A Study on Preprocessing and Feature Extraction in offline Handwritten Signatures

Abstract

In offline handwritten signature verification process preprocessing of the signature is the very fast and most essential part In some cases the raw signature can include extra pixel known as noises or may not be in proper form where preprocessing is mandatory If a signature is preprocessed correctly it leads to a better result for both signature matching and forgery detection Pre-processing includes binarization noise removal thinning orientation etc Many experiments and techniques have already been proposed for implementing these processes and some of them have shown exclusive and spectacular results Regarding to this situation we have studied several preprocessing steps signature features feature detectors and also implemented some of them using MATLAB software We have studied several image processing algorithms and proposed an algorithm to correct the alignment of the input signature which can be used at the preprocessing stage to achieve better results in the signature detection process We have tried to find a baseline of the handwritten signature and align it with respect to the baseline