Font Size: a A A

Research Of Dual-Mode Recognition Algorithm Based On Fingerprint And Finger Vein

Posted on:2011-08-25Degree:MasterType:Thesis
Country:ChinaCandidate:X F LiFull Text:PDF
GTID:2178330332960605Subject:Pattern Recognition and Intelligent Systems
Abstract/Summary:PDF Full Text Request
Biometric identification technology relies on human biological characteristics and it is gradually replacing traditional identity methods and becoming the most effective authentication techniques in our daily lives. The advantages lie in its difficulty to be lost or forgotten and difficulty to be forged and counterfeited. Currently, biometric identification systems research mainly focused on single-mode identification system and much outstanding achievements have been registered in this field. However, there are a lot of problems in the application of identification system and these problems could not be solved only from the single-mode system. The multi-modal system is a new identification system with the advantages of improving the system's accuracy and anti-noise, which combines multiple features. Multi-modal recognition can effectively compensate for the shortcoming of a single identification method and has become more convenient and effective in the authentication.We implemented the multi-modal system which is based on fingerprint and finger vein in this paper. The system consists of the following three parts: fingerprint identification module, finger vein module and multi-modal fusion module.In the fingerprint identification module, we use first a series of pre-processing operation such as gray-level normalization, segmentation, enhancement, binarization and thinning to make the original image into high quality thinned fingerprint image so that we can extract feature easily. Then we use 8-neighborhood method to extract feature from the thinned image. At last, we proposed a feature point matching algorithm and analyze the results.In the finger vein module, we propose a finger vein pattern extraction method using oriented filtering technology. This method extends traditional image segmentation methods, by extracting vein object from the oriented filter enhanced image. Experimental results show that the addition of oriented filter operation, extracts smooth and continuous vein features not only from high quality vein images but also handles noisy low quality images and does not suffer from the over-segmentation problem. We used a recognition method based on MHD matching distance. This algorithm has poor identification performance on finger vein images subject to major translation or rotation. Therefore, we used a relative distance finger vein matching algorithm. The recognition results of this method are much improved as it effectively overcomes the effect of translation and rotation.In multi-modal fusion module, the key technology of multi-modal system was introduced at first. Then a new fingerprint image quality evaluation and finger vein image quality evaluation method was proposed. With this evaluation method, we finish the decision-level fusion operation. Experimental results show that multi-modal system performance is better than a single fingerprint recognition system or a single finger vein recognition system. Furthermore, we proposed another decision-level fusion method based on secondary decision making technology. Our results show that this algorithm is also better than single-mode based recognition systems. The algorithm introduces the concept of feature fusion on the third classifier and laid a good foundation for multi-modal systems based on feature fusion.
Keywords/Search Tags:Fingerprint Recognition, Finger Vein Recognition, Multibiometric Systems, Image Segmentation, Image Enhancement, Feature Extract
PDF Full Text Request
Related items