Off-line signature verification with PSO-NN algorithm

Analysis of signature is a widely used and developed area of research for personal verification. A typical signature verification system generally consists of four components: data acquisition, pre-processing, feature extraction and verification. This paper presents a novel technique for off-line signature verification (SV). The technique is based on a neural network (NN) approach trained with particle swarm optimization (PSO) algorithm. To test the performance of the proposed PSO-NN algorithm three types of forgeries; random, unskilled and skilled are examined and the experimental results are illustrated.