New algorithms are proposed for designing optimal binary vector quantizers. These algorithms aim to avoid the problem of the generalized Lloyd method of easily getting trapped into a poor local minimum. To improve the subjective quality of vector-quantized binary images, a constrained optimal binary VQ framework is proposed. Within this framework, the optimal VQ design can be done via an interesting use of linear codes.

Recursive rectilinear tessellations like quadtree are widely used in image coding, but a regular tessellation, despite simple geometry, may not suit image compression because it is too rigid to reflect the scene structure of an image. The paper presents a new image pyramid formed by adaptive, tree-structured segmentation to be a framework of a predictive multiresolution image coder. Subjectively appealing compression results are obtained at different resolutions by scene-adaptive, tree-structured segmentation and by exploiting the statistical dependency between the layers of the image pyramid. The adaptive segmentation-based image coder is constructed by recursive, least-squares piecewise functional approximation. The seemingly expensive encoding process can be made efficient by an incremental least-squares computation technique. The decoding is simple and can be done in real time if assisted by existing hardware technology.

A practical, fast, bintree-structured variable size block image coder is proposed. The inability of current segmentation-based image coding methods using piecewise functional approximation to represent textures is overcome by a generalized BTC encoding of approximation errors in texture areas. This new texture encoding technique significantly enhances subjective image quality at small extra costs to compression ratio and algorithm complexity.>

The authors present, as an image pyramid, an adaptive, tree-structured segmentation to facilitate progressive transmission. The new image pyramid is semantically more powerful than regular tessellations while syntactically simpler than free segmentation. This good compromise between adaptability and complexity is a key to the high compression ratios achieved by the proposed image coder. The performance is further enhanced by exploiting the statistical dependency between layers of the image pyramid.>

A sequence of medical images is converted to a 4-dimensional volume of bits using gray code. Then an interframe sequential and intraframe interlacing prediction scheme is used for reversible image sequence compression. Higher compression ratios than the current intraframe compression methods are achieved due to interframe decorrelation.>

A fast range search algorithm for visualizing extrema of d-dimensional volume data in real time as the user interactively moves the query range is presented. The algorithm is based on an efficient data structure, called index heap, which needs only O(N/log N) space and O(d2/sup d/N) preprocessing time to be set up, where N is the size of the d-dimensional data volume. The algorithm can answer an extremum query in O(4/sup d/) expected time, and its worst-case time complexity is O(2/sup d/ log N) per query. For dimensions two and three, the range search for extrema is effected in average O(1) time per query independently of the size of query range. Unlike previous range query algorithms in the computational geometry literature, the proposed algorithm is very simple and can be easily implemented.>