Abstract: An image processing method which includes the acquisition (10) of an intensity image (J) formed by a matrix of points (A) having non-binary intensity values and representing filiform objects, and also includes steps for the detection of filiform objects, which steps include: automatic determination (20) of an intensity vector field ({right arrow over (V)}), forming a state image constituted by characteristics concerning magnitudes and angles [∥{right arrow over (V)}∥,&thgr;] of intensity vectors a the various points (A) of the intensity image, an operation (30) for chaining the points of an individual filiform object by means of automatic filtering [Q(i)] of the state image, thus performing a selection of a dense and continuous field of intensity vectors corresponding to points of the filiform object.

Abstract: A watermark detector is disclosed to judge whether multimedia content can be copied or not. The watermark detector examines the multimedia content and outputs a signal indicating whether a watermark is present or not. A decision variable indicating to which extent the watermark is present is determined, for example, the amount of correlation between the input signal and a reference copy of the watermark to be detected. The watermark is detected if the decision variable exceeds a predetermined threshold (y2). The detector also generates a random output signal for a predetermined range of decision values between the threshold (y2) and a further threshold (y1).

Abstract: This invention provides an input apparatus which can correctly recognize and accept hand-written characters as soon as they are input without using such an input device as an optical character reader or a keyboard. when characters, etc., are written using a writing instrument having a pen grip (22) fitted thereto, the pressure of fingers applied to the writing instrument is measured by strain gauges (28a, . . . , 28c). The strain gauges (28a, . . . , 28c) output the pressure change of the fingers as waveforms representing electrical changes to a CPU. The CPU (40) separates the waveform into unit waveforms each corresponding to one character, and analyzes wave characteristics for each unit waveform. The waveform characteristics of the written character are then compared and collated with the waveform characteristics of each character of a writer which are in advance learned and stored. Accordingly, the hand-written characters can be correctly recognized immediately after written. Since mercury switches (29a, .

Abstract: An automatic text inputting method and a system inputs text from multiple pages such as in a book by automatically turning pages, optically converting text image on each page into character data and determining an end of a specified unit of text. For example, the specified unit of text includes an article in a magazine and a chapter in a book. Additionally, in a selected group of text, a representative word is also automatically selected.

Abstract: A method of accurately locating a sub-pixel maximum on a two-dimensional grid having an x axis and a y axis is provided. The method, which is especially suitable for locating a sub-pixel maximum, which falls intermediate grid point locations, begins by identifying a grid point having a maximum grid point value. Then, a quadrangle containing the sub-pixel maximum is determined by locating neighboring grid points along the x and y axes, which have the next greatest grid point values. Next, four one-dimensional, fractional maxima are computed along each side of a quadrangle. Next, two lines connecting one-dimensional maxima along the opposite sides of the quadrangle are computed. Finally, a location for the sub-pixel maximum is calculated where the two computed lines intersect.

Abstract: A technique for compression and expansion of a function defined upon an M-dimensional manifold embedded in N-dimensional space uses a second generation wavelet transform and a modified zerotree bit-encoding scheme. Typically, a function is defined upon a two-dimensional manifold embedded in three-dimensional space, such as a sphere. A geometric base is chosen as a coarse initial model of the manifold. Second generation wavelets for the function are calculated using a triangular subdivision scheme in order to subdivide the geometric base in order to produce a refined triangular mesh. The wavelet coefficients are defined at the vertices of the triangles in the triangular mesh. A tree structure is created in which each node of the tree structure represents an associated triangle of the triangular mesh. Each triangle in the mesh is recursively subdivided into four subtriangles and each associated node in the tree structure also has four children, which correspond to the four subtriangles.

Abstract: A scanner that scans an object using a one-dimensional image sensor in order to capture a two-dimensional object is provided with a detachable remote control device. The scanner is controlled such that scanning will only be performed when the remote control device is detached from the scanner.

