Chair: Martin Vetterli, University of California-Berkeley (USA)
E.A.B. da Silva, University of Essex (UK)
D.G. Sampson, University of Essex (UK)
M. Ghanbari, University of Essex (UK)
A novel coding method of wavelet coefficients of images using vector quantization, referred to as Successive Approximation Wavelet Vector Quantization (SA-W-VQ), is proposed. In this method, each vector is coded by a series of vectors of decreasing magnitudes until a certain distortion level is reached. Analysis of the successive approximation using vectors is given, and conditions for convergence are derived. It is shown that lattice codebooks offer an efficient tool to meet these conditions, with the extra advantage of fast encoding algorithms. In SA-W-VQ, distortion equalization of the wavelet coefficients can be achieved together with high compression ratio and precise bit rate control. Simulation results for still image coding show that SA-W-VQ outperforms both the EZW coder [1] and the standard JPEG.
Patrick Lau, University of Minnesota (USA)
Nikolaos P. Papanikolopoulos, University of Minnesota (USA)
This paper addresses the issues of time and compression efficiency in image transformation. We propose a novel basis selection scheme which improves transformation efficiency by exploiting the energy compacting characteristic in the frequency domain. In a typical complete transformation, a large number of basis functions have low coding efficiency, and thus are not necessary in the encoding process. By removing these functions from the basis set, we improve the time and the compression efficiency of the encoding process while maintaining a high reproduction quality. We have chosen to use the Gabor Transform to demonstrate our proposed method. Experimental results with the Gabor Transform are presented to demonstrate the effectiveness of our method. Finally, issues related to the application of this approach in image sequence encoding and the adaptation of our approach to other transformation schemes are discussed.
Wei Li, Swiss Federal Institute of Technology at Lausanne (SWITZERLAND)
Olivier Egger, Swiss Federal Institute of Technology at Lausanne (SWITZERLAND)
Murat Kunt, Swiss Federal Institute of Technology at Lausanne (SWITZERLAND)
This paper addresses the problem of the quantization noise reduction in subband image coding schemes. Two major artifacts occur for such coding schemes at high compression factors: the ringing effect around high--contrast contours and the blurred false contours in large smooth regions. The first distortion can be considerably reduced by an appropriate design of the subband filters. The second one can be eliminated by using the noise reduction technique proposed in this paper, which consists of applying a noise reduction filter to the DC subband. The advantages of this approach are as follows: First, it can be applied to any kind of subband decompositions. Second, it removes quantization noise to which the eye is most sensitive and third, it is computationally very efficient due to the small size (typically 64 X 64) of the DC subband. The colored quantization noise in the DC subband is rendered white by using the Roberts pseudonoise technique. The proposed noise reduction filter is a Wiener type filter with adaptive directional support. It has the advantage of reducing the noise without blurring the reconstructed image. It is shown that the proposed noise reduction filter augments the visual quality of the reconstructed image as well as its PSNR value.
Charles D. Creusere, Naval Air Warfare Center (USA)
In this paper, we compare for image coding applications a low-complexity IIR wavelet based on an allpass polyphase decomposition to a pair of linear phase biorthogonal wavelets. To code the wavelet coefficients, we use Shapiro's zerotree algorithm which has the virtues of being both efficient and delivering excellent performance (in a rate-distortion sense). We consider a variety of methods for eliminating filter transients at image boundaries including circular convolution, symmetric extension (for the biorthogonal wavelets), and direct transmission (for the IIR wavelet). By also coding the filter states in a zerotree form, we find that direct transmission generally performs better than circular convolution. Finally, we show that the use of this IIR wavelet provides equivalent performance to the biorthogonal wavelets at greatly reduced computational complexity.
Euee S. Jang, SUNY at Buffalo (USA)
Mahesh Venkatraman, SUNY at Buffalo (USA)
In this paper we discuss a method of compression of 3D MRI images. A subband coding scheme with an octave splitting is proposed. The higher frequency subbands are encoded using an octtree structure. Octtrees provide a convenient data structure to store the sparse information present in the higher frequency subbands. Each unit block of the octtree is coded using a hybrid vector and scalar quantization scheme. The baseband is encoded using a DPCM based method using a nonlinear neural network predictor. This scheme provides an elegant way of compression of MRI images with minimum error.
