Recognition of Off-Line Hand-Written Alphabets Using Knowledge-Based Computational Intelligence

Abstract

The handwritten character recognition is considered an active recognition problem under the field of image processing and pattern recognition. Both area of research has gained so much popularity in few last years due to technology advancements and necessity of time. The handwritten character recognition also becomes popular due to globalization and technology improvements. In previous time, the information was stored in written form only. To utilize that information, one needs such an advanced method which can do this task efficiently and without reliance on human. This work provides such a simplified and accurate method for the recognition of handwritten characters. A system is proposed to recognize characters using convolutional neural network and is evaluated on a benchmark dataset named as EMNIST to show the performance of the proposed technique. The proposed system gained the accuracy of 97.62% which is better than various existing methods.

Keywords

• convolutional neural network
• handwritten character recognition
• emnist dataset