Department of Electronics and Communication Engineering, Thiagarajar College of engineering, Madurai-15 Email: tcengg2@md3.vsnl.net.in
*Paper contributed to the Annual Seminar on Electronic Sources of Information, 1–3 March 2000, ISI/DRTC, Bangalore - 560059
In today's technological environment, there is a definite need to identify and classify documents reliably. Existing approaches are inadequate in practical applications because of the large dynamic range and high computational cost. This paper discusses the approach that involves a preliminary data reduction step, gray scale projections and transform technique that greatly reduces the computational complexity of the problem. Further classification is achieved by using neural network. The robustness of this approach to different conditions are examined.