Faouzi Kossentini, Georgia Institute of Technology (USA)
Wilson C. Chung, Georgia Institute of Technology (USA)
Mark J.T. Smith, Georgia Institute of Technology (USA)
An iterative design algorithm for the joint design of complexity- and entropy-constrained subband quantizers and associated entropy coders is proposed. Unlike conventional subband design algorithms, the proposed algorithm does not require the use of various bit allocation algorithms. Multistage residual quantizers are employed here because they provide greater control of the complexity-performance tradeoffs, and also because they allow efficient and effective high-order statistical modeling. The resulting subband coder exploits statistical dependencies within subbands, across subbands, and across stages, mainly through complexity-constrained high-order entropy coding. Experimental results demonstrate that the complexity-rate-distortion performance of the new subband coder is exceptional.
Won-ha Kim, University of Wisconsin- Madison (USA)
Yu-Hen Hu, University of Wisconsin- Madison (USA)
In this work, we propose an algorithm to use in designing a subband coder (SBC) constructed by wavelet packet, to achieve minimum distortion for given bit budget and implementation complexity. We map the QMF tree structures onto a binary tree, then formulate the task as an optimization problem including coding bit and implementation complexity constraints. The problem is dissected into two phases. First, we derive the optimal bit allocation strategy which covers the entire range of bit rate, and second, we search for the optimal subband decomposition by using a fast dynamic program.
Sergio Servetto, University of Illinois (USA)
Kannan Ramchandran, University of Illinois (USA)
Michael Orchard, University of Illinois (USA)
We propose an improved statistical characterization of the field of wavelet coefficients of natural images. Based on this characterization, we introduce Morphological Representation of Wavelet Data (MRWD), a novel coding framework for both image and video coding applications. MRWD departs from existing wavelet-based coders in its use of a radically different set of primitive operations --non-linear, morphological operations--, for efficiently encoding the wavelet data field. Simulation results are very encouraging: a preliminary algorithm based on the morphological data structure is able to achieve about 0.5 dB of gain in SNR over Shapiro's state-of-the-art zerotree-based wavelet coder at a coding rate of 1 bpp for the ``Lenna'' image.
John R. Smith, Columbia University (USA)
Shih-Fu Chang, Columbia University (USA)
In this paper we consider a method for image compression based on frequency and spatially adaptive wavelet packets. We present a new fast directed acyclic graph (DAG) structured decomposition, with both spatial segmentation and orthogonal frequency branching from each node. Whereas traditional wavelet packet decomposition adapts to a global frequency distribution, this technique finds the best joint spatial segmentation and local frequency basis. The algorithm is derived from the fast double tree algorithm proposed by Herley, et al., for 1-D signals [1], with an extension to 2-D and modification to include spatial segmentation of frequency nodes. By collecting redundant nodes in this full adaptive tree, we have derived a directed acyclic graph (DAG) structure which contains the same number of nodes as the double tree, but includes new connections between nodes. We present the adaptive wavelet packet DAG algorithm and examine image compression performance on test images.
Olivier Egger, Swiss Federal Institute of Technology at Lausanne (SWITZERLAND)
Andre Nicoulin, Swiss Federal Institute of Technology at Lausanne (SWITZERLAND)
Wei Li, Swiss Federal Institute of Technology at Lausanne (SWITZERLAND)
n this paper the problem of low bit rate image coding is addressed. The proposed approach is based on the embedded zerotree wavelet (EZW) algorithm. At high compression ratios, visually annoying artifacts appear in the reconstructed images using the originial EZW algorithm. Hence, the first goal of this paper is to improve the visual quality of the coded pictures. The main distortion is due to the Gibbs phenomenon of the linear filter bank and appears as ``ringing effect'' in the reconstructed images. The ringing effect is shown to be considerably reduced by using asymmetrical filter banks. It can even be completely removed by means of a morphological subband decomposition. The major drawback of the morphological filter bank is the loss of textures at high compression factors. The second objective is to make an optimal use of the embedded bit stream by reducing the computational cost of the decoding process. A new decomposition is proposed which reduces drastically the computational load of the synthesis stage.