Abstract: A method and apparatus for reducing power consumption and system bus load caused by a display controller in a unified memory system. A compression engine monitors a bus over which display data from the unified memory flows. The compression engine captures and compresses a copy of the display data corresponding to a display frame. The compressed data is stored in a small memory unit. Subsequent refreshes of the display are serviced by a decompression engine. The decompression engine decompresses the compressed data from the small memory unit and provides the decompressed data to the display controller.

Abstract: A method and an apparatus for compressing and decompressing data employ a splay coding instead of an arithmetic coding to achieve a high speed code entering process. The data compressing apparatus encoding input data according to a history of occurrence of the input data in the past to compress it, includes a pre-data retaining unit retaining a context, a history retaining unit retaining a combination of input data and a context, a code tree retaining unit retaining code trees each independent for each context, a code tree determining unit determining a code tree, a code outputting unit outputting unique data according to bifurcation from a node on the code tree, a code tree rearrangement unit exchanging an encoded leaf with another leaf or an internal node, and a pre-data updating unit entering the data in the pre-data retaining unit.

Abstract: In a digital coding apparatus or a digital coding and decoding apparatus for image data compression and expansion by means of Huffman coding, a Huffman coding circuit converts a combination of ZERO RUN and VALUE into a variable-length code. A code length calculation circuit has an AC code length table for prestoring variable-length codes and their code lengths in corresponding relationship. The code length calculation circuit inputs not a ZERORUN-VALUE combination but a variable-length code from the Huffman coding circuit, thereby calculating from the variable-length code inputted its code length according to the AC code length table. The present invention can provide a downsized AC code length table in comparison with a conventional one that stores ZERORUN-VALUE combinations and code lengths in corresponding relationship.

Abstract: The present invention is a method and apparatus for the efficient processing of binary images represented as a run length encoded data stream. The system employs a CPU and cache or other memory for efficiently scaling or scaling and rotating parsed RLE date. Subsequently, the scaled RLE data can be decoded to produce a scaled image bitmap.

Abstract: A bi-directional data pipeline for interfacing a memory with a communications port includes a series of four pipeline elements comprising two DMA buffers and first and second holding registers. A data word is transferred from memory to the DMA buffers, each holding one data byte of the data word. With each clock cycle, the data bytes are successively transferred through the two holding registers. Two comparators are used to determine if three successive identical data bytes are present in the pipeline. If three identical bytes are detected, run length encoding is enabled, and a run length count register is incremented for each successive identical byte received through the pipeline. The run length count and associated data byte are transferred to a FIFO for transmission over the data path. A tag associated with the run length count distinguishes the run length count from data bytes in the FIFO.

Abstract: A novel method of improving the motion estimation and coding efficiency of Interframes in a video compression system is disclosed. The method teaches a novel minimization function to be used to determine the optimum motion vector for a block within a picture. The minization function takes into account not just the conventional sum of the differences but also the quantization step, the motion vector itself and the predicted motion vector. The minimization function disclosed better balances the tradeoff between the size of the motion vectors and the size of the resulting residuals. The results are particularly useful when the range of the search is increased. Motion vectors closer to the prediction motion vector will be favored over better matches found further away from the prediction motion vector. In addition, a novel variable length coding (VLC) method is also disclosed.

Abstract: A video image compression and decompression system and method operates on frame blocks and uses multiple compression stages to select a best compression method for use with the associated block. The best compression method is selected by performing motion compensation compression on each block and assigning a quantitative image quality rating to the compressed block. During a second compression stage, the blocks having the lowest image quality ratings are selected and compressed using a vector quantization method which improves the quality of the block representation thereby improving the overall image quality. During each compression stage, compression dictionaries are built and dynamically updated. After the compression method is selected for each block, a compressed image frame is built.

Abstract: A method generates a reduced number of values representing a sequence of grouped data values and partitions the reduced number of values by first mapping data values into groups of symbols and then partitioning the resulting stream of symbols. The digits representing the first data value in each group are replaced with symbols from a first alphabet. The most significant digit of the second data value in each group and the sign of that second data value are also represented by a symbol from the first symbol set, while the remaining significant bits of the second data value in each group are represented by symbols from a second symbol set. A stream of symbols which represent a sequence of grouped data values is partitioned into first partition symbol groups and second partition symbol groups. Each first partition symbol group comprises the symbols representing the first data value in each group and also the symbol representing the least significant bit of the second data value which follows.

