Application of Artificial Neural NetworkTechniques in Text Recognition from anImage

Abstract

In this paper, a model is proposed that works on the concept of recognizing text from an image with the help of neural network. Firstly, the text is extracted from the input image. This is one of the most important tasks of this system. Then this extracted text goes through various pre-processing steps that include binarization, normalization, converting text on gray scale, point detection, edge detection, angular rotation etc. After the pre-processing the text is segmented into various sub sections for better functioning of the system and to maintain the accuracy. Important features of the segmented text are extracted in the process of feature extraction. These features helps to differentiate the characters or segments from one another. Finally, the text is classified and is fed to the neural network for the training purpose. Therefore, neural network learns in an unsupervised manner. In the testing phase, the neural network based on the trained data gives the result as recognized text. In this way, the system works in 2 phases i.e. training phase and testing phase and attains state-of-the-art performance