Signature Verification using Convolutional Neural Network and Autoencoder

Signature has been one of the widely used verification biometrics out there. Handwritten signatures are in cheques, forms, letters, applications, minutes, etc. The every individual is unique nature, that why it essential a person’s handwritten signature be uniquely identified. Verification method for authenticating any during absence. Human prone to inaccuracy and sometimes indecisiveness. This paper presents an investigation using Convolutional Neural Network (CNN) Writer-Dependent models verification. Random distortions were generated genuine images autoencoder get forged signatures, which passed classifier training. details all pre-processing steps carried on image shows various test results changing number training sets images. average accuracy Persian dataset 83% when system was trained with 22 There decrease 9.4% model 9