Abstract: A method and apparatus for consolidation of equivalence classes of scanned symbols as may be used for image data compression. The present invention performs run-length symbol extraction and classifies symbols based on both horizontal and vertical run length information. An equivalence class is represented by an exemplar. Feature-based classification criteria for matching an exemplar is defined by a corresponding exemplar template. The feature-based classification criteria all use quantities that can be readily computed from the run endpoints. Reducing the number of equivalence classes is achieved through a process called equivalence class consolidation. Equivalence class consolidation utilizes the symbol classifier to identify matched exemplars indicating equivalence classes which may be merged. For a consolidated equivalence class, the exemplar matching the most symbols is selected as the representative for the class.

Abstract: In order to estimate a character size of a subject copy in a facsimile system, the number of white runs is detected with respect to each of run length detected. Subsequently, an average white run length is determined. Thereafter, the character size is estimated by referring to data indicating a relationship between an average white run length and a character size.

Abstract: Data items associated with different locations in a spatial field of locations are written in a memory. The memory comprises a length memory in which for each location there is stored a length code which indicates the memory length required for the data item. For the reading of a data item associated with a location, the address of the data item is calculated by reading the length codes for the data items of preceding locations from the memory and by summing the lengths indicated.

Abstract: A method of image processing includes the steps of allocating image data of a document into a plurality of blocks of a predetermined pixel matrix; determining a gradient range exponent and a mean value information for each of the blocks based on the image data contained in each of the blocks; encoding the image data of each pixel of each of the blocks into code data based on the mean value information and gradient range exponent for the respective block so that the code data defines each pixel with fewer gradation levels than the image data; determining whether or not a mutually adjoining block is related to a solid image of a same density relative to a block subject to the encoding process; and executing a run length encoding process for a batch of mean value information, gradient range exponent, and code data based on a run length of adjoining blocks discriminated as related to a solid image of the same density.

Abstract: In a modified Huffman (MH) code decoding apparatus, an MH code is composed of a first part and second part. The apparatus includes a code conversion table for outputting a run data in response to an index data and a code length table. The index data is composed of a color data indicative of a color of a run corresponding to the current MH code, a first code length data associated with a length of the first part of the current MH code, and a second part data indicative of the second part of the current MH code. The run data is composed of the color data, a code type data indicative of whether the current MH code is a terminate code or a make-up code, and a run length data indicative of a length of the run.

Abstract: A DCT VLC decoding method substantially lessens the computational burden of picture dequantization. The method of decoding variable length code (VLC) data includes the step of receiving a bitstream sequence organized into a hierarchy of levels including, in order, a sequence layer, a group of pictures layer, a picture layer, a slice layer, a macroblock layer, and a block layer. In the sequence layer, the method defines a quantization matrix. In a macroblock layer, the method defines a quantizer scale parameter in the macroblock layer, multiplies the quantization matrix times the quantizer scale parameter and stores the product in a quantization multiplication factor. In the block layer, the method parses an encoded data bitstream to detect a VLC code in the block layer and dequantizes the VLC code by multiplying the VLC code times the quantization multiplication factor.

Abstract: The present invention relates to a bit-map compressing method and a bit-map compressing device. The bit-map data compressing method includes a main scanning step for scanning bit-map data in a main scanning direction; a first region separating step for separating a region including black pixels and a region including no black pixels in line unit; a secondary scanning step for scanning in a secondary direction perpendicular to the main scanning direction; a second region separating step for separating a region including black pixels and a region including no black pixels, in column unit; and an encoding step for encoding as an element each region obtained in the first region separating step and the second region separating step. Bit-map can be transferred by effectively performing data compression on simple algorithm, without using special hardware.

Abstract: A method and apparatus for classification of scanned symbol into equivalence classes as may be used for image data compression. The present invention performs run-length symbol extraction and classifies symbols based on both horizontal and vertical run length information. An equivalence class is represented by an exemplar. Feature-based classification criteria for matching an exemplar is defined by a corresponding exemplar template. The feature-based classification criteria all use quantities that can be readily computed from the run endpoints. Reducing the number of equivalence classes is achieved through a process called equivalence class consolidation. Equivalence class consolidation utilizes the symbol classifier to identify matched exemplars indicating equivalence classes which may be merged. For a consolidated equivalence class, the exemplar matching the most symbols is selected as the representative for the class.

