Abstract: Systematic use of infrared imaging characterizes marks made on items and identifies the particular marking tool with better accuracy than use of visual imaging. Infrared imaging performed in total darkness eliminates shadows, glint, and other lighting variations and artifacts associated with visible imaging. Although normally used to obtain temperature measurements, details in IR imagery result from emissivity variations as well as thermal variations. Disturbing an item's surface texture creates an emissivity difference producing local changes in the infrared image. Identification is most accurate when IR images of unknown marks are compared to IR images of marks made by known tools. However, infrared analysis offers improvements even when only visual reference images are available.

Abstract: Optical flow is calculated for each sub-block on the basis of two images, the sub-blocks are then partitioned into a region having low confidence and a region having high confidence depending on the reliability of the optical flow, and the optical flow in the region having low confidence is filled by dilation processing using the optical flow in its surrounding regions.

Abstract: The invention presents The Estimated Spectrum Adaptive Postfilter (ESAP) and the Iterative Prepost Filter (IPF) algorithms. These algorithms model a number of image-adaptive post-filtering and pre-post filtering methods. They are designed to minimize Discrete Cosine Transform (DCT) blocking distortion caused when images are highly compressed with the Joint Photographic Expert Group (JPEG) standard. The ESAP and the IPF techniques of the present invention minimize the mean square error (MSE) to improve the objective and subjective quality of low-bit-rate JPEG gray-scale images while simultaneously enhancing perceptual visual quality with respect to baseline JPEG images.

Abstract: A method, system and program product for compressing a compound image, comprising forming compound image data into a plurality of blocks; obtaining classifying data that designate one from a plurality of classes for each of a plurality of said blocks, based on predominate compression properties of the block; for each of a plurality of said blocks, obtaining transformed data therefor from either a lossy or a lossless transform selected based on the classifying data for the block; and forwarding that transformed data to one entropy encoder. In one embodiment, obtaining transformed data comprises obtaining transformed data from at least one of a lossy transform, a first lossless transform, or a second lossless transform, based on the classifying data for the transform.

Abstract: Methods for identifying a note with a partial OCR of one or more characters in the note's serial code. The methods record the characters read and associates each character read with its field position in the serial code. The method can use this information or combine the character and field information with positional information of the note within a stack in order to identify a note with a reasonable degree of statistical probability.

Abstract: A two-dimensional (2D) image model models the layout structure of a class of document images as an image grammar and includes production rules having explicit layout parameters as data items that indicate information about the spatial relationships among image constituents occurring in images included in the class. The parameters are explicitly represented in the grammar rules in a manner that permits them to be automatically trained by a training operation that makes use of sample document images from the class of modeled documents. After each sample image is aligned with the 2D grammar, document-specific measurements about the spatial relationships between image constituents are taken from the image. Optimal values for the layout parameters are then computed from the measurement data collected from all samples.

Abstract: An electronic database for image interpolation is generated by a computer. The computer generates a low-resolution image from a training image, a plurality of representative vectors from the low-resolution image, and a plurality of interpolation filters corresponding to each of the representative vectors. The interpolation filters and the representative vectors are generated off-line and can be used to perform image interpolation on an image other than the training image. The database can be stored in a device such as computer or a printer.

Abstract: A three-dimension distance time-of-flight system is disclosed in which distance values are acquired by a plurality of sensors independently from each other. For use with this and similar systems, Z-distance accuracy and resolution are enhanced using various techniques including over-sampling acquired sensor data and forming running averages, or forming moving averages. Acquired data may be rejected if it fails to meet criteria associated with distance, luminosity, velocity, or estimated shape information reported by neighboring sensors. A sub-target having at least one pre-calibrated reflectance zone is used to improve system measurement accuracy. Elliptical error is corrected for using a disclosed method, and reversible mapping of Z-values into RGB is provided.

Abstract: Two different recognition processing units respectively perform a plurality of different recognition processing operations for the pattern to be recognized input from a character input unit, and acquire a plurality of recognition candidates and recognition scores. An accuracy calculation unit converts the recognition scores obtained by the two recognition processing units into accuracies based on a scale common to the plurality of different recognition processing operations using parameters stored in a parameter storage unit. A determination unit generates a recognition candidate sequence by determining the order of the recognition candidates obtained by the two recognition processing units on the basis of the accuracies converted by the accuracy calculation unit.

Abstract: During the operation of a document processing system machine and job data are collected from a document processing system. Optionally, machine information, both for the specific machine and population based data are acquired from a database or server. Having accumulated the job data and machine data, the diagnostic inference engine performs an analysis to determine the initial diagnosis of the document processing system. After obtaining the initial diagnosis, the system determines the test patterns to be printed and the image quality tests to be performed. The system then prints test patterns, and scans the patterns to determine image quality parameters and/or to automatically identify image defects. Optionally, image defect information may also be provided by the customer or the service engineer via a user interface. Next, a diagnostic inference engine uses the results of the image quality analysis to refine the initial diagnosis.

