Text compression
Elements of information theory
Elements of information theory
Efficient Context-Based Entropy Coding Lossy Wavelet Image Compression
DCC '97 Proceedings of the Conference on Data Compression
An Overhead Reduction Technique for Mega-State Compression Schemes
DCC '97 Proceedings of the Conference on Data Compression
Lossless Image Data Sequence Compression Using Optimal Context Quantization
DCC '01 Proceedings of the Data Compression Conference
Applications of universal context modeling to lossless compression of gray-scale images
IEEE Transactions on Image Processing
Lossless compression of continuous-tone images via context selection, quantization, and modeling
IEEE Transactions on Image Processing
High performance scalable image compression with EBCOT
IEEE Transactions on Image Processing
The LOCO-I lossless image compression algorithm: principles and standardization into JPEG-LS
IEEE Transactions on Image Processing
Context modeling based on context quantization with application in wavelet image coding
IEEE Transactions on Image Processing
Optimal context quantization in lossless compression of image data sequences
IEEE Transactions on Image Processing
Mutual information-based analysis of JPEG2000 contexts
IEEE Transactions on Image Processing
Context-based adaptive binary arithmetic coding in the H.264/AVC video compression standard
IEEE Transactions on Circuits and Systems for Video Technology
Motion compensation algorithm based on color orientation codes and covariance matching
ICIRA'10 Proceedings of the Third international conference on Intelligent robotics and applications - Volume Part II
Hi-index | 0.00 |
Context-based lossless coding suffers in many cases from the so-called context dilution problem, which arises when, in order to model high-order statistic dependencies among data, a large number of contexts is used. In this case the learning process cannot be fed with enough data, and so the probability estimation is not reliable. To avoid this problem, state-of-the-art algorithms for lossless image coding resort to context quantization (CQ) into a few conditioning states, whose statistics are easier to estimate in a reliable way. It has been early recognized that in order to achieve the best compression ratio, contexts have to be grouped according to a maximal mutual information criterion. This leads to quantization algorithms which are able to determine a local minimum of the coding cost in the general case, and even the global minimum in the case of binary-valued input. This paper surveys the CQ problem and provides a detailed analytical formulation of it, allowing to shed light on some details of the optimization process. As a consequence we find that state-of-the-art algorithms have a suboptimal step. The proposed approach allows a steeper path toward the cost function minimum. Moreover, some sufficient conditions are found that allow to find a globally optimal solution even when the input alphabet is not binary. Even though the paper mainly focuses on the theoretical aspects of CQ, a number of experiments to validate the proposed method have been performed (for the special case of segmentation map lossless coding), and encouraging results have been recorded.