mxT*L.

BImL4L. PRocEsSlNG 1,4-5

(1991)

How I Came Up with the Discrete Cosine Transform
Nasir Ahmed
Electrical and Computer Engineering Albuquerque, New Mexico 87131 Department, University of New Mexico,

During the late sixties and early seventies, there was a great deal of research activity related to digital orthogonal transforms and their use for image data compression. As such, there were a large number of transforms being introduced with claims of better performance relative to others transforms. Such comparisons were typically made on a qualitative basis, by viewing a set of “standard” images that had been subjected to data compression using transform coding techniques. At the same time, a number of researchers were doing some excellent work on making comparisons on a quantitative basis. In particular, researchers at the University of Southern California’s Image Processing Institute (Bill Pratt, Harry Andrews, Ali Habibi, and others) and the University of California at Los Angeles (Judea Pearl) played a key role. In this regard, the variance criterion and the rate distortion criterion were developed and used extensively as performance measures for evaluating image data compression. In addition, the Karhunen-Loeve transform (KLT) evolved as the optimal transform for comparison purposes. With this as background, I can now address the DCT issue. What intrigued me was that the KLT was indeed the optimal transform on the basis of the meansquare-error criterion and the first-order Markov process model, and yet there was no efficient algorithm available to compute it. As such, the focus of my research was to determine whether it would be possible to come up with a good approximation to the KLT that could be computed efficiently. An approach that I thought might be worth looking into was Chebyshev interpolation, a neat discussion of which was available in a text book (Computer Evaluation of Mathematical Functions, by C. T. Fike, Prentice-Hall, 1968, Sections 7.4 and 7.5). This was in early 1972, and I wrote a proposal to the National Science Foundation (NSF)
1051.2004/91$1.50 Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.

to study a “cosine transform” nomials of the form T,(m) = (l/N)‘/“, T,(m) = (2/N)‘%os

using Chebyshev poly-

m = 1, 2, . . . , N em- 2N lh) h = 1 2 ,...) t N.

The motivation for looking into such “cosine functions” was that they closely resembled KLT basis functions for a range of values of the correlation coefficient p (in the covariance matrix). Further, this range of values for p was relevant to image data pertaining to a variety of applications. Much to my disappointment, NSF did not fund the proposal; I recall one reviewer’s comment to the effect that the whole idea seemed “too simple.” Hence I decided to work on this problem with my Ph.D. student Mr. T. Natarajan and my friend Dr. Ram Mohan Rao at the University of Texas at Arlington. In fact, I remember dedicating the whole summer of 1973 to work on this problem. The results that we got appeared too good to be true, and I therefore decided to consult Harry Andrews later that year at a conference in New Orleans. We were both invited speakers at a session on Walsh Functions. Harry suggested that I check out the performance of this “cosine transform” using the rate distortion criterion. He then sent me the computer program to do so. The results again showed that this transform performed better than all the others, and its performance compared very closely with that of the KLT. When I sent the results back to Harry Andrews, he suggested that I publish them. As such, I sent them to the IEEE Computer Transactions, and the paper was then published in the January 1974 issue. I recall that we decided to send it in as a correspondence item in order to get it published with mini-

Dr. India. degree in electrical engineering from the University of Mysore. He became the Chairman of this department in July 1989. Manhattan. and Discrete-Time Signals and Systems (Reston. degrees from the University of New Mexico in 1963 and 1966. St. Minnesota. Since 1983 he has been a Professor in the Electrical and Computer Engineering Department at the University of New Mexico. He received the B. In August 1985 he was awarded one of the twelve Presidential Professorships at the University of New Mexico. From 1966 to 1968 he worked as Principal Research Engineer in the area of information processing at the Systems and Research Center. Ahmed was an Associate Editor for the IEEE Transactions on Acoustics. in 1961. Albuquerque.. respectively. and Signal Processing (1982-1984) and is currently an Associate Editor for the IEEE Transactions on Electromagnetic Compatibility (Walsh Functions Applications). from 1968 to 1983. NASIR AHMED was born in Bangalore. and Ph. Little did we realize at that time that the resulting “DCT” would be widely used in the future! It is indeed gratifying to see that the DCT is now essentially a “standard” in the area of image data compression via transform coding techniques. 1979). Speech.D. Paul. 1983). . and coauthor of Computer Science Fundamentals (Merrill. and the MS. Honeywell. He was with Kansas State University. He is also the author of a number of technical papers in the area of digital signal processing. India. He is the leading author of Orthogonal Transforms for Digital Signal Processing (Springer-Verlag. Inc.mum delay. in 1940.S. 1975).