Abstract: An encoding and decoding system that includes a model implemented hardware or software that is capable of operating on arbitrary data (e.g., data of a different or variety of types.) The model generates a context and binary decision for each symbol, which a binary entropy coder that uses to generate a compressed bit stream. One context model is operable with arbitrary data and uses context bins of various orders to generate a context and a decision for each symbol in an input symbol stream. An entropy coder estimates probability and generates compressed data stream in response to contexts and decisions from the model and according to generated probability estimates.

Abstract: A computer system constructs a compression dictionary for compressing a character string by interrogating an initial substring portion to determine input string characteristics that are used to select one or more dictionary segments from a library of predetermined dictionary segments individually adapted for compressing strings with particular characteristics. The initial substring portion is dynamically determined during the interrogation. A first set of dictionary segments that meet predetermined automatic selection criteria are selected and a second set of candidate dictionary segments that meet second-level selection criteria are identified for a sampling phase. During the sampling phase, the candidate dictionary segments are alternately used to compress the initial substring portion and determine compression performance statistics.

Abstract: An efficient, fast-decoding, order-preserving, easily implementable, length-based (L-based) arithmetic coding method, apparatus, and manufacture for an m-ary alphabet {1, . . . , i, . . . , m} is provided. A coding method in accordance with the invention combines recursive division of intervals on a number line into sub-intervals whose lengths are proportional to symbol probability and which are ordered in lexical order with the constraint that probabilities be estimated as negative powers of two (1/2, 1/4, 1/8, etc.). As a consequence, the advantageous preservation of lexical order and computational efficiency are both realized. Also, a coding system in accordance with the invention is simple to implement, and high speed operation is achieved, because shifts take the place of multiplications. A coding apparatus in accordance with the invention preferably includes either a single decoding table to achieve fast decoding, or two decoding tables to achieve fast decoding as well as order preservation.

Abstract: The invention provides an improved method and apparatus for compression of palettized images. Input symbols in an M-ary alphabet are binarized based on a context model of the input data, where the binarization is selected to provide good compression by a binary encoder. The particular binarization is determined from a reindexing table which maps each input symbol to a number of binary values. The mapping is determined from the images to be compressed, and is typically transmitted with the compressed images as overhead. The mapping is a local minimum of the bitwise entropy of the binarization. With or without reindexing the input, the symbols can be converted compressed in parallel, with the bits of the input symbols buffered and reordered as necessary to ensure that bits needed for context of a bit being decoded are available before the decompressor decodes the bit being decoded.

Abstract: A method for data compression of records in storage that offers the decoding speed of Variable-to-Fixed codes without loss of sort order characteristics when stored in coded form. The methods include special treatment of the first ordered symbols in both source and decoder alphabets and do not depend on the use of prefix or arithmetic codes for their efficacy. The method saves database storage space by compressing actual records without loss of record sort order characteristics. The lexical ordering or "strong alphabetical" property of the encoding technique are left undisturbed by EOR codes and code-string terminations for byte-padding purposes because both EOR and code-string termination is accomplished with a zero-valued code word. This maintains record order through a proper magnitude relationship between the corresponding encoded records even if one record compresses to a different length than another.