Abstract: An automated analysis method and tool for a digitally imaged financial item can digitally recognize a plurality of characters across an area of a digital image containing character indicia at predetermined positions in the area and also possibly containing non-character indicia in the area. This finds exemplification in an automated check processing method and check sorting system in which a check's MICR line that has been read and interpreted as invalid can be digitally analyzed, via a digital image of the check, to identify MICR characters that could not be validly read or interpreted by the conventional MICR processing equipment. This includes digitally applying character recognition processing to an invalidated MICR line in a digital image of the financial item (e.g., a check). Previously invalidated MICR line data can then be corrected.

Abstract: Upon observation of multi-viewpoint images, even when the viewpoint moves continuously, the image cannot normally be displayed in correspondence with the continuous movement of the viewpoint, and an image from a viewpoint closest to the current viewpoint is displayed. To prevent this, each two adjacent images of the multi-viewpoint images are interpolated by detecting corresponding pixels and generating an interpolated image in accordance with the correspondences among pixels. The interpolated image is generated in such a manner that an epipolar plane image is generated by juxtaposing corresponding lines of the multi-viewpoint image, straight lines are detected from the epipolar plane image, and the intersecting points between interpolated lines and the detected straight lines are detected as interpolated pixels. In this case, as for a point for which a straight line cannot be detected, a straight line passing the point is estimated, thereby generating an interpolated pixel.

Abstract: A method for identifying a detected object in multi-dimensional imagery data is disclosed. The method includes, (a) providing (1) a set of comparative feature values associated with one or more known objects in one or -more specified observation conditions and (2) a set of observation-related values including (i) a set of feature values for one or more features of an object as observed in an environment, (ii) a set of condition values for one or more conditions of the observation, including one or more of an orientation value, a depression value, a resolution value, and an obscuration value; and (b) generating one or more identification values for the object by comparing the set of feature values to the set of comparative feature values, including compensating for the set of condition values.

Abstract: When additional information is embedded into image data, both image quality and tolerance are improved. The image processing method includes the steps of: providing a plurality of pixels or a plurality of spatial-frequency components into which additional information is to be embedded; and changing the data value of the plurality of embedding pixels or spatial-frequency components in accordance with a value corresponding to the data value of the spatial-frequency components in order to embed the additional information.

Abstract: A method and apparatus for preventing or inhibiting effective reproduction of documents such as currency, checks, stock certificates, and any other printed document including a pre-defined security mark printed therein. The subject method and apparatus operate to effect a multi-step review of all digital image data acquired from a printed document to be reproduced for purposes of locating any potential security marks and further examining same for purposes of positively identifying a potential security mark as an actual security mark. If a mark is located and verified to be an authentic security mark, effective reproduction of the printed document will not be permitted and/or other security measures will be taken.

Abstract: A method of image processing in three-dimensional (3-D) head motion tracking is disclosed.

Abstract: An input image is distorted so that it has less correlation to a watermark and the distorted image is subtracted from the input image to provide a preprocessed image. If the correlation between the preprocessed image and a watermark is sufficient then the watermark is detected in the input image. The distortion may be a spatially-variant distortion such as stretching, shrinking, shearing or rotation or the distortion may be a non-linear transforming or filtering such as median, local minimum or local maximum filtering.

Abstract: The present invention relates to a method for three-dimensional image reconstruction of basal ganglion. A novel geometrical algorithm has been developed to calculate the correction coordinates of the target based on the reference axial shift in the CT scan coordinate system. Furthermore, wavelet transform along with interpolation techniques are used to obtain continuous sectional images and three-dimensional image reconstruction is then performed to form the stereotactic atlas of basal ganglion. Therefore, the stereotactic atlas of basal ganglion established in this invention can be used as references for assisting operation and training for neurosurgeons.

Abstract: A system for image object tracking and segmentation, includes (i) a modified adaptive resonance theory-2 (M-ART2) model for detecting changes of scenes, (ii) a two-dimensional correlative autopredicitve search (2D CAPS) method for object tracking, (iii) a derivation of contour energy for use in modelling an object edge, and (iv) an active contour model with global relaxation for defining optimal image object boundaries. The modelled object boundary is derived from an edge potential energy calculation and a set of current edge points. For each edge point, several candidate data points are also selected. Rather than calculate an energy value for each potential contours, a solution is achieved by comparing energy differences along a travel path. The selected contour is built in steps. An optimal path is selected from the potential paths.

Abstract: A system for monitoring and reporting incidences of traffic violations at a traffic location is disclosed. The system comprises a digital camera system deployed at a traffic location. The camera system is remotely coupled to a data processing system. The data processing system comprises an image processor for compiling vehicle and scene images produced by the digital camera system, a verification process for verifying the validity of the vehicle images, an image processing system for identifying driver information from the vehicle images, and a notification process for transmitting potential violation information to one or more law enforcement agencies.