HomeChair: Jelena Kovacevic, Lucent Technologies, USA
Home
Faramarz Fekri, Georgia Institute of Technology (U.S.A.)
Russell M. Mersereau, Georgia Institute of Technology (U.S.A.)
Ronald W Schafer, Georgia Institute of Technology (U.S.A.)
In this paper we discuss the problem of reconstruction of a high resolution image from a lower resolution image by a jointly optimum interpolative vector quantization method. The interpolative vector quantizer maps quantized low dimensional 2x2 image blocks to higher dimensional 4x4 blocks by a table lookup method. As a special case of generalized vector quantization (GVQ), a jointly optimal quantizer and interpolator (GIVQ) is introduced to find the corresponding codebooks for the low and high resolution image. In order to incorporate the nearest neighborhood constraint on the quantizer and also to obtain the desired distortion in the interpolated image, a deterministic annealing based optimization technique has been applied. With a small interpolation example, we will demonstrate the superior performance of this method over nonlinear interpolative vector quantization (NLIVQ), in which the interpolator is optimized for a given input quantizer.
ic982124.pdf (From Postscript)
TOP
Konstantinos Masselos, University of Patras (Greece)
Thanos Stouraitis, University of Patras (Greece)
Costas E Goutis, University of Patras (Greece)
Novel algorithms for vector quantization codebook design are presented in this paper. Two basic techniques are proposed. The first technique takes into consideration specific characteristics of the blocks of the training sequence during the generation of the initial codebook. In this way a representative initial codebook is generated. Starting from a high quality initial codebook the iterative optimization procedure converges fast to a representative final codebook which in turn leads to high output image quality. The second proposed technique extends small codebooks computationally. The main idea is the application of simple transformations on the codewords. This technique reduces the memory requirements of the traditional vector quantization making it useful for applications requiring low-power consumption.
ic981820.pdf (From Postscript) TOP
Shin-Chou Juan, National Chiao-Tung University (Taiwan)
Chen-Yi Lee, National Chiao-Tung University (Taiwan)
Side-match VQ(SMVQ) is a well-known class of FSVQ used for low-bit rate image/video coding. It exploits the spatial correlation between the neighboring blocks to select several codewords that are very close to the encoding block from the master codebook. But if the block boundary is in the region edge area, the spatial correlations are not high and the SMVQ can't select proper codewords to encode blocks. In this paper, an Entropy-Constrained Gradient-Match VQ (ECGMVQ) is proposed. Instead of exploiting the spatial correlation, the ECGMVQ uses the gradient contiguity property to select the codewords. State function of ECGMVQ can select better codevectors than the SMVQ. In addition, the entropy-constrained rule is applied to the encoding process to reduce bit rate. Simulation results show that the improvement of ECGMVQ over the SMVQ is up to 4~5 dB at nearly the same bit rate. Further, the perceptual image quality is better than that of SMVQ, especially in the region edge area.
ic981275.pdf (Scanned) TOP
Hsien-Chung Wei, National Tsing Hua University (Taiwan)
Pao-Chin Tsai, National Tsing Hua University (Taiwan)
Jia-Shung Wang, National Tsing Hua University (Taiwan)
Several effective low bit rate still image compression methods have been presented in these two years, such as SPHIT [9], Hybrid VQ [7], Wu and Chen method [10]. These methods exercise the analysis techniques (wavelet or subband) before distributing the bit rate to each piece of image, thus the tradeoff between bit rate and distortion can be resolved. In this paper, we try to propose a simple but comparable method that adopts the technique of side match VQ only. Side match vector quantization (SMVQ) is an efficient VQ coding scheme for low bit rate coding. Conventional side match (two-sided) utilizes the codeword information of two neighboring blocks to predict the state codebook of an input vector. In this paper, we propose a hierarchical three-sided side match finite-state vector quantization (HTSMVQ) method that can (1) make the state codebook size as small as possible, it can be reduced to 1 if the prediction is performed perfectly; (2) improve the prediction quality for edge blocks; (3) regularly refresh the codewords to alleviate the error propagation of side match. In the simulation results, the image ""Lena"" can be coded with PSNR 34.682 dB at 0.25 bpp. It is better than SPIHT, EZW, FSSQ and hybrid VQ with 34.1, 33.17, 33.1 and 33.7 dB, respectively. At the bit rate lower than 0.15 bpp, only the enhanced versions of EZW perform better than our method about 0.14 dB.
