Discrete orthogonal moments (DOM) are powerful tool for characterization of image shape features in applications of pattern recognition and image analysis. In this paper, a new set of DOM based on the weighted discrete Racah polynomials are considered. This set generate linear unitary transform of initial image called in the paper as Racah Transform (RT). Due to its unitarity, the RT meets formal requirements to transform for use in image compression. Some properties of Racah polynomials and results of comparative experimental research of an information performance of the RT versus the Discrete Fourier and the Discrete Cosinus Transforms are discussed in terms of compression ratio, quality of reconstruction, noise robustness, and computational efficiency.
image compression, coding with transform, quality of reconstruction, noise robustness