprint and knuckleprint's characteristics. The experimental results show that the handmetric recognition based on the condition positive kernel function has the characteristics of anti-translation and emphasizing local information, and improves the recognition rate. 3.The multi-biometric feature algorithm has been proposed by fusion KPCA according to feature-level fusion under the relational measure framework, including the feature-level kernel matrix fusion methods by palmprint and knuckleprint, and a variety of feature- level fusion algorithm based on kernel function fusion. According to the experimental results, the fusion can achieve higher recognition rate and lower error rates of authentication.
Keywords/Search Tags:KPCA, feature-level fusion, kernel function, CPD, fractal dimension
PDF Full Text Request
Related items