Abstract: A zero-run-length encoder for a JPEG compression system comprises an addressable memory for storing 63 input values (quantized AC DCT coefficients), zero-detection logic, a shift register, a value generator, an accumulator, a Huffman encoder, done-detection logic, and last-value-detection logic. For each input value, the zero-detection logic stores zero/nonzero indications in a respective bit position of the shift register. The value generator includes a leading-zero counter that determines the number of leading zeroes in the leading fifteen bit positions of the shift register. This count is used to determine an offset value which is added to a previous address value (initially zero) to yield a present address value. The present address value is used to select a memory location from which an input value is read from memory into the Huffman encoder. The Huffman encoder generates an output code as a function of the addressed input value and the leading zero count.

Abstract: A data encoding system for encoding input color pixel data and outputting encoded data. The data encoding system includes a reference pixel generating device which outputs reference pixel data corresponding to the input color pixel data to be encoded; a predictor having a color order table which sets the color ranks of color codes for every reference pixel pattern, reads and outputs the color rank of the corresponding color code from the color order table on the basis of the color pixel data to be encoded and its reference pixel data; and an entropy encoding device which converts the color ranking data into encoded data and outputs the encoded data.

Abstract: An original image is encoded using difference signals that correspond to the difference between the original image and a blur image that approximates the original image. The blur image is generated using DC signals, wherein each DC signal corresponds to a different block of the original image.

Abstract: A zig-zag and alternate scan conversion circuit for encoding/decoding videos which is structured only by logic circuits using a regularity of scan address patterns, thereby achieving an improvement in processing speed and a reduction in area.

Abstract: A method is provided for identifying, correcting, modifying and reporting imperfections and features in pixel images that prevent or hinder proper OCR (Optical Character Recognition) and other document imaging processes. One embodiment of this invention provides that run length compressed images can be analyzed and corrected directly for improved performance. Major steps included in this invention for the enhancement of images for OCR and document imaging are: The detection of undesired printed matter and the deletion of undesired printed matter. Detection of undesired printed matter includes successive generation of entries into an array listing of undesired printed matter to be deleted. Deletion of undesired printed matter follows in a separate step for all types of undesired printed matter.

Abstract: The invention implements a run-length file format with improved space-sav qualities. The file starts with a header in ASCII format and includes information such as the file format, maximum X and Y values, and a color palette. The binary raster data include at least one byte for each pixel run to indicate the color selected from the color palette and the run length. Concatenation flags demarcate groups of bytes. According to a first embodiment, each run is indicated by a variable-length sequence of bytes with the color in a fixed-length field and the repeat number in a variable-length field. According to a second embodiment, the color and run length are given in separate byte strings, each having a variable length. The file also includes line offset pointers to permit navigation of the file without decompressing it and thus to facilitate operations such as panning and zooming.

Abstract: Image data, read by an image reading device and binarized, is stored in a buffer memory by one scanning line. Then a compressive coding on the stored image data by one scanning line and a compressive coding on image data developed form codes, such as codes written in a page description language (PDL), or image data read from a predetermined memory are performed by a single encoder.

Abstract: An apparatus for processing an input image, represented by an input bit stream including a plurality of first bit sets, is provided. The apparatus includes a device for processing the input bit stream in such a manner that each of the first bit sets is represented by a second bit set with each of the second bit sets having less bits than each of the first bit sets and a selected number of bits in the input bit stream, corresponding to image-related information, being discarded as a result of the encoding. The apparatus is provided with an electronic volatile memory for storing a resulting encoded bit stream and a bit stream analyzer for analyzing the stored encoded bit stream to generate positional signals. The apparatus is further provided with a reconstruction circuit for reconstructing the stored encoded bit stream, by reference to the positional signals.

Abstract: An image processing apparatus for compressing and decompressing image data. The apparatus includes: a composition circuit for composing composed data, having a predetermined bit width, from the given image data which correspond to spatially continuous pixels of an image; a compression circuit for compressing the composed data with Lempel-Ziv algorithm so as to obtain compressed data; a memory for storing the compressed data therein; and a controller for controlling the composition circuit and the compression circuit.

Abstract: A data compression system separates input data into color planes prior to compression. If needed for legacy game consoles, compression is performed by a game cartridge transparent to the console. To minimize the number of passes required of a coder/decoder, color planes are ordered by density and the densest color plane is coded first. After the first color plane is coded, other color planes are coded, but pixels which are known to have colors from previously coded color planes are not coded. The last color plane is not coded, but is deduced from all the other color planes. Alternatively, pixel color values are represented by vectors with components thereof separately coded by subcolor planes. Also, each color plane can be coded until a threshold number of pixels are coded, and the remaining pixels coded by bit plane. The image data could be coded using pixel position information as context.

