Abstract: The disclosure provides a method for use in a robotic system for identifying fasteners used in aircraft and other applications. The fasteners include rivets (with no slots), bolts (with two slots) and screws with one or two slots. The method accepts as input a family of related signals which are known to contain symmetry. By computing the mean variance of the zero-meaned cross-correlation signals, the method automatically selects those signals which contain pertinent information while disregarding those signals which do not. The accepted signals are then used to create a composite reference signal which can be used to determine the particular fastener type.

Abstract: Documents represented as bitmap images are transformed into coded textual data and coded graphics data by graphics and textual recognizers, which use a standard notation for recording the results of the document recognition processes, including any ambiguities, in a document description language. Recognized portions of the document, represented as editable coded data, such as for example ASCII, are placed in elements, defined in the document description language, with all contents of an element sharing some common characteristic. Elements can include, for example: character-string-elements, questionable-character-elements, questionable-word-elements, verified-word-elements, alternate-word-elements, segment-elements, and arc-elements.

Abstract: A device and a process for judging the contour of a road by processing data from an image taken of the road by a camera. In the data processing, Hough transform is used to obtain a plurality of straight lines approximate to the edges of the road. In order to recognize the contour of a road, it is most important to recognize the edges of the road. For the recognition of a road, for example, an image taken of the road is divided into upper and lower regions. In the lower region, two straight lines are given approximate to the right and the left edges of the road. In the upper region as well, two straight lines are given approximate to the right and the left edges of the road. Then, positional relationships between the intersection of the former two straight lines and that of the latter two straight lines are compared. Based on positional relationships, it can be judged very easily whether the road is straight, curves, inclines or declines.

Abstract: A method for identifying and removing noise from digitized images includes the steps for judging whether or not a group of black pixels represents a noise by analyzing the number of pixels in the group, the degree of flattening of the group and the curvature of the region. According to the inventive method, since the entire region is scanned at once, allowing the detection of various sized pixel groups, the black pixel groups meeting a predetermined level can be selected and removed from a corresponding predefined region. Thus, noise can be removed from several predefined regions using a different noise level for each region.

Abstract: A device for recognizing and matching fabric pattern-forms for cutting is constituted by a marking CAD, a pattern-match control computer, a pattern recognition device body, a camera, a monitor and console, a mouse, a camera positioning robot, a cutter, a camera and video signal changeover mechanism, an iris controller, a pattern-match and cutting table, and so on. In the marking CAD, information concerning cutting point sequence data and pattern-matching points is generated and transferred to the control computer. The pattern-matching control computer moves the camera above each of the pattern-matching points to fetch an image to thereby measure the pattern position. The cutting point sequence data are revised on the basis of the result of the measurement. When poor recognition or erroneous-recognition occurs in the pattern recognition based on the image, the pattern position is determined manually through the monitor and console and the mouse.

Abstract: A binarization method obtains an optimum threshold value for binarization which is used when converting a multi-level image data which describes an input image into a black-and-white bi-level image data. The binarization method includes the steps of counting a first number of black picture elements by varying a threshold value from a darkest tone level to a lightest tone level, counting a second number of picture elements having a tone level other than the lightest tone level, obtaining a percentage of the first number with respect to the second number for each of the varied threshold values so as to normalize a density of the input image, and determining the optimum threshold value for binarization based on the percentage which is obtained for each of the varied threshold values.

Abstract: The present invention relates to a process for the acquisition and processing of screen images or radiographs and to an apparatus making it possible to inspect or control mechanical parts. The process consists of obtaining at least two screen images of the part to be inspected, converting said images into digital video images coded on several grey levels, correcting each image with respect to itself and then with respect to a reference image and finally binarizing the corrected images in order to superimpose them, so as to determine the image points which correspond and thus detect defects. Application to the control and inspection of mechanical parts.

Abstract: A system for texture-based automatic detection of man-made objects or targets or features of scenes (automatic target recognition, ATR or detection, ATD) in representations of sensed natural environmental scenes spatially partitions digitized picture elements (pixels) of the scene into plural spatially coordinated groups of pixels and simultaneously determines texture measures including a composite texture measure for each group of pixels. Following self-calibration of the spatially coordinated composite texture measure values along spatially delineated row and orthogonal column directions of groups of pixels, areas of interest (AOIs) are identified as the groups of pixels most likely to contain a man-made object, target or feature in a decision logic which performs a group of statistical tests.

Abstract: An estimate of a threshold value for feature extraction from image data containing noise and quantized into a plurality of quantization levels is made by filtering, such as recursively averaging n adjacent quantization levels and counting the extrema of the result. The recursive averaging is repeated with increasing values of n until only three extrema are present in the filtered data. The estimate of the threshold value is refined by comparing averages or summations of n quantization levels above and below the estimate and altering the estimated threshold value based on a comparison of the averages or summations. By basing both the estimation and refinement processes on the original data and using a value of n which is smaller in the refinement process than in the estimation process, a highly effective threshold value for the feature extraction process is obtained. Adaptive adjustment of n during the estimation process minimizes loss of data as well as computing time.

