Offline signature verification method using weighted block analysis

Handwritten signatures are socially and legally well accepted as a convenient means of writer verification, document authentication and authorisation. It is a very difficult task to decide automatically whether two signatures, given as scanned images are written by the same person or not. In this paper, an effective signature verification method that depends on the structural property of different parts of the image has been proposed. It uses a new technique for feature extraction by splitting the image into several blocks. The main features which play the key role in this technique are image center and block centers, Euclidean distance between image center to each block center and number of pixels in each block. Each block is assigned some weightage according to its successful appearance in image formation. Performance of the scheme has been measured and presented in terms of FAR (False Acceptance Rate) and FRR (False Rejection Rate). The result shows that the method eliminates random and simple forgeries and also reduces skilled forgery to a great extent.