ic981907.pdf (From Postscript) TOP
Kazuhiko Terada, Kyoto University (Japan)
Masahiro Takeuchi, Kyoto University (Japan)
Kazutoshi Kobayashi, Kyoto University (Japan)
Keikichi Tamaru, Kyoto University (Japan)
In this paper, we propose a low bit-rate coding algorithm for wireless communication based on multi-stage hierarchical vector quantization, motion compensation and differential pulse code modulation. Our method adapts bit allocation to spatial and temporal correlation. Conventional schemes based on discrete cosine transform (DCT) need a large amount of computation on both encoding and decoding. On the other hand, our proposed method consists of addition, subtraction and shift operation. It does not use multiplication. It can decode in real time on a conventional serial processor. Encoding by vector quantization (VQ), however, consumes a large amount of computation. We developed a new LSI to accelerate VQ. Our scheme can send 10 frames of QCIF video sequences through a 29.2kbps line. The quality of reconstructed image is over 30dB.
ic982278.pdf (From Postscript) TOP
Lai-Man Po, City University of Hong Kong (Hong Kong)
Ying Zhang, Guangdong Posts & Telecom (China)
Kwok-Wai Cheung, City University of Hong Kong (Hong Kong)
Chun-Ho Cheung, City University of Hong Kong (Hong Kong)
In this paper, a novel wavelet subtree partitioning algorithm is proposed, which divides a subtree into scalar quantized wavelet coefficients and fractal coded sub-subtree. Based on this new technique, a variable size wavelet subtree fractal coding scheme for still image compression is developed. Experimental results show that the new scheme can achieve nearly optimal partition of wavelet subtree with substantially computational reduction as compared with Davis' scheme.
ic981161.pdf (Scanned) TOP
Lowell L Winger, University of Toronto (Canada)
Anastasios N Venetsanopoulos, University of Toronto (Canada)
The selection of filter bank in wavelet compression is crucial, affecting image quality and system design. Recently, the biorthogonal coiflet (cooklet) family of wavelet filters has been constructed [2][4], and explicit frequency domain formulae have been developed [2] in the Bernstein polynomial basis. In this paper we use the Bernstein basisfor frequency domain design and construction of biorthogonal nearly coiflet wavelet bases. In particular, we construct a previously unpublished nearly coiflet 17/11 biorthogonal wavelet filter pair. Key filter quality evaluation metrics due to Villasenor demonstrate this filter pair to be well suited for image compression. Comparison is made to the 17/11 biorthogonal coiflet (cooklet), Villasenor 10/18, Odegard 9/7, and classical CDF 9/7 wavelet bases. Simulation results with the SPIHT algorithm due to Said and Pearlman [3], and with our SRSFQ [7] [5], confirm that the new 17/11 wavelet basis outperforms the others for still image compression.
ic981265.pdf (From Postscript) TOP
Sang-hyun Joo, Niigata University (Japan)
Hisakazu Kikuchi, Niigata University (Japan)
Shigenobu Sasaki, Niigata University (Japan)
Jaeho Shin, Dongguk University (Korea)
We introduce a new zerotree scheme that effectively exploits the inter-scale self-similarities found in the octave decomposition by a wavelet transform. A zerotree is useful to code wavelet coefficients and its effectiveness was proved by Shapiro's EZW. In this paper, we analyze symbols produced from the EZW and discuss the entropy per symbol. Since the entropy depends on the produced symbols, we modify the procedure of symbol generation. First, we extend the relation between a parent and children used in the EZW to increase the probability such that a significant parent has significant children. The proposed relation is flexibly extended according to the fact that a significant coefficient is likely to have significant coefficients in its neighborhood. Our coding results are compared with the published results. Our proposed coder owes its improved performance to the use of lower entropy per symbol. The comparison of the number of produced symbols is also given.
ic981719.pdf (Scanned) TOP
Sergio Rodrigues Neves, Instituto de Pesquisas da Marinha (Brazil)
Gelson Vieira Mendonça, COPPE/EE/UFRJ (Brazil)
The increasing interest on image compression is due to the increasing necessity of transmission velocity and memory to keep digital images, on many kind of midia. In order to reduce the quantity of bits needed to store images and, consequently, increase their transmission velocity, many coding methods have been studied, especially those which utilize transforms. This paper presents an image compression technique that codifies edges and textures separately using the wavelet transform. Several test images have been coded with the proposed method and the results show that the reconstructed images have good performance in relation to their PSNR and, especially, in terms of visual perception.
ic981808.pdf (Scanned) TOP
Jill R. Goldschneider, University of Washington (U.S.A.)
Eve A. Riskin, University of Washington (U.S.A.)
In this paper, an algorithm for wavelet packet trees that can systematically identify all bit allocations/best-basis selections onthe lower convex hull of the rate-distortion curve is presented. The algorithm is applied to tree-structured vector quantizers used to code image subbands that result from the wavelet packet decomposition. This method is compared to optimal bit allocation for the discrete wavelet transform.
ic981036.pdf (From Postscript) TOP