Abstract: The contour fidelity of reproduction of images is enhanced by matching groups of pixels with templates based on a hierarchical set of patterns. This is accomplished by isolating a symmetrical window of pixels in the image including a central pixel and matching the groups of pixels in the window with hierarchical templates. Enhancement values for the central pixel are provided based on the results of the comparison. Windows are formed so that each pixel in the image is a central pixel for at least one window. The hierarchial set of patterns is developed so that each pattern in the hierarchy is a combination of an extension pattern and the pattern immediately below it in the hierarchy. This significantly reduces the hardware requirements for implementation of the technique. Templates are developed by creating horizontal, vertical, reversed and inverted permutations of each pattern in the set.

Abstract: Methods and apparatus for automatically tracking the position of a moving object in real time, particularly the face of a person who is being fed by a robotic system, is disclosed. The object can be tracked by comparing a prestored object model image with the current image of the object using the square-distance criteria. The search area can be limited to a region in which the face is most likely to be found and the prestored object model image can be limited to robust points. The method can include motion prediction, including both continuous motion and sudden motion, such as the motion cause by a person sneezing. Alternatively, a computationally efficient approach employing a one-dimensional algorithm can be used.

Abstract: A data processing system augments compression of non-zero values of significant coefficients by coding entries of a significance map independently of coding the values of significant non-zero coefficients. A dedicated symbol represents a zerotree structure encompassing a related association of insignificant coefficients within the tree structure. The zerotree symbol represents that a coefficient is a root of a zerotree if, at a threshold T, the coefficient and all of its descendants that have been found to be insignificant at larger thresholds have magnitudes less than threshold T. The zerotree structure is disclosed in the context of a pyramid-type image subband processor together with successive refinement quantization and entropy coding to facilitate data compression.

Abstract: A process for converting characters arranged circularly, as for example about the center hole of a compact disk, into a linear arrangement. Points are assigned to locations on the circular arrangement. These points are mapped to a linear arrangement. The number of points is selected to be greater than those of the original image to enhance the resolution of the resulting image. The location of the points is stored in an address array. The values of the pixels in the original image are then copied to a target array. The pixel values are then converted to binary values serving as input to a recognition neural network and a verification neural network.

Abstract: A character recognition system wherein the flexibility of the recognition task is expanded for orientation independence, position independence, and orientation and position independence. The system also provides a technique for implementing concurrency in the processing to achieve high speed without sacrificing performance. The system is readily implemented on conventional machine vision computing systems.

Abstract: A dither processing method comprising steps of:a) Defining a non-rectangle dither cell and a macro dither cell consisting of a plurality of the dither cells arranged in a predetermined form;b) Defining an order of magnitudes of threshold of pixels in the dither cell;c) Defining an order of magnitude of threshold of the dither cells in the macro dither cell;d) Defining a threshold for each the pixel of each the dither cell in the macro dither cell according to the order of magnitude of threshold of the pixels in the dither cell as well as the dither cells in the macro dither cell.

Abstract: A method provides for extracting features from a handwritten character which is input to an on-line input device. The on-line input device successively outputs stroke information indicating points on strokes of the handwritten character. The method includes a first step of detecting strokes of the handwritten character based on the stroke information. Thereafter, singular points are detected, based on the stroke information. Each of the singular points is defined as a pair of points on the strokes of the handwritten character which are written at different times but have substantially the same coordinates. Then, primitive sequences forming each of the strokes are obtained, and a connective relationship between each two formative sequences of each of the strokes is determined. Finally, the structure of each of the singular points is determined.

Abstract: An optical recognition method and system utilizes a scanner to digitize a complex document comprising regions of text and graphics. A recognition engine, illustratively in the form of a general purpose computer, separates regions of text and graphics and identifies lines of text. This is accomplished by dividing the digitized image into columns. The columns are then separated into small rectangular units by obtaining a horizontal histogram or shadow thereof. The units are arranged into a two dimensional linking for ease of processing and access. Units which are too short are eliminated as noise and units which are too tall are eliminated or otherwise identified as comprising graphics. The remaining units are organized into lines of text by connecting units horizontally whose shadows on the vertical axis overlap.

Abstract: An imager and an analog to digital converter yield a series of digital signals from successive frames of images of scenes in which selected types of high-speed events may take place. The signals are sampled, and changes in the sampled signals typically indicative of such events are detected. Random access memory temporarily stores a sequence of the signals including each such changed signal. Analytic circuitry determines whether the sequence contains a record of such an event, and, if it does, provides a videotape record of the event.

Abstract: Apparatus which segments visual scenes into discrete objects. The apparatus is able to perceptually group the elements corresponding to a coherent figure, and to segregate them from the background or from another figure. Reentrant signaling among rhythmically active neuronal groups is used to correlate responses along spatially extended contours. The efficacy of the connections is allowed to change on a fast time scale resulting in active reentrant connections which amplify the correlations among neuronal groups and provides the desired segmentation.

Abstract: A method of apparatus for automatic page orientation of a scanned image which compares the number of character ascending pixels to the number of character descending pixels in the image to determine if the image is properly aligned or is 90.degree. or 180.degree. out of orientation. The method and apparatus includes morphologically processing the bitmap of the scanned image using structuring elements for isolating the character ascenders and descenders. When page orientation is improper, the bitmap image of the scanned image is rotated to correct the misalignment.