Abstract: A method and apparatus for storing and retrieving images of documents, e.g. checks.

Abstract: A pipelined parallel processor (PPP) integrated onto a single integrated circuit. The PPP includes a filtering unit, internal routing circuitry such as a crosspoint switch, an internal frame store controller, and one or more function circuits. The function circuits may be, for example, arithmetic units, lookup tables, timing compensators, adders/subtractors, statistics modules, image shifting circuitry, and other useful processing devices. The components of the PPP are interconnected with the crosspoint switch which routes data between the frame store controller, the filtering unit, function circuits, external input channels, and external output channels.

Abstract: An image processing system and methods of operation within such a system are disclosed. The image processing system utilizes a class model defined by one or more relative relationships between a plurality of image patches. The relative relationships describe the overall organization of images within an image class. The relative relationships are encoded in a global deformable template which can be used to classify or detect images. The class model may be predefined or generated by an image processing system.

Abstract: An image data encoding apparatus encodes image data at a high speed while minimizing the deterioration of the image quality on a reproduced screen. The apparatus includes an encoding circuit, a code quantity counter and a control circuit. The control circuit compares a total code quantity with a reference code quantity. When the total code quantity exceeds the reference code quantity, the control circuit further compares an excessive quantity with a brought forward code quantity. The control circuit instructs an interruption of generation of the code image data by the encoding circuit when the excessive quantity is larger than the brought forward code quantity.

Abstract: The present invention provides a marker detection and thinning process for a photocopier which is executed during a single scan of a document and does not require a prescan of the document. The system realtime analyzes pixel image data, which is stored in a memory during the scanning process, to detect a marker line of a predetermined color and to thin the marker line. This analysis begins with image data stored in the memory that corresponds to a first pixel. The first pixel is associated with the scanline being presently scanned. It is determined if the image data of the first pixel represents a color that is equivalent to the predetermined color. Marker data is then stored in a buffer at a location corresponding to the first pixel if it is determined that the image data of the first pixel represents a color that is equivalent to a predetermined marker color. The data, stored in the buffer at a second pixel location, is analyzed.

Abstract: A method for mosaicing medical images, particularly volumes acquired by a 3D ultrasound device. The method relies on an image-based technique that transforms a target volume to align with a reference volume. The transformation is composed of a coarse rigid transformation and a fine elastic warp. The fine transformation assumes that the two volumes are close enough for the mapping between them to be expressed in terms of their gradient values.

Abstract: Image data decompression apparatus and method for receiving data words comprising a first sub-word and a second sub-word, at least some possible values of the first sub-word specifying a run of at least m instances of a predetermined data value, corresponding second sub-words specifying a number of instances of the predetermined data value in a range from at least m instances. The apparatus includes a first decoder for decoding the first sub-words to detect whether a current first sub-word specifies a run of at least m instances of the predetermined value, and a second decoder for decoding the second sub-words to generate output decoded data.

Abstract: A method of locating handwritten words in handwritten text images under a variety of transformations including changes in document orientation, skew, noise, and changes in handwriting style of a single author which avoids a detailed search of the image for locating every word by pre-computing relevant information in a hash table and indexing the table for word localization. Both the hash table construction and indexing can be done as fast operations taking time quadratic in the number of basis points. Generally, the method involves four stages: (1) Pre-processing where features for word localization are extracted; (2) Image hash table construction; (3) Indexing where query word features are used to look up hash table for candidate locations; and (4) Verification, where the query word is projected and registered with the underlying word at the candidate locations. The method has applications in digital libraries, handwriting tokenization, document management and OCR systems.

Abstract: An object of the present invention is to construct a sensing system capable of extracting change information as new time series information from a continuous moving picture in a non-contact manner and remotely without requiring a special illuminating and irradiating device and driving the succeeding processing system as required, to clearly show a phenomenon in an arbitrary frequency band existing in the moving picture.

Abstract: A digitized image is divided into blocks and a wavelet transform is performed, producing wavelet blocks which are quantized, scanned into linear sequences, and then coded. The scanning of each wavelet block starts with a purely low-frequency element. The high-frequency elements are scanned in a sequence that depends on the occurrence of non-zero elements among the high-frequency elements. When a non-zero element is scanned, this information is used to proceed quickly to related high-frequency elements. This scanning method tends to produce a linear sequence in which non-zero elements are clustered together. Such sequences can be coded efficiently.

Abstract: A method for halftoning a digital image having an x,y array of pixels each having an associated input pixel value, includes the steps of providing a memory containing a dither bitmap for each input pixel value of the digital image where the dither bitmaps are partially correlated; selecting a dither bitmap corresponding to the input pixel value for a pixel in said digital image; modularly addressing the selected dither bitmap with the location of said pixel to obtain an output halftone image value; and repeating steps for each pixel in the digital image.

