Abstract: Segmentation of a character-containing image, made by placing a dark mark against a light background, is accomplished by establishing a vertical pixel projection for each pixel column in the image. For more exact segmentation, the image is horizontally scanned in each row of the image after detecting edges of characters. The scanned results and thresholds are compared to verify whether edges and spaces of the scanned row are noise or discontinuous edges. The thresholds correlate with neighboring row data of the image. According to the results of the comparison, the discontinuous edges are connected and the noise is eliminated. Thereby, the character-containing image can be segmented into each character without segmentation errors.

Abstract: An image processing method, image processing device and recording medium are provided in which the effects of erroneous extraction of a region assumed to correspond to a main portion within an image are mitigated. On the basis of image data of an image which is an object of processing, plural face candidate regions, which are assumed to correspond to faces of persons, are extracted. A degree of overlapping is computed for a pair of overlapping face candidate regions. A weight point is set for each face candidate region such that a weight point of a face candidate region which overlaps with another face candidate region is higher than a weight point of a face candidate region which does not overlap with another face candidate region, and the higher the degree of overlapping with another face candidate region, the higher the weight point.

Abstract: A method for characterizing an image where a number of test areas of predefined shape and size are located on the image. The color or the texture of the image over each of the test areas is quantified. The image can be characterized by statistical descriptions of the frequency distribution of color or texture of the test areas.

Abstract: A video editing system which resizes a video stream in a transform domain is disclosed. The video editing system partial decodes a compressed video stream in order to obtain a transform domain representation of the video stream. The transform domain representation of the video stream includes both frame encoded blocks and field encoded blocks. The field encoded blocks result from reordering of pels prior to transform domain encoding the pels. The frame encoded blocks, however, result from transform domain encoding pels without reordering pels. Accordingly, the video editing system of the present invention accounts for the encoding differences between the field encoded blocks and the frame encoded blocks when resizing the video stream in the transform domain. To this end, the video editing system includes field resizing matrices for resizing the field encoded blocks and frame resizing matrices for resizing the frame encoded blocks.

Abstract: An automated method, storage medium, and system for analyzing bone. Digital image data corresponding to an image of the bone are obtained. Next there is determined, based on the digital images, a measure of bone mineral density (BMD) and at least one of a measure of bone geometry, a Minkowski dimension, and a trabecular orientation. The strength of the bone is estimated based upon the measure of BMD and at least one of the measure of bone geometry, the Minkowski dimension, and the trabecular orientation. To improve bone texture analysis, the present invention also provides a novel automated method, storage medium, and system in which digital image data corresponding to an image of the bone is obtained, and a region of interest (ROI) is selected within the bone. A fractal characteristic of the image data within the ROI using an artificial neural network is extracted. The strength of the bone is estimated based at least in part on the extracted fractal characteristic.

Abstract: A word recognition device uses an associative memory to store a plurality of coded words in such a way that a weight is associated with each character of the alphabet of the stored words, wherein equal weights correspond to equal characters. To perform the recognition, a dictionary of words is first chosen; this is stored in the associative memory according to a pre-determined code; a string of characters which correspond to a word to be recognized is received; a sequence of weights corresponding to the string of characters received is supplied to the associative memory; the distance between the word to be recognized and at least some of the stored words is calculated in parallel as the sum of the difference between the weights of each character of the word to be recognized and the weights of each character of the stored words; the minimum distance is identified; and the word stored in the associative memory having the minimum distance is stored.

Abstract: A method of binarization used in an OCR system involves in determining text pixels by checking, for each pixel, that the difference between its value and the values of a plurality of pixels located at a predetermined distance therefrom is greater than a relative threshold corresponding to the difference in intensities between the text and the background of the image, subsampling the image at a rate corresponding to at least two pixels in order to detect kernels of text, and then binarizing the image pixels only in tiles of several stroke width sides containing text kernels by using in each tile, an absolute threshold estimated in that tile.

Abstract: A message in the form of a sequence of bits is encoded into relatively short, non-binary signals. The message (or vector) is divided into two halves. The encoder first determines an encoding of the message into a vector half the length of the eventual encoded vector. The half-length vector forms the first half of the encoded, and is negated to form the second half of the encoded vector. An identical synchronization signal is added to both halves of the encoded vector. The synchronization signal is used to compensate for the possibility that the received vector (or message) is cyclically rotated from the original vector. During decoding, the decoder first obtains the synchronization signal by adding the two halves of the vector together, canceling the encoded message. The synchronization signal is used to find the original cyclical shift of the vector. The decoder obtains the half-length encoded message vector by taking the difference between the two halves of the aligned full-length encoded vector.

