Abstract: An apparatus and method for inspecting skived glue, hemming and skew, and backfold of cartons by means of real time image comparison. The apparatus comprises channel sensors, means for measuring conveyor travel distance, a system controller coupled to the channel sensors, a memory for storing images, a processor coupled to the outputs of the system controller and the memory and a bad carton locator means. A known good carton is passed through the channel sensors and a reference synthetic image is stored in a memory. A test carton is passed through the channel sensors producing a real time image which is compared to the stored reference synthetic image for detecting defects.

Abstract: A two-dimensional inverse discrete cosine transformation circuit of an MPEG2 video decoder including a one-dimensional inverse discrete cosine transformation circuit, an input switching circuit for receiving input of new data and data already subjected to first one-dimensional inverse discrete cosine transformation and sending one of them to the one-dimensional inverse discrete cosine transformation circuit, an input switching control circuit for controlling the input switching circuit so as to alternately and continuously output data output from a first serial-parallel conversion circuit and data output from a second serial-parallel conversion circuit, and a data allocation circuit for switching and controlling an output destination of output data of the one-dimensional inverse discrete cosine transformation circuit based on the timing of switching by the input switching control circuit.

Abstract: A method for determining spatial characteristics of a selected object in an image that has pixel values representing heights of corresponding portions thereof is characterized by the steps of finding one or more heights of portions of the object, determining a volume of the object at and above (or below) a selected height by summing a count representing a cross-sectional area of the object at that height with a count representing a volume of the object at a greater (or lesser) height, and outputting the volume so determined. A related method determines a cross-sectional area of an object in such a "height" image by finding one or more heights of portions of the object and summing a count of pixels in the object at the selected height with a count representing a cross-sectional area of the object above that height.

Abstract: A method and system for compositing graphical images, wherein an advanced adjustment layer may be applied during a compositing process to a set of image layers 1 . . . n, or to any subordinate subset of such image layers. One or more adjustment layers are applied to an intermediate merged image, generated by compositing previous image layers, and the adjusted result is stored as a temporary image. The temporary image is then in turn composited with the intermediate merged image. Any remaining image layers are then composited in with the intermediate merged image to generate a final merged image. The invention allows a user to apply a vast array of effects without requiring significant new knowledge on the part of the user. For example, if there are "A" adjustments and "T" transfer modes, the present invention allows A.times.T effects which leverage existing knowledge of the user of only A+T functions.

Abstract: The reconstruction region of a hologram reconstructed image is segmented. A fast Fourier transform (FFT) is carried out after multiplying the hologram measurement data V(R) by a weighting function. A fast Fourier transform (FFT) is carried out after multiplying the antenna sensitivity inverse matrix A.sup.-1 (R'.sub.a,b) for each segment by inverse propagation function r'.sub.a,b exp(j2.pi.r'.sub.a,b /.lambda.) to find B.sup.-1.sub.a,b (s, t), and then multiplying this by a weighting function. After the two FFT outputs are multiplied by each other, inverse fast Fourier transform (IFFT) is carried out. Compensation is performed for the weighting function and the output is made the segment reconstructed image. In addition, radio hologram observation is carried out at two scan surfaces, a horizontal polarization antenna and a vertical polarization antenna are placed at each scan surface, and the voltage vector distribution received by the antennas is stored.

Abstract: Machine vision methods for determining a characteristic (such as position, orientation, size, center of mass, and boundary) of an object in an image include finding points in the image on the boundary of the object; identifying bounding boxes, or regions, in the image that correspond to edges of the object; and labeling boundary points to denote which respective edges, if any, to which they belong based on (i) the locations and orientations of those points, and (ii) locations of the plural bounding boxes. Points apparently lying on a boundary of the object, but outside a bounding box, can be ignored--and, more particularly, are denoted as not corresponding to an edge. Likewise, apparent boundary points lying within a bounding box, but at an angle not associated with the corresponding to the respective edge of the object, can also be ignored. Characteristics of the imaged object are determined as a function of those boundary points found to correspond to an edge of the object (e.g.

