Recognition of Linearly Oriented Hand Written Characters using k-Nearest Neighbor

Abstract

Optical Character Recognition System offers the human machine interaction which is commonly used for several important applications. A lot of study has been conducted to accomplish the work on character recognition in different languages. This project represents a technique for recognizing the characters from an image with noise using Optical Character Recognition (OCR). The important steps of this method are preprocessing the text including converting the text image to black / white and removing the noise from the text image, segmenting the text image into each character, extracting features using K-nearest neighbor (KNN) and classifying. The System is implemented using Anaconda software application program. Noise from all images of the text is excluded. The quality of the input document is very important to achieve high accuracy rate.