Feature Ensemble based method for verification of Offline Signature images
Pooja Chaturvedi, Anamika Jain · 2022 International Conference on Machine Learning, Big Data, Cloud and Parallel Computing (COM-IT-CON) · 2022
Signature verification is the most researched area in computer vision as it has less intra-class and less inter-class similarity. Our objective is to reduce the intra-class variation and increase the inter-class similarity. This paper has proposed an ensemble based approach where method of ensemble of feature created by using geometrical and Mobile Net features along with the ensemble of classifier has utilized. The proposed method has been tested over the publicly available dataset BHSig260. This paper has achieved 99.4% and 99.3% accuracy over Bengali and Hindi dataset respectively.