Summary form only given. We address a new lossy coding method for binary images using high-order context model based on vector quantization (VQ). Binary images, such as engineering drawings and weather maps, usually contain structured patterns as well as large white space. In order to utilize structured information, the proposed approach employs the new block-based high-order context modeling. The advantage of having block context is a reduced complexity for the same amount of context coverage. In the proposed approach, the context is determined based on adjacent blocks. The number of block pattern is quite large even for a moderate block size. In practice, the probability distribution of all the block pattern is highly non-uniform. Therefore it is possible to reduce the number of contexts significantly without sacrificing compression performance. We use a node merging method reduce the number of block patterns for context nodes. The proposed compression algorithm utilizes two methods; a context model based binary VQ and JBIG algorithm. The compression results are good with the compression ratios ranging from 20 to 60. It was observed that the computer generated characters have better visual quality than hand written characters. The test images were engineering drawing maps for pipelines and CCITTT test images.

We present a new lossy technique for document image compression by using straight line extraction and a block context model. Straight line segments are extracted from a binary document image and subtracted from the original image. Their endpoint coordinates and width can then be efficiently coded. The remaining part of the image, which mainly contains text and other symbols, is coded using a high-order block context model (HOBCM) based on vector quantization (VQ). The proposed method is particularly effective for document images containing a large number of straight line segments, such as engineering or architectural drawings. It achieves much higher compression than conventional lossless techniques, such as the JBIG and CCITT G3 and G4 standards, with little loss of visual quality. In the experiments we carried out, a group of engineering drawings digitized at 200 dpi, compression ratios ranging from 30 to 70 were obtained.

A new speech LSP parameters coding scheme for noisy channels is proposed. In this scheme, 5 out of the 10 LSPs are scalar quantized. Non-linear prediction is used for coding the other 5 LSPs and the prediction errors are vector quantized. The 5 LSPs are chosen in the structure so that the channel error will not affect more than two frame LSPs. This new scheme can support various levels of channel coding. In noiseless channels it needs 27 bits/frame at 1 dB average spectral distortion with a frame size 30 ms. With additional 4 bits/frame to protect a part of the 27 bits, a (15,11) Hamming code achieves 1 dB average distortion at the error rate 0.1%.

A new finite-state vector quantization (FSVQ) algorithm is developed based on state space optimization and the derailment prevention requirement. The proposed derailment-free FSVQ (DF-FSVQ) achieves good performance through (i) state space reduction, which allows a practical implementation of high-order FSVQ, and (ii) derailment free state transitions. Experimental results show that our approach outperforms other known FSVQ schemes in terms of performance, system complexity, and processing speed, especially at low bit rate image coding.

A new approach to the design of a finite-state vector quantizer (FSVQ) is proposed. FSVQ essentially exploits correlations between adjacent blocks for efficient coding. Previous FSVQ design schemes had ad-hoc features in defining states and resource allocation using equal number of bits for state codebooks regardless of their probabilities of occurrence in a given source. We propose a FSVQ design approach which improves the compression performance by merging states and using variable state-codebook sizes. Another undesirable feature of the FSVQ is a derailment problem which degrades the performance in many practical applications. We propose a structurally constrained state-codebook design approach that eliminates the derailment problem. The performance of the proposed algorithm outperforms previously known FSVQ methods. Further development of the algorithm utilizing mean-removed VQ is described which gives less block artifact even though PSNR is a little bit inferior.

The authors present a new pruned tree structured vector quantization (TSVQ) algorithm, called incremental tree growing (ITG), which incrementally grows and quantizes the context tree locally on a level by level basis, thus drastically reducing both the memory requirements and the computational complexity. After the incremental tree growing is completed, terminal branches are globally vector quantized again, which is possible due to significant reduction of the number of initial branches. Using a technique similar in spirit to the mean removed VQ (MRVQ) of Baker and Gray (1983), significant reduction in the number of probability tables is achieved. In one of their simulations, they reduced the number of probability tables from 262 K(=2/sup 18/) to 108 by using the ITG algorithm, and from 108 to 10 using the modified MRVQ at the conditional entropy increase of only 1.26%. In summary, the proposed ITG algorithm combined with a modified MRVQ provides an efficient framework for the design of context-trees with reduced memory and computational requirements.>

High order statistical modeling is a promising technique for data compression. Previously we introduced an algorithm for the design of efficient high order statistical models of digital images. The system consisted in mapping the context set to a tree-structured vector quantizer, the context-tree, and using vector quantization techniques to reduce the size of the tree and, thus, simplify the model. While the models obtained through this off-line design procedure are very efficient they can only be used to code images in the same class as those represented in the training set. In this paper, we show that high order models can be obtained adaptively by constructing the context-tree with a wider range of image classes and allowing the coder to adaptively collect the statistics as the input is processed. Since the structure of the model (context-tree) is kept fixed and only the conditional distributions are allowed to vary the overall scheme is semi-adaptive. The compression efficiency is better than the conventional DPCM+entropy scheme by 8% to 36% for the images used for the simulations.>

An approach to obtaining high-resolution image reconstruction from low-resolution, blurred, and noisy multiple-input frames is presented. A recursive-least-squares approach with iterative regularization is developed in the discrete Fourier transform (DFT) domain. When the input frames are processed recursively, the reconstruction does not converge in general due to the measurement noise and ill-conditioned nature of the deblurring. Through the iterative update of the regularization function and the proper choice of the regularization parameter, good high-resolution reconstructions of low-resolution, blurred, and noisy input frames are obtained. The proposed algorithm minimizes the computational requirements and provides a parallel computation structure since the reconstruction is done independently for each DFT element. Computer simulations demonstrate the performance of the algorithm.

An approach to high-resolution image reconstruction from blurred noisy multiframes is presented. A recursive least square estimation with regularization method is applied for high resolution estimation. This approach provides a unified algorithm for deblurring, noise removal, and high resolution reconstruction. Due to the recursive nature of the algorithm, computations are minimized. The algorithm is highly parallel since reconstruction is done independently for each pel in the transform domain. Computer simulations demonstrate the performance of the algorithm.>