الفهرس 
Biometric TechniquesOnly 14 pages are availabe for public view
 |  | from | 166 |  |  |
| | | from | 166 | | |
Abstract
Because of the uniqueness and complication of hand vein patterns the hand vein authentication gives a great level of recognition accuracy. The hand vein patterns are inside the body so that it can be a challenging technique to forge. Moreover, the system is contactless and hygienic for operate in society regions. The contactless hand vein recognition technology comprises of image sensor and software technology. The user’s hand is hold under an infrared ray scanner to capture an image of hand. The illumination of the infrared beam is limited relying on the light round the sensor, and the sensor is capable of take the hand image in spite of the location and movement of the hand. Afterward, the software compares the decoded vein image with the enrolled images, though determining the location and alignment of the hand by a pattern matching technique.
This research reports a novel multimodal biometric system employing intelligent technique to authenticate human by fusion of dorsal hand, palm and finger veins pattern alternately. By explains an image analysis technique to detect the region of interest (ROI) from hand vein image. Once detecting ROI, construct a series of preprocessing steps for eliminate the transformation and rotation of hand vein images presented in the data gathering procedure and decrease several data quantity devoid of missing any effective information this stage done by used canny edge detector, dilation filter and erosion filter. Next construct a series of preprocessing steps to improve the contrast among hand vein patterns and background using median filter, 2D Wiener filter, Applied Contrast Limited Adaptive Histogram Equalization (CLAHE) filter and Homomorphic filter. Once hand vein pattern extracted it managed to obtain the features to use them in matching stage. This study use Principal Component analysis (PCA) algorithm to extract features. In matching step K-nearest neighbor (KNN) classifier with the Euclidian distance used as a similarity measure by one-to-one match method. Bosphorus Hand Vein Database, CASIA Multi-Spectral Palmprint Image Database V1.0 (CASIA database) and the Shandong University Machine Learning and Applications - Homologous Multi-modal Traits (SDUMLA-HMT) databases used in experiments in this study. Finally, fusion methodology adopted at the decision level which is a post-classification method, and it follows the AND rule. It compared the accuracy of the three unimodal (dorsal hand, palm and finger veins) and the four multimodal (fusion of finger and dorsal hand, fusion of palm and dorsal hand, fusion of palm and finger and fusion of dorsal hand, palm and finger veins) and the results showed that the fusion of dorsal hand, palm and finger veins has the best accuracy with correct recognition rate (CRR) 99.21%.