Image representation using nonorthogonal basis images with adaptive weight optimization

Image-coding techniques that employ subsets of nonorthogonal basis images chosen from two or more transform domains have been shown consistently to yield higher image quality than those based on one transform for a fixed compression ratio. However, these techniques have not been widely employed due to the very high computational complexity of existing formulations. This paper presents a new adaptive parallel mixed transform technique for image representation. Formulations are derived, adaptive algorithms are described, and results of simulations are presented. Excellent performance is achieved with greatly reduced computational complexity compared to existing mixed-transform techniques, making mixed-transform-based image coding practical.