Abstract: A method and mechanism for displaying partial results of full context handwriting recognition. As handwritten characters are entered into a system, a shape matcher associates the character with a plurality of alternate code points, with each alternate code point having probability information associated therewith. The alternate code points are placed at the end of a queue, and a cost is determined from each alternate code point to any immediately preceding alternate in the queue. The cost is based on the probability information of the alternates and a transition cost therebetween. Then, the lowest cost path back from each of the alternates at the end of the queue to an alternate at the beginning of the queue is determined. If each lowest cost path back converges to a common alternate in the queue, the common alternate and any previous alternates on the path back are recognized as the code points for each of the handwritten characters associated therewith.

Abstract: A method and apparatus for extracting cluster shape features in two and three-dimensional images in a single scan is provided. The images, consisting of one or more classes, may be unbounded in one of their dimensions. As an image is scanned, each cluster fragment is assigned with one or more cluster labels. These labels are used to merge the cluster fragments into larger cluster fragments. An enhanced Hoshen-Kopelman algorithm is employed to determine the cluster shape features for the merged cluster fragments. Cluster label reuse is employed to enable the processing of substantially large images including unbounded images. After completing the scanning of an image section, cluster shape features data are outputted for completed clusters that extended into the section previous to said section. Optionally, cluster shape features of cluster fragments and completed clusters that extend into the said section can also be outputted.

Abstract: A system and method for video data processing for searching a region of overlapping candidate blocks to locate the best match to a comparably sized reference block comprising plurality of tree adder computational structures which each calculate the sum of absolute difference values for the corresponding data pixels in the comparison. The system further comprises a memory and addressing configuration comprising a dual part RAM, a funnel shifter, and control logic to supply the necessary data pixel values to allow the plurality of tree adders to compute the sum of absolute difference values for adjacent locations concurrently. Another aspect of the system and method for video processing is accumulating two values representing mathematical comparisons of two multi-dimensional images performed in a pixel-wise manner to determine which of the comparison produces less of a difference between the compared images.

Abstract: A method for automatically generating a knowledge database in an object classification system having a digital image data source, and a computer, includes the steps of inputting digital image data corresponding to a plurality of training images, and characterizing the digital image data according to pre-defined variables, or descriptors, to thereby provide a plurality of descriptor vectors corresponding to the training images. Predetermined classification codes are inputted for the plurality of training images, to thereby define object class clusters comprising descriptor vector points having the same classification codes in N-dimensional Euclidean space. The descriptor vectors, or points, are reduced using a similarity matrix indicating proximity in N-dimensional Euclidean space, to select those descriptors vectors, called extreme points, which lie on the boundary surface of their respective class cluster. The non-selected points interior to the class cluster are not included in the knowledge database.

Abstract: A method and apparatus for locating geometric shapes or edges thereof in data collection symbols initially samples and stores an image of light reflected from the symbol. Thereafter, two routines are performed. A first routine performs low level vision processing by identifying linked points along edges, lines, curves or within shapes. A rapid pixel linking subroutine identifies edge pixels in symbol images lacking distortion. If the edge or shape suffers from distortions, then one or more distortion compensating subroutines locate sequential linked pixels despite such distortions. The resulting coordinates of linked points, which represent lines, curves and other geometric shapes, are then employed by a second routine which identifies patterns within the identified lines/curves. Based on these identified patterns, types of symbols from various symbologies can be identified and located within the stored image.

Abstract: A reference image R is selected and a region of interest (ROI) is interactively selected encompassing a desired structure from a sequence of images of a moving structure. This ROI is cross-correlated with other real-time images by multiplication in the Fourier frequency domain, to determine if the desired structure is present in the image. If the structure is present, this image may be averaged with other images in which the structure is present to produce higher resolution adaptively averaged images. This invention is particularly useful in imaging coronary vessels. In an alternative embodiment, the offset of the desired structure may be calculated in a series of images. The images may then be sorted by this offset, and played back in that order to provide a "movie-like" display of the desired structure moving with the periodic motion.

