Reversible DCT and It's Application to Image Coding

Abstract

In this paper, we propose a reversible discrete cosine transform. The transform and inverse transforms are represented by simple equations so that the amount of computations is very small. If floor functions are ignored, the proposed transform is exactly the same as DCT. No redundancy exists in transform domain, since the determinant is equal to 1,that is the density of transformed points is equal to 1. Furthermore, the transform is normalized so that we can avoid the problem that dynamic range is nonuniform. The lossless and lossy compression efficiencies of the proposed method are compared with conventional methods in consideration of it's application to unified lossless and lossy coding of still images. It is shown that lossless compression efficiency of the proposed method is almost the same as reversible wavelet transform and lossy one is almost the same as reversible wavelet transform and discrete cosine transform.