Abstract: A method of transferring the digital representation of a bit-mapped image which includes the steps of: (1) identifying predefined patterns in the bit-mapped image data, and (2) transmitting pattern codes corresponding to those patterns to a receiving device. The identification of the patterns involves the partitioning of the bit-mapped image into a series of pattern cells and the scanning of those pattern cells to find preselected patterns. The preferred method of searching the pattern cells is based upon the use of a hashing function and hash tables. In response to the received pattern codes, the receiving device reconstructs the bit-mapped image associated with the pattern codes. After all patterns in the bit-mapped images are identified and sent to the receiving device, the non-recognized portions of the bit-mapped image are then sent. The overall bit-mapped image is recreated by combining the recognized patterns with the non-recognized portions in their original order.

Abstract: Compression of a plurality of signal samples y(t.sub.0),y(t.sub.1),y(t.sub.2) . . . y(t.sub.z) may be accomplished by performing a first linear regression to obtain a curve fitted thereto which is represented by a polynomial having a first set of coefficients (a.sub.0,a.sub.1,a.sub.2,a.sub.3 . . . a.sub.n) which is stored in memory (18) or transmitted by a transmitter (20). The error between each of the actual signal samples (y(t.sub.0),y(t.sub.1),y(t.sub.2) . . . y(t.sub.z) and the value approximated by the curve fitted to the values is computed and then thresholded. A second regression analysis is then performed to fit a curve, having a second set of coefficients (b.sub.0,b.sub.1,b.sub.2,b.sub.3 . . . b.sub.q) to the thresholded error values. The second set of coefficients is also stored or transmitted and used, in conjunction with the first set of coefficients, to closely approximate the original set of signal samples.

Abstract: The process for enhancing images of scanned documents identifies a variety of items in the scanned document which could make optical character recognition and other document image processing difficult or impossible. These items include identifying skew, registration, speck, lines intersecting printed matter, reverse printing, shaded printed matter. The process allows to register such items, take steps to correct skew and image registration, as well as reverse invert printing. Items to be removed are listed for deletion in one step. The image of the scanned document is stored and processed as a run-lngth coded image, except for a some operations in which parts of the run length coded image are converted to pixel image code for particular substeps of the process. The result of such a substep, e.g. a modified image, is then re-converted to run-length code.

Abstract: A method and system including a high speed computer algorithm for rotating images stored in a computer memory, such as those captured by a CCD array. The method and system may be used with an over-the-belt optical character recognition (OCR) reader that captures a pixelized image of the surface of a parcel as it moves along a conveyor. A run-length encoded output image is obtained from a run-length encoded input image, wherein the output image is a rotated version of the input image. A predefined pixel mapping sequence for a rotation angle with a rational tangent is used to map image pixel data. Mapping image pixel data using a predefined pixel mapping sequence allows rotation without floating point computations, and allows the use of a reverse pixel mapping technique to map output runs from an input image in order to avoid the creation of holes or artifacts in the output image. Mapped output runs are scaled to preserve the aspect ratio of the rotated image.

Abstract: A coding of images and originals, for example, for telefax and color television, allows transmission time to be shortened, in particular in the case of black and white digital or numeric longitudinal coding. For that purpose, the same code words for black and white are provided also for different numbers or digits. As discriminating criterion, the input sequence of black and white is used. During coding of several successive white lines, transmission time can be further shortened in that the coded number of white lines is provided before or after the code word for the white line. During gray coding, transmission time is in particular reduced by subdividing the gray scales or binary code words, if necessary with deliberate redundancy, since in this case many characters occur successively and are then transmitted in the same way as the white lines.

Abstract: An encoding system uses a modified 8/9 rate modulation code to encode 8-bit data symbols into 9-bit cells in a conventional manner in accordance with the modified code and 9-bit ECC symbols into 10-bit cells by (i) encoding 8 bits of the symbol into a 9-bit cell in accordance with the modified code, and (ii) inserting into the 9-bit cell the remaining, that is, the non-encoded, bit of the ECC symbol. The system reproduces the 8-bit data symbols by decoding the 9-bit cells in a conventional manner in accordance with the modified code, and the 9-bit ECC symbols by (i) removing from the associated 10-bit cell the bit inserted during encoding, (ii) decoding the remaining 9 bits to reproduce 8 bits of the symbol, and (iii) inserting into the 8 bits the bit that was earlier removed. In an exemplary embodiment, the 8 least significant bits of the ECC symbol are encoded using the modified 8/9 rate code. The 9 bits produced by the code are used essentially as the first "c" bits and last "10-c" bits of a 10-bit cell.

