Title: Reversible Denoising and Lifting Based Color Component Transformation for Lossless Image Compression

Abstract: An undesirable side effect of reversible color space transformation, which consists of lifting steps, is that while removing correlation it contaminates transformed components with noise from other components. To remove correlation without increasing noise, we integrate denoising into the lifting steps and obtain a reversible image component transformation. For JPEG-LS, JPEG 2000, and JPEG XR algorithms in lossless mode, we find that the proposed method applied to the RDgDb color space transformation with a simple denoising filter is especially effective for images in the native optical resolutions of acquisition devices, but may lead to increased bitrates for typical images. We also present an efficient estimator of image component transformation effects.