Topological and textural features for off-line signature verification based on artificial immune algorithm

This work presents a new system for off-line handwritten signature verification. Specifically, Artificial Immune Recognition System (AIRS) is employed to achieve the verification task. Also, to provide a robust signature character-ization, two new features are used. The first data feature is the Orthogonal Combination of Local Binary Patterns (OC-LBP), which aims to reduce the size of LBP histogram while keeping the same efficiency. In addition, we propose a topological feature that is based on the image Longest-Run-Features (LRF). The proposed features are evaluated comparatively to the state of the art methods. The results obtained for CEDAR dataset, highlight the efficiency of the proposed system.