Bidimensional Empirical Mode Decomposition for multiresolution image coding

Abstract

Bidimensional Empirical Mode Decomposition (BEMD) provides a tool for image processing for its special ability to locally separate spatial frequencies. The Intrinsic Mode Functions (IMFs) other than the first are images of lower frequency components. The coding method presented in this paper uses the BEMD process to create two sub-band images from the original image: hi and low frequency. These images are differently subsampled and compressed by JPEG standard with different quality factor and then sent to a feedback system to compensate the quality loss in the compression process. The proposed paper proposes a technique to enhance jpeg encoding of images. Reconstructed images produce increased visual appearance of correspondent jpeg coded ones at same compression rate.