Abstract: A method and apparatus for characterizing the performance of either a printer used to print a two-dimensional symbol or an imaging system used to generate an gray-scale image of the printed symbol. After the image is generated, an acquisition target of the symbol is located. The quality of the acquisition target is characterized by thresholding the gray-scale image, run-length encoding the resulting binary image, and computing the correlation coefficient between the run-length-encoded image and a run-length-encoded representation of a defined acquisition target.

Abstract: Regions of a video frame are classified into different classes of regions (e.g., inter encoded blocks, intra encoded blocks, and skip blocks). The classes of regions are then run-length encoded using an encoding sequence that is a function of the content of the video frame. The run-length encoded signals are used to generate an encoded bit stream for the video frame. In preferred embodiments, the run-length encoding sequence either is selected from a set of possible sequences or is generated based on a selected location within the video frame. The selected sequence or location is identified explicitly in the encoded bit stream and used by a decoder to select/generate the run-length decoding sequence used to decode the run-length encoded signals in the encoded bit stream that correspond to the classes of regions in the video frame.

Abstract: In an image compressing apparatus of this invention, a table of extended MH codes is prepared with respect to combinations of consecutive black/white pixels. When a portion where frequent changes between black and white occur is to be coded, consecutive black/white pixels are coded at once without measuring the run lengths of pixels. These extended MH codes are obtained by combining MH codes corresponding to black and white runs whose run lengths are determined in one byte of an original image.

Abstract: This invention provides a method and an apparatus for synthesizing two image signals, at least one of which is run-length coded, at a high speed.

Abstract: An image data coding apparatus and method to achieve a scale-down of a circuit without sacrificing compression efficiency. Input image data is delayed in a delay circuit and a coincidence detector compares the image data and delay image data to output a coincident code when these two are the same. An adaptive run length counter adaptively switches to select either a coincident code run length of a continuous length of the coincident codes calculated at an output time of the coincident codes or an image data run length of a continuous length of the same codes of the image data calculated at a non-output time of the coincident codes and outputs an adaptive run length. A Huffman coder carries out a Huffman coding of the adaptive run length.

Abstract: The decompression side of the system has apparatus for forming a sequence of arrays of pixel display values for driving a display device from a stream of compressed data containing compressed data codes. Each of the arrays of pixel display values represents a frame of a moving image. Pixel copy codes within the stream of compressed data trigger copying of one or more previously formed pixel display values to a current position within a sequence of positions within the array along which decompression progresses. New pixel codes within the stream of compressed data trigger mapping of one or more pixel appearance values specified by the new pixel codes to respective pixel display values. The pixel display values derived from the new pixel codes are then written to a current position within the sequence. The compression side provides a complementary mechanism for generating such pixel copy codes and new pixel codes from source image data.

Abstract: A pixel data encoder for use in a facsimile apparatus features a first in first out memory for storing one-dimensional binary pixel data sequences in synchronism with a data acquisition clock; a run length determiner coupled to receive the one-dimensional binary pixel data sequences stored in the first in first out memory, the run length determiner calculating a run length for each of the one-dimensinal binary pixel data sequences received; and an encoding section which issues a one-bit address signal for indicating raw data will be issued and N-bit raw data in the event that the run length is less than a predetermined run length N, the encoding section issuing an address code and a remainder code in the event that the run length exceeds N, the address code consisting of consecutive binary level "1"s with a bit length (M-2) and a binary level "0", and the remainder code consisting of a binary sequence representing the run length wherein the most significant bit of the binary sequence is omitted, M representin

Abstract: In a character recognition system, a method and apparatus for segmenting a document image into areas containing text and non-text. Document segmentation in the present invention is comprised generally of the steps of: providing a bit-mapped representation of the document image, extracting run lengths for each scanline from the bit-mapped representation of the document image; constructing rectangles from the run lengths; initially classifying each of the rectangles as either text or non-text; correcting for the skew in the rectangles; merging associated text into one or more text blocks; and logically ordering the text blocks.