G.S. Desoli

A spline-like scheme for least-squares bilinear interpolations of images

A novel image interpolation algorithm which uses a least-squares approach by means of a spline-like scheme is presented. Such an interpolation can be utilized for image-data compression both in a stand-alone way and as a boundary response in a two-source decomposition scheme. Conventional linear-interpolation approaches are described first. Then, a least-squares bilinear interpolation is proposed. The major advantages of this type of interpolation for image coding are the following: reduction in the tile effect by means of the continuity constraints imposed on the reconstruction and through the noncorrelation of the residual; lower computation complexity: more effective exploitation of interblock correlation compared with traditional block-coding methods; and robustness against noise compared with standard interpolation algorithms. Experimental results show that the efficiency of the proposed algorithm is high, and that higher reconstruction quality can be obtained by an appropriate encoding of the residuals.<<ETX>>

Adaptive DCT for image‐data compression

This paper deals with the development of a coding system for image data compression based on the Discrete Cosine Transform (DCT). The aim is to integrate several methods into a single approach in order to obtain an adaptive coder characterized by higher performances. In particular, a variable blocksize scheme is used, which is based on quadtree segmentation driven by edge information. To improve data compression, DCT coefficients are then coded by a vector quantizer. Experiments have shown that the values of compression factors can be notably increased, while maintaining a good reconstruction quality.

Adaptive Image Sampling and Interpolation for Data Compression

Abstract A new interpolation algorithm for 2D data is presented that is based on the least-squares optimization of a spline-like scheme. This interpolation technique is then integrated into a two-source decomposition scheme for image-data compression. First, an optimum least-squares bilinear interpolation is implemented and applied to a uniform image sampling. Second, the problem of spatial adaptivity is addressed in order to fully exploit the compression capability of the technique. The solution lies in a nonuniform sampling strategy based on a recursive splitting algorithm. Experimental results show that the proposed image-interpolation algorithm is very efficient. The major advantages of this scheme over traditional block-coding methods are the absence of the tiling effect (thanks to the continuity constraints imposed on the reconstruction process) and a more effective exploitation of interblock correlation.

An edge-based splitting criterion for adaptive transform coding

A novel splitting strategy for variable block-size image coding based on edge information is presented, which makes it possible to reduce the number of blocks to be encoded while maintaining an in-depth split in edge areas. This approach has been applied to a standard transform coding algorithm, and results have been compared with those obtained by a classical splitting structure. Tests showed a notable increase in the compression factor under the same SNR conditions. Reconstructed images are affected by the blocking effect to a smaller extent, as the problem of generating a large number of small blocks in the uniform areas surrounding an edge is overcome by the proposed split topology.<<ETX>>

Adaptive 3D interpolation and two-source video coding

A new strategy for video coding at very low bit rates is proposed in the paper, where an image sequence is encoded in a two-source way. The main difference from state-of-the-art approaches lies in the use of a 3D spatio-temporal interpolation, based on a non-uniform grid (frames are processed in overlapped blocks), in order to allocate the information in an adaptive way. Experimental results show a good fidelity of the decoded sequences, without blocking effect.<<ETX>>

High-compression video coding: a novel approach

A novel approach to video coding at very low bitrates is presented, which differs significantly from most of previous approaches, as it uses a spline-like interpolation scheme in a spatiotemporal domain. This operator is applied to a non-uniform 3D grid (built on sets of consecutive frames) so as to allocate the information adaptively. The proposed method allows a full exploitation of intra/inter-frame correlations and a good objective and visual quality of the reconstructed sequence. Keywords: image processing; trilinear interpolation; 3D splitting; splines; adaptive approximation; image data compression; video coding.

Data Compression for Image Archiving Applications

A novel approach to still-picture. data compression is presented that is based on the inte- gration of two different coding techniques. Its aim is to obtain hightompression factors by exploi- ting the characteristics of a second-generation method (i.e., polynomial approximation) in suitable scene areas, while achieving higher fidelity, where required by the local image characteristics, th- rough a first-generation technique (i.e., vector quantization). Such a region-driven hybrid approach performs the integration by utilizing a control image (re- sulting from the detection OF areas characterized by high spatial frequencies) that drives the seg- mentation and approximation processes. The results obtained on a variety of natural color images are quite satisfactory. The visual quality of the reconstructed images is, in general, very good, and SNR is kept within an acceptable range. Moreover, the bit rate obtained is always below 0.15 bit/pixel (bitlpixel), with peak values of 0.07 bitlpixel.

A novel tree-structured video coder

A novel approach to video coding at very low bit rates is presented, which differs significantly from most of previous approaches, as it uses a spline-like interpolation scheme in a spatiotemporal domain. This operator is applied to a non-uniform 3D grid (built on sets of consecutive frames) so as to allocate the information adaptively. The proposed method allows a full exploitation of intra/inter-frame correlations and a good objective and visual quality of the reconstructed sequences.<<ETX>>

A quadtree schema for vector quantization of image sequences

An approach to image sequence coding based on the classical vector quantization is presented. The novelty lies in the use of a quaternary tree structure to transmit the position information about moved blocks, thus exploiting the temporal interframe correlations. Moreover, an error-free predictor is used in order to remove residual intra- and interframe redundancies. Results obtained by this coder are satisfactory and allow the transmission of an image sequence over a 64-kb/s channel.<<ETX>>