Choong Woong Lee

A VLSI chip for motion estimation of HDTV signals

Presents a VLSI chip for realtime implementation of a motion estimator in an HDTV encoder for the purpose of transmission and/or broadcasting. Using this chip, it is possible to implement the estimator of HDTV signals based on SMPTE-260M, a HDTV digital interface standard. The search area can be easily expanded to the required range also at high system clock-rate upto 80 MHz. >

Motion-compensated wavelet transform coder for very low bit-rate visual telephony

Abstract This paper proposes a new motion-compensated wavelet transform video coder for very low bit-rate visual telephony. The proposed coder sequentially employs: (1) selective motion estimation on the wavelet transform domain, (2) motion-compensated prediction (MCP) of wavelet coefficients, and (3) selective entropy-constrained vector quantization (ECVQ) of the resultant MCP errors. The selective schemes in motion estimation and in quantization, which efficiently exploit the characteristic of image sequences in a visual telephony, considerably reduce the computational burden. The coder also employs a tree structure encoding to represent efficiently which blocks were encoded. In addition, in order to reduce the number of ECVQ codebooks and the image dependency of their performance, we introduce a preprocessing of signals which normalizes input vectors of ECVQ. Simulation results show that our video coder provides good PSNR (peak-to-peak signal-to-noise ratio) performance and efficient rate control.

Fast competitive learning with classified learning rates for vector quantization

Abstract In this paper, we present a new competitive learning algorithm with classified learning rates, and apply it to vector quantization of images. The basic idea is to assign a distinct learning rate to each reference vector. Each reference vector is updated independently of all the other reference vectors using its own learning rate. Each learning rate is changed only when its corresponding reference vector wins the competition, and the learning rates of the losing reference vectors are not changed. The experimental results obtained with image vector quantization show that the proposed method learns more rapidly and yields better quality of the coded images than conventional competitive learning method with a scalar learning rate.

Sequential vector quantization of directionally decomposed DCT coefficients

A new transform domain vector quantizer has been introduced for efficient image coding which employs directional decomposition and sequential vector quantization of DCT coefficients (DCT-SVQ): the DCT coefficients are decomposed into 16 slightly-overlapping subvectors for a subvector to convey key information about one directional image, and then the decomposed subvectors are vector quantized sequentially. The scheme has the capability of the rate control with no need for changing the codebooks and reveals good performance over a wide range of code rates. This paper addresses the improvement of that scheme by (1) adjusting the quantization step size of the DC quantizer as well as the AC quantizers according to the target distortion of the reconstructed image and (2) introducing the activity of an image block to the decision of the appropriate stopping condition of the sequential vector quantizer for that block.<<ETX>>

Sequential vector quantization of perceptually windowed DCT coefficients

In order to provide a high quality of the reconstructed images with a reduced transmission bit rate, taking into account both statistical redundancy and perceptual characteristics of the DCT coefficients, we apply perceptual windowing and sequential vector quantization to encoding of the coefficients: first, the coefficients are decomposed into 16 slightly overlapping subvectors so as that each subvector has reasonable dimension and conveys key information about one directional image, and then the decomposed subvectors are quantized in a sequential manner. The proposed scheme is good at encoding images with a wide range of transmission bit rates which can be easily controlled by adjusting only the stopping criterion of the sequential vector quantizer.

Wavelet transform coding for MCP error images using quadtree decomposition and ECVQ

Abstract This paper proposes a new wavelet transform video coder which employs motion compensation, wavelet decomposition, and entropy-constrained vector quantization (ECVQ), in sequence. Each of layered subimages obtained from wavelet decomposition is segmented into basic blocks, and then the blocks are selectively encoded by ECVQ according to the energy of the samples. We introduce an efficient method to encode the map representing which blocks are encoded, based on inter-band prediction followed by a quadtree encoding. The proposed coder uses a simple forward analyzer in order to optimize the encoding parameters and introduces a preprocessing of signals which normalizes the input vectors of ECVQ in order to reduce the image-dependency of ECVQ codebooks. Simulation results show that our video coder provides good PSNR (peak-to-peak signal-to-noise ratio) performance and efficient rate control.

Optimal codebook design for a trellis-searched vector quantization

This paper proposes a new trellis-searched vector quantization scheme with an optimal codebook design algorithm. The proposed vector quantizer employs a conditional index encoder to efficiently exploit high interblock correlation, where each index is adaptively encoded using a different Huffman table which is selected according to the index of the previously decoded block. We also modify the Viterbit algorithm in order to find the optimal path through the trellis of indices. For an optimal codebook design an iterative descent method is used, which is very similar to that for ECVQ (entropy-constrained vector quantization) codebook design. Simulation results show that at the same bitrate the proposed method provides higher PSNR performances than the conventional ECVQ by 1.0 approximately 3.0 dB when applied to image coding.

Edge preservative VQ based on the projection of images

The authors present an image compression scheme which has a good image quality based on vector quantization (VQ) of projections instead of the image itself and on simple postprocessing. The projection procedure makes it possible to transmit the edge preserving information. Usually the vector dimensions of projections are much smaller than that of the block, so it is possible to process a larger block in the projection domain than in the spatial domain with similar computational complexity. Processing large blocks can reduce the transmission bit rate by exploiting more interpixel correlation. As the block size increases, the blocking effect is more pronounced on background areas. To reduce this artifact, postprocessing is required. A multiorder polynomial interpolator for postprocessing the pixels was applied at and near the boundaries of background blocks. Simulation results show that the performance of the suggested scheme was superior to that of the normal VQ scheme. >