الفهرس 
ContentsOnly 14 pages are availabe for public view
 |  | from | 129 |  |  |
| | | from | 129 | | |
Abstract
This thesis proposes a novel palmprint recognition methodology in the state of the art of fusion feature extraction criteria. All feature extractors have been measured and tested according to seven machine learning algorithms. The features fusion have been designed between novel principal lines morphological features and image-based features extraction techniques. Firstly, the Region of Interest (RoI) has been extracted based on the valley points methodology. Additionally, some image enhancement techniques such as morphological operations and edge detection have been implemented in order to formulate the RoI image more appropriate for the Hough transformation. Furthermore, Hough transform has been applied in order to extract the palmprint RoI principal lines. Accordingly, the most important morphological characteristics of those lines; length and slope have been obtained. In addition, the thesis proposed different features extraction techniques including Principal Component Analysis (PCA), transformed domain, and invariant moments. Finally, different machine learning methods after implementing fusion features vectors have been performed in order to achieve the highest effective palmprint recognition system. The accuracy of recognition was evaluated by measuring sensitivity, specificity, precision, dice, Jaccard, correlation coefficients, accuracy, and training time. Seven different supervised methods for machine learning were utilized and implemented. Experimental findings show that, among all the tested machine learning methods, the feed forward neural network with back propagation based on features vector of the fusion between invariant moments and principal lines morphology have achieved 99.99 percent accuracy of correct recognition.