Abstract: A data compression system which generates several types of symbols from bands of data derived from the digital image. Each group of data bits located in the same bit position across the data band is represented by a mode symbol, which signifies the decimal value of the binary combination of the data contained in that group. The mode symbol is followed by a run symbol, which represents the number of times or incremental positions that the same mode occurs successively until a different mode is generated. Mode and run symbols are transmitted in alternation, with each mode symbol followed by its corresponding run symbol, to an encoder, which generates variable length code words that are transmitted to a receiving station and decoded to provide symbols which are used to form a facsimile image. In one embodiment of this system, the mode and run symbols are combined to form a different type of symbol, known as a ternary octade symbol, to further improve the compression performance of the system.

Abstract: A method and apparatus for compressing digital data derived from an image using non-adaptive predictive techniques. A prediction table, which is pre-generated based on a number of sample images, generates a predicted pixel and source state for each pixel of the original image, as the original image is scanned. The predicted pixel is the expected value of the pixel when considering the values of a group of adjoining pixels, while the source state is indicative of the probability that the predicted pixel is in error. Prediction error pixels are then generated and grouped according to their respective source states to form a plurality of run length symbols with each symbol comprising a white portion and a black portion, which symbols are stored sequentially in order of formation in a memory device according to their respective source states.

Abstract: A digital image data compression system which generates a plurality of run length symbols, with each symbol comprising a white portion and a black portion. The white portion may contain up to a specified maximum number of white pixels, while the black portion may contain up to a second specified maximum number of black pixels. A symbol may be composed of a black portion followed by a white portion (a BW type run), or a white portion followed by a black portion (a WB type run). If the second portion of any run length of a particular type run exceeds the maximum number of pixels allowed for that color, that run length is terminated at the maximum number, the corresponding symbol is generated, and a special symbol, indicating a change in the type of run, is inserted. The symbol following the special symbol, which is the opposite type from the original symbol, is generated using the remaining pixels from the last run length and the following run length of the opposite color.