Design a New System for Recognition of Devnagri Numeric Characters using Neocognitron

Handwriting is a way of communicating with one another through written media. In today’s era of the digital world where everything is getting digitalized, there is a requirement of a system which can recognize the handwritten text. OCR is used to recognize handwritten data which is in form of images and convert it in into editable and searchable data. Neocognitron is a multilayered neural network which is used to recognize handwritten text. Neocognitron also has an ability to recognize a defective and noisy pattern in correct form. The main objective of this research paper is to design a new system for character and pattern recognition of above said characters in any mode e.g. Normal, Noise and defected patterns of devnagri numeric characters. We also outline the main challenges of using CNNs in patterns in before learning and after learning systems. On the each Devnagri numeric characters (0 1 2 3 4 5 6 7 8 9) using Neocognitron's 3 stages model with Normal, Noise and defected input patterns.