Palmprint identification system using shape matching and k-nearest neighbor algorithm

Palmprint identification is one of biometric methods to ascertain the identity of person. Although palmprint is not the most popular characteristic that we currently use, it is a powerful alternative for identifying in the future technology due to its uniqueness, stableness, convenience and ease of use. In recent surveys, there are many features on palm that can be used for automated recognition such as principle lines, wrinkles, ridges and datum points. Principle line which is the main line is proposed in this thesis. The system is divided into two main subsystems. The first part concerns with palmprint feature extraction using a cascade of consecutive filters to obtain the principle lines. To achieve the proposed system, recognition is provided as the second part to classify principle line image. Firstly, shape histogram is constructed from the extracted image by projections along vertical and horizontal axes. Then, the histogram of query image and the image in database are compared by using cosine similarity measure. Finally, K-Nearest Neighbor is employed to identify a person. The experimental results demonstrate the efficient system yielding high recognition accuracy (98.53%).