Abstract: A method of locating a machine-readable two dimensional marker, such as a barcode, in an image. An image of a medium containing the two dimensional marker is processed to generate a mask containing one or more filled regions each of which is sufficiently large to contain the predetermined dimensions of the marker. To generate the mask, small image areas having a first pixel polarity which are internal to larger image areas having a different pixel polarity are first filled with pixels having the same polarity as the surrounding larger image areas. A morphological opening operation is then performed on the filled image to leave only such filled regions as are equal in size or larger than the predetermined dimensions of the marker. After the mask has been generated, the mask is logically combined with the original image to extract one or more candidate regions.

Abstract: A method for scanning and detecting number and skew angle of photographs (12) and removing edge artifacts is disclosed. A plurality of photograph is placed on a scanner (11). A low resolution scan (34) is conducted of the photographs (12) to produce a plurality of low resolution images. A plurality of polygons is constructed, wherein each polygon contains at least one low resolution image. A number of polygons is determined and compared to the number of photographs. If the number of polygons is less than the number of photographs, the photographs are repositioned to separate overlapped photographs and the low resolution is repeated. A skew angle and location is determined for each of the photographs. A first photograph is selected and the first photograph is scanned at a high resolution to produce a high resolution image. The high resolution image is deskewed and cropped to remove the edge artifacts.

Abstract: A signature card is used with a computing system. The card plugs into a host computer and is, for example, a personal computer memory card internal association (PCMCIA) card. The card includes a write-on pad for receiving input from a stylus. A memory stores the input received by the write-on pad. An interface transfers the input stored by the memory to the host computer. The card also includes an independent power source which is independent from the host computer.

Abstract: In a picture-information processing apparatus, information of an edge in a picture is generated from information of luminances in the picture. Data of a mask corresponding to the edge is generated from the information of the edge. Information of feature points on an equal-luminance line in the picture is generated from the information of the luminances in the picture. The equal-luminance line is composed of segments extending between the feature points. A decision is made as to whether or not each of the segments of the equal-luminance line agrees with an edge line segment in response to the data of the mask and the information of the feature points. An edge decision signal is generated which represents the result of the decision. The edge decision signal is added to the information of the feature points.

Abstract: A method for preprocessing a binary file for data compression under a dictionary-based data compression algorithm takes advantage of redundancy in a two-dimensional binary image. The method rearranges a linear representation of a binary image, i.e. a representation based on pixels of horizontal lines, to a two-dimensional representation, i.e. a representation based on a sequence of adjoining picture areas, to achieve an improvement of compression ratio. The present invention is applicable to dictionary-based data compression methods, such as LZW, LZ77 and LZ78.

Abstract: A method for quantitatively assessing diffusion anisotropy according to an invariant anisotropy index that accounts for orientational coherence of the measured principal directions between different localized regions of an object to counteract the bias and increased variance effects of noise inherent in the diffusion measurement. A diffusion weighted imaging sequence is performed on a two-dimensional slice of an object to provide raw diffusion weighted image signals, which are processed by conventional Fourier transform and magnitude reconstruction to provide diffusion weighted images, from which a diffusion tensor is estimated for each voxel of the imaged slice. In each voxel a lattice anisotropy index is calculated as a function of both the eigenvalues and eigenvectors of neighboring voxels such that intervoxel orientational coherence compensates noise-induced bias effects. The orientational coherence measure between two voxels is calculated according to an intervoxel deviatoric tensor dot product.

Abstract: A neural network based improving the performance of an omni-font classifier by using recognized characters for additional training is presented. The invention applies the outputs of the hidden layer nodes of the neural net as the feature vector. Characters that are recognized with high confidence are used to dynamically train a secondary classifier. After the secondary classifier is trained, it is combined with the original main classifier. The invention can re-adjust the partition or boundary of feature space, based on on-line learning, by utilizing the secondary classifier data to form an alternative partition location. The new partition can be referred to when a character conflict exists during character recognition.

Abstract: A display screen inspection method according to the present invention takes a display screen of a display device as an inspection subject and makes decisions on defects of the display screen. The method includes performing a separation between display operating portions and display non-operating portions of the inspection subject present in a gray image from image data of individual pixels of the gray image obtained by picking up an image of the inspection subject, selectively extracting image data of only proper display operating portions and compressing the image data to be inspected for defects, and inspecting the selectively extracted image data for any defects.

