Font Size: a A A

The Research On 3d Palmprint Feature Extraction And Matching Methods

Posted on:2010-01-15Degree:MasterType:Thesis
Country:ChinaCandidate:C Y LiFull Text:PDF
GTID:2198360332457868Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
Traditionally, palmprint verification is based on 2-dimension image information which could be called 2D palmprint verification. However, it has confronted with some bottlenecks, such as: higher anti-counterfiting, more robust to noise and so on. To improve the accuracy and robustness, the 3D palmprint verification is proposed.First, in the developed system, a cost-effective grey LCD projector with LED light source is employed, and some shift light patterns are projected onto the palm. With the phase shifting and unwrapping techniques, we can retrieve the depth information of the object surface by projecting a series of different phase stripes on it. The range data of the palm surface can then be obtained. Moreover, 3D palmprints have more rich information than 2D palmprint data. Meanwhile, the system can get 3D palmprint data and 2D palmprint data, it has a great contribution to the fusion algorithm of 2D and 3D.Next, with the help of Region of Interest (ROI) extraction algorithm on 2D palmprint, the corresponding ROI on the 3D palmprint is available. According to the surface curvature characteristics of classical differential geometry, we obtain the curvature information of 3D palmprint images. We mapped the original curvature images into grey level images with integer pixel—Mean Curvature Images (MCI) and Gauss Curvature Images (GCI).Finally, we select MCI to take a step forward operation by comparing the MCI and GCI. We mainly apply dimension reduction methods which can reduce the time complexity and space complexity effectively and also obtain better recognition effect. In this paper, Principle Component Analysis(PCA), Locality Preserving Projections(LPP) and Independent Components Analysis(ICA) methods are applied. More importantly, a new method on palmprint—Regularize Linear Discriminant Analysis(RLDA) is proposed, which has obtained the best result.The experiment results strongly supported that the method presented in this paper has a higher accuracy and implementation of procedures compared with traditional LDA, PCA, ICA and LPP etc. Meanwhile, the method eliminates the existing problems of traditional LDA—small sample size problem and the problem that optimization criterion function is not directly related to recognition rate.
Keywords/Search Tags:Palmprint Identification, 3D Palmprint Identification, 3D Curvature Feature, RLDA
PDF Full Text Request
Related items