Combined Global And Grid Features Based Handwritten Signature Recognition And Verification Using Support Vector Machine

Abstract– The paper presents a Hand Written Signature Recognition and Verification System based on haar wavelet transform and Support Vector Machines (SVM). Signature recognition and verification systems are existing research area in which a number of research works are there and ongoing. This paper presents a system which recognizes and verifies signatures within less time, good accuracy in a less complicated way. For feature Extraction global features and grid features are considered mainly based on the Discrete Wavelet Transform (DWT) using the haar wavelets. Recognition of signatures done using Support Vector Machines (SVM) and Artificial Neural Network (ANN) and compared with the output accuracy. The inputs are images and outputs are the name of the signer if the test input is the genuine signature otherwise shows as forged signature. The features are extracted after pre-processing of the images and Principal Component Analysis is used for global wavelet feature reduction. Support Vector Machine training for combined grid and global features gives higher accuracy than other comparable methods here. Keywords-Handwritten signature; Signature Recognition; Signature Verification; Support Vector Machine; Discrete Wavelet Transform