Abstract: The present invention comprises a method for the detection and segmentation of bright images within a digital image using wavelets. One example of such a bright image is microcalcifications within a mammogram. Multiresolution analysis may be used to detect and segment the possible microcalcification areas by combining Bayes classifiers. By analyzing the time-frequency characteristics of clustered microcalcifications, the inventors first choose the optimized wavelet for the detection of microcalcifications. A wavelet packet analysis is then used to detect different size microcalcifications. An adaptive method of choosing the threshold for detection by using a one-dimensional wavelet transform to analyze the PDF of the images at different scales is used. Then, a scheme to detect different size microcalcifications in different scale wavelet packet transformed images is developed.

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 linear transformation of the image signal on a segment with an arbitrary geometrical shape includes the following steps: the sub-division of the segment into sub-segments of regular shapes (rectangular, square or linear shapes), the application of an orthogonal linear transformation to each sub-segment, the combining of the coefficients coming from the first transformation into classes of coefficients according to a predetermined criterion, and the iteration of the transformation on the classes of coefficients, and which can be applied to the processing of images for encoding.

Abstract: A novel wafer or circuit board automatic scanning optical inspection system and technique using the reference comparison principle for finding defects in scanned sample images, that enables a new degree of universality in finding all defects including very small single pixel conductor breaks and shorts and defects of irregular shapes, and together with simultaneous design rule processing, through the use of processed skeletal reference images and with separately programmable alignment tolerance and detection parameters.

Abstract: A method is disclosed for finding a pose of a geometric model of a fiducial mark within an image of a scene containing the fiducial mark. The method includes receiving input from a user including geometric model shape information and geometric model dimensions, selecting a pre-defined model having sub-models by using the model shape information, the pre-defined model being parameterized by the geometric model dimensions; and then using the pre-defined model to find an image of the fiducial mark in the image of the scene. The act of selecting a pre-defined model having sub-models includes pre-defining a model such that the sub-models and the relationships among the sub-models are selected by exploiting knowledge of features and properties of the fiducial sought so as to provide more accurate location of the fiducial. The invention provides enhanced ease of use of a sub-model-based search method by requiring only the shape and dimensions of the fiducial sought.

Abstract: A figure is input, and structural features of the input figure are extracted. Structural features of deformed figures are produced, in accordance with a specific transformation rule, from the extracted structural features. The input figure is classified based on the structural features thus extracted and thus produced.

Abstract: An image data compression apparatus capable of reducing false color in image data to a minimum when image compression or decompression is performed, including a preliminary image compressing portion for compressing image data in a sampling region set in a specific position on an image plane for obtaining preliminary compressed image data; a data size evaluation value calculating portion for calculating a data size evaluation value from a data length of said preliminary compressed image data; a compression ration determining portion for determining compression ratios of the image data to color and luminance data in accordance with the data size evaluation value such that a data size of compressed image when image compression is performed for the entire image plane of the image data equals to a prescribed desired data size and such that the compression ratio of the image data to the color data equals to or exceeds the compression ratio of the image data to the luminance data; and an image compressing portion fo

Abstract: A method and apparatus for inspecting a feature formed on the surface of a semiconductor wafer predicts the profile of the feature and pinpoints the stepper settings of the inspected feature by inspecting the feature using standard SEM imaging techniques. Embodiments include forming and SEM-imaging a plurality of reference features, comparable to the target feature to be inspected, on a reference semiconductor wafer, each of the reference features associated with a known profile and stepper setting. The reference SEM waveform associated with an optimal profile is selected as a golden waveform, then a waveform corresponding to the target feature is compared with the golden waveform. If the target waveform does not substantially match the golden waveform, the reference waveform which most closely matches the target waveform is identified, to determine the profile and stepper settings of the target feature.

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 background subtraction apparatus of the present invention includes a key point locator for locating key points on a known object type, a boundary point locator for locating boundary points of the known object that make up the edges of the known object, and an edge processor for processing the edges to provide a clean-edged extraction of the known object from a background image. Preferably, the key point locator includes an alignment detector for detecting alignment of an image of the known object type with a skeleton image. Still more preferably, the skeleton image is an exoskeleton image and the known object type is a human being.

Abstract: A method for comparing a ranking scheme for an object, such as a document image, with a known standard, or ideal ranking scheme for a comparative assessment of the performance and reliability of the two ranking schemes, is taught. The same set of objects are processed and ranked by both ranking schemes. The ranking scheme being tested is examined for all subject objects ranked lower than its ranking by the ideal ranking scheme. The test ranking scheme is charged for each swapped object which it ranked higher than the subject object, which the ideal ranking scheme ranked lower.

Abstract: When a GC (i.e., computer graphics) image is synthesized with an actual photographic image (dynamic image), it occurs that the synthesized image appears unnatural. An image pickup section 2 picks up a plate 1 while moving. When synthesizing a ball formed through CG with a continuous image picked up by the image pickup section 2, how 12 patterns (in the portion enclosed by a frame 21) Aj (j=1 . . . , 12) present a ball 5 are transferred in the continuous image picked up by the image pickup section 2 is shown by a rotational transform function and a rectilinear transform function to obtain the synthesis position of the projected image of the ball 5 in each image by using the rotational transform function and rectilinear transform function.