An efficient coding scheme for biomedical x-ray images based on the DCT basis decomposition is presented. The algorithm is adaptive to the local activities by changing the block sizes. A hierarchical binary tree structure is constructed. The residual sub-images (errors) of the original and the estimation achieved from an interpolation method are decomposed based on the DCT basis functions. Each subimage, regardless of its size, is reconstructed by only three low-frequency DCT coefficients. The computation is rather simple because only three components are manipulated. The experimental results are provided to evaluate the performance of the proposed algorithm.
Hai-Shan Wu, Robert A. King, Joseph Barba, Joan Gil, "DCT Basis Decomposition Coding of Biomedical X-Ray Images Using Vector Quantization" in Journal of Imaging Science and Technology, 1997, pp 158 - 165, https://doi.org/10.2352/J.ImagingSci.Technol.1997.41.2.art00010