Offline Writer Identification Using Local Derivative Pattern

Abstract

In this paper, we propose a scheme for identifying the authorship of off-line handwritten documents based on a histogram-based descriptor. The idea of our work is inspired from that of the Local Derivative Patterns (LDP), that has found much success in the application of face recognition. However, to the best of our knowledge, this work is the first of its kind that utilizes them for characterizing the writing style of an author. The efficacy of the algorithm has been tested on the handwritten documents of the CVL database, using two strategies. The performance of writer identification rate on this database indicate that the proposed descriptor is effective for the problem of text independent off-line writer identification.