Abstract
Image data compression is an important research area in image processing. A basic image data compression technique is to use an invertible linear transformation, such as Fourier, cosine, sine, and Hadamard transformations, to generate approximate uncorrelated coefficients in the transform domain. The optimal uncorrelated transform, the so-called Karhunen-Leove, coefficients are image dependent. Here, computationally efficient linear, invertible integer, the so-called Newton and Stirling, transforms are proposed. Integer arithmetic promises substantial computational savings. Some of the properties of these transforms are described.
© 1989 Optical Society of America
PDF ArticleMore Like This
O. K. Ersoy and D. Y. Kim
MD1 OSA Annual Meeting (FIO) 1986
N. S. Tzannes, John S. Bodenscharz, and M. A. Tzannes
MO3 OSA Annual Meeting (FIO) 1987
V. Michael Bove
WB3 Image Understanding and Machine Vision (IUMV) 1989