Awarded by National ICT R & D Commission, Pakistan to Abasyn University Islamabad Campus
Mobile Computing is considered to be the next big thing on the face of technology and availability of platforms like smart phones is going to play a very important part in enabling the mobile computing to achieve its true potential. Even in the time of economic crunch in 2008-2009 smart phones market has been seeing continuous improvement and big guns like Google, RIM and PALM have made heavy investments in this market in one form of the other. Some of the companies have focused on the hardware while some of the companies have realized that availability of killer applications on the smart phones can enhance their revenue many times.
Applications development on the smart phones is comparatively newer field. Although the techniques like J2ME (Java2 Micro Edition) have existed for quite some years but as the name indicates they had micro set of features. Only after the availability of iPhone and Android enabled smart phones, the world has seen the availability of the functionally complete APIs for smart phone applications development. This was a big step as the future of smart phones market depends heavily on the availability of killer applications.
Mobile environment, though constrained, provides some interesting features and facilities not directly available on the desktop computers. One such feature is the availability of integrated cameras on most of the smart phones. This feature can be used to provide killer applications related to Character Recognition but before that some API / framework and engine should be there that can provide the basic building blocks for the development of applications related to Character Recognition. This proposal has been inspired by the motivations described above. We want to develop a state of the art OCR Engine which can be used by the developers to develop their own applications. We intend to divide the project into two different phases. In the first phase we want to tackle the character recognition problem for the constrained document images and in the next phase we want to tackle character recognition problem for the unconstrained document images. This document gives a top level overview of the techniques and algorithms that we intend to use in this project. Besides giving the top level design, we have also discussed some of the candidate algorithms. Almost all the commercially available OCR systems have image pre processing modules associated with it. We have also divided our system into four steps: Image acquisition, Image preprocessing and segmentation, Classification / Recognition of the text and Post Processing. Experience of our Team in different dimensions of character recognition namely Image Pre Processing, Pattern Recognitions, and Smart Phones Development augurs well for the success of this project.
Success of this project would not only result in research experience and related advantages for the developers inside Pakistan but will also stimulate the smart phones software development in Pakistan. Last but not the least it would give a great opportunity for the students of Abasyn University and other local universities to work on the top notch research problems of the field. Open source nature of the project would encourage collaboration from developers all over the world.