Abstract: A fast position tolerant and economical fingerprint verification algorithm and software implementation is disclosed. The algorithm has very important advantages in comparison to existing fingerprint verification algorithms like minutia extraction or full image analysis based on a series of two dimensional fast Fourier transforms (FFT's) and two dimensional cross correlation in speed, compactness of the biocript and finger positioning for verification. It is based on a small "yardstick" pattern being moved inside the image to be verified, and when a match is found the rest of the verification process is started. The software code implements the algorithm in the most efficient way for utilization on a digital signal processor architecture. A main area of application is in fingerprint identification and verification, but it could be implemented with same success in other image verification systems like iris, eye, face etc., where two images should be compared.

Abstract: A data resolution synthesis algorithm takes low resolution source input data (e.g., data degraded from compression or data acquired from low quality imaging devices) to synthesize high resolution output data. In an exemplary embodiment for color printing, the algorithm is performed by characterizing a multi-pixel area, or window, around a pixel that can benefit from resolution enhancement. To interpolate to a high resolution output a set of spatial filters is applied to the data area based on the window characterization. The output of the resolution synthesizer is a set of multiple pixels for each input pixel, representing the source input pixel in a higher resolution enhanced version. The filters are chosen from a stored data base (generic or specifically applicable data base for each type input device) created to fit input/output device requirements.

Abstract: An improved method and mechanism for recognizing chirographs (handwritten characters) input into a computer system. A primary recognizer is provided for converting chirographs to code points, and secondary recognizers such as binary CART trees are developed and trained to differentiate chirographs which produce certain code points at the primary recognizer. Each such secondary recognizer is associated with each selected code point. When a chirograph is received, the chirograph is provided to the primary recognizer whereby a code point corresponding thereto is received. If the code point corresponds to one of the secondary recognizers, the chirograph is passed to the secondary recognizer, and a code point is returned from the secondary recognizer. If not, the code point provided by the primary recognizer is returned.

Abstract: A method is presented for identifying clutter in an image such as a sonar age. A detection scheme identifies portions of the image having a signal-to-noise ratio greater than a given threshold. A classification scheme is then applied to each such portion to generate fractal feature values associated therewith in order to classify the portion as a target or non-target. For clutter identification, each portion is assigned to a group based on a distance from the position of each portion to another of the portions that is its nearest neighbor. A series of tests are then performed for each group using the fractal feature values associated with each portion in each group. A failure of any of the series of tests by a group identifies each portion associated with that group as clutter in the image.

Abstract: An image processing coefficient determination method of an image processing system comprising the steps of:outputting test pattern image data having one or more image areas each containing a specific spatial frequency component by an image output device for preparing an output image,measuring a spatial frequency characteristic of the output image, anddetermining spatial frequency processing coefficients from a relationship between a spatial frequency characteristic of the test pattern image data and the measurement result of the spatial frequency characteristic of the output image.

Abstract: A method and apparatus for enhancing optical character recognition comprises a data processor and memory for maintaining an error detection and correction log. The data processor maintains a memory table of a plurality of rules for generating a rule base determined by recognition of a particular context type of an electronic bit-map portion. The appropriate rule base comprises rules and combinations of rules for application to bit-map portion data. A rule, a rule base or data may be selected and obtained from an internal or external memory. Upon application of the rule base, the error detection and correction log maintains a record of clear errors, corrected data, failed rules of the rule base and the original bit map. Possible errors are flagged and clear errors are automatically corrected provided a confidence level in the correction is reached or exceeded.

Abstract: A method for matching objects based on spatial layout of regions based on a shape similarity model for detecting similarity between general 2D objects. The method uses the shape similarity model to determine if two objects are similar by logical region generation in which logical regions are automatically derived from information in the objects to be matched, region correspondence, in which a correspondence is established between the regions on the objects, pose computation in which the individual transforms relating corresponding regions are recovered, and pose verification in which the extent of spatial similarity is measured by projecting one document onto the other using the computed pose parameters. The method of the invention can be carried out in a microprocessor-based system capable of being programmed to carry out the method of the invention.