Abstract: A special method recognizes patterns in measurement signals. Speech signals or signals emitted by character recognition apparatuses are thereby meant. For the execution of the invention, the hidden Markov models with which the patterns to be recognized are modeled are expanded by a special state that comprises no emission probability and transition probability. In this way, the temporal position of the sought pattern becomes completely irrelevant for its probability of production. Furthermore, the method offers the advantage that new and unexpected disturbances can also be absorbed without the model's having to be trained on them. In contrast to standard methods, no training on background models need be carried out. However, this means a higher expense during the recognition of the patterns, since the individual paths of the Viterbi algorithm have to be normed to the current accumulated probabilities in the path with respect to their probabilities, in order to be able to compare them.

Abstract: In order to divide a dot pattern of touching characters having a thickness into patterns of each character without deformation, it is analyzed in the invention, as an ensemble of run continuations non-branched, continuations of neighboring `run`s, a succession of black dots ranged in a column being defined as a run. Some peculiar `run`s are nominated as segmentation candidates for dividing the dot pattern thereby, and the dot pattern of touching characters is divided by most appropriate one of the segmentation candidates.

Abstract: An image reading and recording apparatus includes a reading section for optically reading an image for output of image signals, and a recording section for generating and controlling a light beam to record the read image on the basis of the image signals. The apparatus further includes a light sensor for detecting the light beam at a predetermined position and for generating a record timing signal which determines start timing for control of the optical beam, a counter for determining lapse of a predetermined time since the generation of the record timing signal, and a read timing signal generator for generating, upon the lapse of the predetermined time, a read timing signal which determines timing for output of the image signals from the reading section.

Abstract: A value of the degree of centralization of gradient vectors, which has been calculated from an image signal, is compared with a threshold value, and a region of a prospective abnormal pattern is thereby determined. A picture element corresponding to a position, at which the center of gravity on the region of the prospective abnormal pattern is located, is taken as the picture element of interest, and a picture element corresponding to an end point that is associated with a mean value of index values for each radial direction line in iris filter processing, which mean value takes the maximum value, is thereby specified as a marginal point of the region of the prospective abnormal pattern in the direction along which the radial direction line extends. The thus set marginal points are connected by predetermined lines, and the region surrounded by the connecting lines is extracted as the prospective abnormal pattern.

Abstract: A method for compressing and decompressing multidimensional data based on approximating the data space by a smoothed particle cluster representation. The smoothed particle cluster representation is obtained by simulation for relaxation motion of a simulated particle corresponding to a sampling point to obtain a statistical average arrangement of the simulated particles on the basis of an equation of motion based on a potential theory given by an assumption that the simulated particles are moved upon receiving a repulsive force between the simulated particles, a velocity resistance, and a force proportional to a functional potential gradient having a large negative value in a region having a large original data value. The coordinates and an interval of the sampling points are obtained in accordance with an arrangement state of the simulated particles. The volume associated with each sampling point for obtaining the associated data value is the volume of the simulated particle.

Abstract: An image processing method and an image processing apparatus in which a DCT section transforms image data to a frequency conversion coefficient using a discrete cosine transform coefficient, a quantizing section quantizes the transformed frequency conversion coefficient using a quantization coefficient indicating weighing in quantization, a Huffman encoding section encodes the quantized data, and a CPU detects a code rate for encoded data by a prespecified time interval during execution of processing for encoding the data quantized by the encoding means and changes the quantization coefficient during processing for encoding according to the detected code rate.

Abstract: A multifunction peripheral which can perform copying, printing, faxing and scanning. The MFP is made up of a hard output unit, essentially a copier, and a control unit, essentially a print server. User interface definitions for copying are embedded in the hard output unit. User interface definitions for printing, faxing and scanning are stored in the control unit. When the hard output unit resets, it loads the UI definitions from the control unit. The UI definitions in the control unit are stored as a file and are easily updated from a LAN.

Abstract: An image signal compressing device having a spatial frequency data amount setting unit and a quantization table generation unit. A set data amount is set in the spatial frequency data amount setting unit for each of a plurality of spatial frequencies based on a preset total data amount and a discrete cosine transformation (DCT) data statistical amount corresponding to quantized DCT coefficients using a quantization table in which all quantization coefficients are "1". An amount of encoded data obtained using a predetermined quantization coefficient is estimated in the quantization table generation unit. The predetermined quantization coefficient is adopted as a resultant quantization coefficient of the quantization table when the estimated data is not greater than the set data amount set in the spatial frequency data amount setting unit.

Abstract: Video encoding and decoding processes provide compression and decompression of digitized video signals representing display motion in video sequences of multiple image frames. The encoder process utilizes object- or feature-based video compression to improve the accuracy and versatility of encoding interframe motion and intraframe image features. Video information is compressed relative to objects or features of arbitrary configurations, rather than fixed, regular arrays of pixels as in conventional video compression methods. This reduces the error components and thereby improves the compression efficiency and accuracy. The decoder process decompresses the encoded video information to reconstruct the objects or features of arbitrary configurations.