Offline Signature Recognition using Pretrained Convolution Neural Network Model
Kamlesh Kumari1, Sanjeev Rana2
1Kamlesh Kumari ,Research Scholar, Department of Computer Science Engineering , M. M (D.U), Mullana, Ambala.
2Dr. Sanjeev Rana, ,Professor, Department of Computer Science Engineering , M. M (D.U), Mullana, Ambala.
Manuscript received on September 23, 2019. | Revised Manuscript received on October 15, 2019. | Manuscript published on October 30, 2019. | PP: 5497-5505 | Volume-9 Issue-1, October 2019 | Retrieval Number: A2016109119/2019©BEIESP | DOI: 10.35940/ijeat.A2016.109119
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Offline Signature recognition plays an important role in Forensic issues. In this paper, we explore Signature Identification and Verification using features extracted from pretrained Convolution Neural Network model (Alex Net). All the experiments are performed on signatures from three dataset (SigComp2011) (Dutch, Chinese), SigWiComp2013 (Japanese) and SigWIcomp2015 (Italian). The result shows that features extracted from pretrained Deep Convolution neural network and SVM as classifier show better results than that of Decision Tree. The accuracy of more than 96% for Japanese, Italian, Dutch and Chinese Signatures is obtained with Deep Convolution neural network and SVM as classifier.
Keywords: Decision Tree (DT), Deep Convolution neural network (DCNN), Support Vector Machine (SVM), Boosted Tree (BT), Writer Dependent (WD), Writer Independent (WI), K-nearest neighbor (KNN).