Hindi Optical Character Recognition and Its Applications

Introduction

Hindi HCR is considered as one of the grave situation today in the world. Typed or handwritten Hindi characters are identified by a computer. Though it is not easy to recognize the Hindi hand written text (Garg, N. K. 2015). They are not recognized efficiently or more accurately by the computer or any optical character recognition machine. There have been a lot of researches in the same context and may different algorithms have been proposed for the recognition purpose along with the availability of the software in the market. For recognizing characters, there are a series of procedures and processes are done. A lone process or only machine is developed so far that can do the entire recognition itself.

The setback in this recognition is recognized by many researchers. The simplest technique is the template matching in which the templates of each word are already taken and preserved for checking the input image in terms of error and the difference between the two is thus calculated. There are techniques which works very fine for different fonts but owes a limitation of giving poor output in terms of the performance for the hand written characters. The next approach is the feature extraction which takes the statistical distribution of points and is then analyzed. Furthermore the properties are extracted which are orthogonal. Features are calculated for each and every symbol and then are stored in the database. Though this technique can be used for the handwritten characters but it is subjected to noise and thickness. We can also use geometric approach the feature extraction is quite precise and can be easily understood.

The ANN (artificial neural networks) has been so far found the best alternative for the recognition due to the ease in the design and the can be used easily worldwide. Hindi character recognition is a significant part of the modern world as it makes the jobs of the people easier than it was used to be. The complex problems can be solved with much accuracy and in lesser time. Even if there is a richness in the ongoing research still there is a certain need in the area of the research. With the increase in the use of the office automation it is an urgent need of the hour to provide practical and effective solutions. However there are many different factors such as topological, arithmetical, structural does not help in the identification process.. The main focus is on the handwritten or hand printed character recognition due to the differences in character shape and size.

Over the past few years there are many companies which are involved in the thorough research of the same which is rising repeatedly. However, handwritten text identification is not a new technique but still it is not popular among masses. The ultimate goal of designing is achieving 100% accuracy which is just an illusion as of now and it is certain that an advanced technology is however necessary to enable the people who cannot even read their own notes.