The development of handwriting recognition systems began in the 1950s when there were human operators whose job was to convert data from several documents into electronic format, making the process rather long and often affected by errors. Automatic text recognition aims to limit these errors using image preprocessing techniques that increase the speed and accuracy of the entire recognition process. Writing recognition has been one of the most fascinating and challenging areas of research in the field of image processing and pattern recognition in recent years. It contributes greatly to the advancement of the automation process and improves the interface between man and machine in numerous applications. Optical character recognition is a field of study that can cover many different resolution techniques. Neural networks (Sandhu and Leon, 2009), vector support machines and statistical classifiers appear to be the preferred solutions to the problem because of their proven accuracy in the classification of new data.
Optical Character Recognition is actually a converter that translates handwritten text images into text based on the machine. In general, handwriting recognition is classified into two types as off-line and on-line. In off-line recognition, writing typically captures a scanner optically and full writing is available as an image. In other words, handwritten text is when handwritten text is scanned by a scanner in a digital format. But in the online system, the two-dimensional coordinates of the successive points are represented as a function of the time and the order of the strokes made by the writer. In other words, the X-Y coordinates are given as a result indicating the location of the pen and the force applied by the user during writing and the speed as well. Handwritten text online is written with a stylus on a tablet. There is also a third method which is not as famous as the first two methods mentioned above in which laser, inkjet devices, can be used to obtain machine-printed text.
There is extensive work in the field of handwriting recognition, and there are several revisions. Online methods have proven to be superior to their off-line counterparts in handwriting recognition because of the temporary information available with the former. However, various applications, including mail sorting, bank processing, document reading, and postal address recognition require offline handwriting recognition systems. In addition, in off-line systems, neural networks and support vector machines have been successfully used to achieve comparatively high levels of recognition accuracy. As a result, offline handwriting recognition remains an active area for research toward exploring new techniques that would improve recognition accuracy. Therefore, for this report, I have decided to work on an alphabetic character recognition system by hand written online using the Neural Network of Rear Propagation, the LAMSTAR Neural Network and the Vector Support Machine (SVM).
The Artificial Neural Network (ANN) is a computational model of the brain, which has parallel distributed processing elements that are learned by adjusting the weights connected between neurons. Due to its flexibility and strength, it has now been widely used in different fields, such as pattern recognition, decision-making optimization, market analysis, robotic intelligence. ANN may be more notable as computational processors for different tasks such as data compression, sorting, combinatorial optimization of problem solving, pattern recognition, etc. ANN has many advantages over other classical methods. While computational complexity, ANN offers many advantages in recognizing patterns of adaptation from a very little context of human intelligence. In the off-line recognition system, neural networks have emerged as the fast and reliable tools for classification to achieve high recognition accuracy. Since the 1990s, classification techniques have been applied to the recognition of manuscript characters. These methods include statistical methods based on the Bayes decision rule, Artificial Neural Networks (RNAs), Kernel Methods including Support Vector Machines (SVM) and multiple classifier combination.