Abstract: Described above is a method of enlarging a character that is represented by a reference pattern of pixels. The method includes a step of converting pixels of the reference pattern to corresponding rectangles of pixels in an enlarged pattern of pixels. Further steps includes identifying acute right angles in the reference pattern, wherein each such acute right angle has horizontal and vertical edges with corresponding lengths, and determining the lengths of the horizontal and vertical edges of the acute right angles. Right angles neighboring identified acute right angles are also found and the lengths of their edges are calculated. A determination of whether to smooth a right angle in the enlarged pattern is made by examining the corresponding right angle in the reference pattern and its neighboring right angles. Specifically, the determination is made based upon the lengths of the edges of the various right angles.

Abstract: An automatic recognition method capable of accurately determining edges of a concavity or convexity in order to achieve an improved measuring accuracy in recognition of a concavity or convexity. This method comprises the steps of: converting a photographic image of an object having a concavity or convexity defined therein into data in a specified feature space; obtaining a plurality of points that exist outside a virtual circle determined by a virtual center of the concavity or convexity, the virtual center being established for each elementary area of the image converted into the feature space data; obtaining the equation of a plane that passes through the plurality of points; detecting symmetric pairs of pixels in each elementary area of the feature space data with respect to the corresponding virtual concavity or convexity center; and recognizing the position of the concavity or convexity of the object when the symmetric pixels satisfy the equation of the plane.

Abstract: A sequential polygon approximation apparatus for a contour and a method thereof which are capable of sequentially extracting a vertex in the order of a contour pixel when performing a polygon approximation so as to code a shape information, which includes an error computation unit for obtaining a contour pixel between two contour pixels of the vertex extraction unit from the contour pixel memory and computing an error which is a distance between a straight line between two contour pixels and a pixel from the contour pixel memory, a comparison unit for comparing an error obtained by the error computation unit with a threshold value, and a controller for controlling the vertex extraction unit and the error computation unit in accordance with the comparison result of the comparison unit, setting the pixel just before the final pixel of a first region as a first vertex when the error obtained in the first region which is the region of the contour pixel sequentially determined for the vertex extraction from a fixe

Abstract: An electronic scanner having a light source and a light sensitive head is connected via a cable to a computer. The scanner head detects reflected light from the surface of the composite material and generates an electronic representation of the surface. A conventional software driver interprets the scanner output to produce an electronic bit-mapped image. The electronic image is then displayed so that gaps are readily visible. The electronic representation is also analyzed to determine the presence of laps and gaps, to measure gap widths, to measure the distance between points on the display, and to determine the percentage of the surface covered by fiber material. The electronic representation may also be stored for later analysis.

Abstract: Surface relief of a finger etc. is read using an optical-fiber prism unit, with fiber terminations at one end to contact the surface, and at the other for light passage along fibers from the first. Light enters where NA<0.5 and fiber diameter is constant with longitudinal position. The device is in a 1.4-2 L case, with a battery or power input, converter to form a corresponding data array for verifying, digital signal processor to do the verifying, and output to indicate or implement a decision. A video controller (with custom-programmed logic circuit) operates the sensors to develop the data array; an ADC digitizes the array; memory holds an authorized-user skin-pattern template, firmware for the processor, and data used in verifying; an output register holds the decision signal--all on a control, address, and data bus. High-power, radiative elements and a fast high-impedance data reader are on a common board in an isolating layout.

Abstract: The detection method and apparatus detects abnormalities having a first signal propagation property within a host medium having a second signal propagation property based upon signals which have been reflected, scattered and/or absorbed by the host medium and any abnormalities therein. The detection apparatus includes a signal source for introducing a signal into the host medium which varies with respect to a predetermined signal parameter, such as time, frequency or source location. The detection apparatus also includes a plurality of detectors disposed along some or all of the boundary of the host medium for detecting the signal following propagation through the host medium and any abnormalities within the host medium. The detection apparatus further includes a signal processor for constructing at least one partial differential equation to describe the propagation of a regularized signal through a medium having the same shape as the host medium.