Local directional pattern based signature verification using weighted fractional distance classification

Abstract

Although handwritten signature verification has been extensively researched, it has not achieved optimum accuracy rate yet. Therefore, efficient and accurate signature verification techniques are required since signatures are still widely used as a means of personal verification. This paper presents an alternative efficient classification technique to supervised learning classification techniques. The signature features are extracted using the Local Directional Pattern algorithm, and classified using a combination of multiple distance-based techniques: weighted Euclidean distance, fractional distance and weighted fractional distance. This combination of multiple distance-based classification techniques achieved accuracy rate of 87.8%, which is comparable to a similar system that used Support Vector Machines, a supervised learning technique. Therefore, competitive levels of accuracy can be obtained using distance-based classification.