Handwritten signature recognition with Gabor filters and neural network
Iman Subhi Mohammed · AIP conference proceedings · 2019
Due to the necessary of verification systems and identification of persons from Handwritten signatures as a means of identification. This process has become an important requirement to ensure security and prevent fraud and impersonation. As biometric techniques progressed, it was necessary to use computer vision and image processing to increase the efficiency of this verification process. The aim of this work is to integrate the methods of image processing and intelligent techniques such as neural networks as well as statistical methods to build an efficient biometric system for the process of distinguishing signatures and identity verification. Multi-scale and multi-orientation analysis methods were used as a Gabor filter to distinguish the different signatures and their attribution to their true owners to upgrade this process when examining signature images and identifying owners. This work involves converting images to the frequency space using Gabor converter and then extracting the statistical and engineering characteristics of the images to form the matrix of the input to the back-propagation neural networks BPNN. A set of live images were collected as a database of people affiliated with the University of Mosul. The proposed work showed a 88.57% success rate compared with the results of similar work in the same field. The system was built using MATLAB 7.10.0 + (R2016).