Kyeong Ho Yang

A normalized rate-distortion model for H.263-compatible codecs and its application to quantizer selection

This paper proposes a new normalized parametric rate-distortion model which can be efficiently applied to H.263-compatible video codecs and other codecs of similar type. The model is very efficient in terms of computational complexity and amount of required memory since it approximates all 31 rate-distortion relations (one for each quantization parameter QP) with only one curve by normalizing the energy of signals and the quantization distortion with functions of QP. We show how the model parameters can be obtained from experimental data. We also propose a quantizer selection scheme which uses the model to estimate the number of bits required for encoding 8/spl times/8 residual blocks. Experiments show that the model provides very good approximations of the rate and the distortion characteristics for all QPs, and that the proposed quantizer selection scheme works better than the scheme used in the MPEG-4 Video Verification Model Version 5.1.

A fast and memory efficient algorithm for down-conversion of an HDTV bitstream to an SDTV signal

In the future of the digital television (DTV) broadcasting, a standard definition television (SDTV) decoder should have the function of receiving a high definition television (HDTV) signal and display it on its monitor. In order to do so, it is required to convert an HDTV bitstream down to an SDTV signal. In this paper, we propose a fast and memory efficient algorithm for the down-conversion. The proposed algorithm saves lots of computation of the down-conversion by combining inverse discrete cosine transform (IDCT) with the pixel-domain down-conversion. It also requires a smaller memory by performing motion compensation (MC) after the down-conversion, compared to the conventional scheme where MC is performed before the down-conversion. In the brute-force IDCT case, the computation savings of our proposed algorithm are 62%, 74% and 84% for down-sampling ratios of 2:1, 8:3, and 4:1, respectively, with the same picture quality compared to the conventional memory efficient pixel-domain approach. In the case of using a fast IDCT algorithm, the computational savings are respectively 25%, 31% and 37% for down-sampling ratios of 2:1, 8:3, and 4:1.

An efficient memory allocation scheme for Huffman coding of multiple sources

In this paper, we propose an efficient memory allocation scheme for memory-constrained Huffman coding of multiple sources, which can be applied to many adaptive variable-length coding systems. The allocation of a given memory is performed in two stages. At the first stage, the iterative bisection algorithm based on the Lagrange optimization method is used to find a Lagrange allocation, which is either optimal or close to the optimal allocation. In the latter case, the Lagrange method does not fully allocate the given memory and a sequential allocation method is introduced as the second stage to allocate the remaining memory, which is performed in a greedy manner. To explain the proposed allocation scheme, we introduce a function, b(l), which approximates the relation between the average bitrate b and the Huffman table size l, and discuss its properties. The use of this function considerably reduces the computational burden for memory allocation. We apply the proposed memory allocation scheme to memory-constrained conditional entropy coding of vector quantization indices for image compression. Simulations show that the proposed memory allocation scheme provides almost optimal performance, which is far better than can be achieved with conventional simple allocation methods.