Abstract: In the case where a three-dimensional image including a plurality of element figures, such as a quadrangular plane, a curved surface including a ridgeline, and a cylinder, is edited using an image editing apparatus such as a three-dimensional CAD apparatus, when one element figure in the three-dimensional image is specified, other element figures connected to the specified one element figure, or other element figures included in the same three-dimensional image in which the one element figure is included are retrieved, and other element figure in the retrieved other element figures, which was determined as having a specific relationship to the one element figure is related to the one element figure. Then, the specified element figure and the element figures related to the specified element figure are edited collectively based on an operation received by the image editing apparatus.

Abstract: This document describes techniques that utilize a learning method to generate a ranking model for use in image search systems. The techniques leverage textual information and visual information simultaneously when generating the ranking model. The tools are further configured to apply the ranking model responsive to receiving an image search query.

Abstract: A method and system for computer-aided detection of abnormal lesions in digital mammograms is described, wherein digital films are processed using an automated and computerized method of detecting the order and orientation of a set of films. In one embodiment, anatomic features are used to detect the order, orientation and identification of a film series. In another embodiment of the invention, a technologist feeds films into the system in any order and orientation. After processing, the system provides an output on a display device to a radiologist that is in an order and orientation preferred by the radiologist. In yet another embodiment of the invention, films from one case are distinguished from films of another case. In this manner and through the use of a bulk loader, a large number of films can be stacked together and fed into the system at one time.

Abstract: An image retrieval method includes learning multiple object category classifiers with a processor offline and generating classifications scores of images as the semantic attributes; performing vocabulary tree based image retrieval using local features with semantic-aware co-indexing to jointly embed two distinct cues offline for near-duplicate image retrieval; and identifying top similar or dissimilar images using multiple semantic attributes.

Abstract: A unified system of programming communication. The system encompasses the prior art (television, radio, broadcast hardcopy, computer communications, etc.) and new user specific mass media. Within the unified system, parallel processing computer systems, each having an input (e.g., 77) controlling a plurality of computers (e.g., 205), generate and output user information at receiver stations. Under broadcast control, local computers (73, 205), combine user information selectively into prior art communications to exhibit personalized mass media programming at video monitors (202), speakers (263), printers (221), etc. At intermediate transmission stations (e.g., cable television stations), signals in network broadcasts and from local inputs (74, 77, 97, 98) cause control processors (71) and computers (73) to selectively automate connection and operation of receivers (53), recorder/players (76), computers (73), generators (82), strippers (81), etc.

Abstract: Multiple discrete objects within a scene image captured by a single camera track are distinguished as un-labeled from a background model within a first frame of a video data input. Object position and object appearance and/or object size attributes are determined for each of the blobs, and costs determined to assign to existing blobs of existing object tracks as a function of the determined attributes and combined to generate respective combination costs. The un-labeled object blob that has a lowest combined cost of association with any of the existing object tracks is labeled with the label of that track having the lowest combined cost, said track is removed from consideration for labeling remaining un-labeled object blobs, and the process iteratively repeated until each of the track labels have been used to label one of the un-labeled blobs.

Abstract: A method for fine-grained image classification on an image includes automatically segmenting one or more objects of interest prior to classification; and combining segmented and original image features before performing final classification.

Abstract: Provided is a system for enumerating local alignments, comprising a local alignment enumeration module for enumerating local alignments, in a case where the cell of a second matrix corresponding to a cell at a transition source indicates that the cell belongs to a local alignment, and in a case where the maximum score calculated is larger than the predetermined value, registers, in the cell of the second matrix corresponding to the cell to be calculated, an identifier of a local alignment registered in the cell of the second matrix corresponding to the cell at the transition source, and further, in a case where the maximum score calculated is larger than a maximum score of the cells belonging to the same local alignment, stores the cell to be calculated as an end point of the local alignment.

Abstract: Aspects of the present invention are related to systems and methods for automatic content-boundary detection in a digital image. According to one aspect of the present invention a received image may be preconditioned to form a normalized image from which image-content edges may be detected. Projection histograms in the direction of a known skew angle associated with the image and the normal to the skew angle may be formed to determine image-content boundaries.

Abstract: A nearest-neighbor-based distance metric learning process includes applying an exponential-based loss function to provide a smooth objective; and determining an objective and a gradient of both hinge-based and exponential-based loss function in a quadratic time of the number of instances using a computer.

Abstract: The present method relates to digital color image processing, in particular restoration and recovery of lost contrast and details in highlights due to overexposure or clipping by reconstructing components based on the signal remaining in at least one non-clipped signal component.

Abstract: A method of image processing comprising receiving a plurality of interpolated images, interpolated from two adjacent camera positions having different image planes, applying a transformation to each interpolated image to a respective one of a plurality intermediate image planes, wherein each intermediate image plane is oriented intermediate to the image planes of the two adjacent camera positions depending on a viewing angle of that interpolated image relative to the adjacent camera positions. Also an integrated circuit or processor, an apparatus for capturing images and an apparatus for displaying images.

Abstract: An exemplary method includes prompting a user to capture video data at a location. The location is associated with navigation directions for the user. Information representing visual orientation and positioning information associated with the captured video data is received by one or more computing devices, and a stored data model representing a 3D geometry depicting objects associated with the location is accessed. Between corresponding images from the captured video data and projections of the 3D geometry, one or more candidate change regions are detected. Each candidate change region indicates an area of visual difference between the captured video data and projections. When it is detected that a count of the one or more candidate change regions is below a threshold, the stored model data is updated with at least part of the captured video data based on the visual orientation and positioning information associated with the captured video data.

Abstract: A three-dimensional ultrasonic inspection apparatus 10 can make increased inspection accuracy in comparison to the conventional apparatus and includes: an ultrasonic transducer 11 disposed m×n piezoelectric vibrators in a matrix; a signal processing unit 17 to receive, detect an echo, and generate a three-dimensional image data by processing an electric signal of the echo detected; a peak detecting element 51, 52 to detect a first peak and a second peak of an intensity distribution of the three-dimensional image data in a depth (z) direction; a joint portion image creation unit 36 to create a three-dimensional image of the joined area 15 by mapping z direction distance of the first peak and the second peak to x-y plane; a determination unit 37 to determine whether the joined area 15 is sound by two-step determination; and a display unit 38 to display the three-dimensional image and the determination result, of the joined area 15.

Abstract: An optical navigation subsystem may be used when obtaining a candidate fingerprint for a mobile communication device (e.g., for authentication purposes). To accommodate such use of the optical navigation subsystem, the optical navigation subsystem may be adapted to automatically adjust a processed image sensor window from a first set of window dimensions (e.g., suitable for optical navigation) to a second set of window dimensions (e.g., suitable for fingerprint capture). Alternatively, a single, static set of window dimension may be employed in conjunction with stitching algorithms, or other methods suitable for forming candidate fingerprint images by combining a plurality of images obtained using the static set of window dimension.

Abstract: Systems and methods are disclosed to categorize images by detecting local features for each image; applying a tree structure to index local features in the images; and extracting a rank list of candidate images with category tags based on a tree indexing structure to estimate a label of a query image.

Abstract: A computing device reads a reference image and a real-time of a printed circuit board (PCB), determines feature points and feature information of the feature points in the reference image; and creates two 1×N matrices based on the feature points. Furthermore, a mapping matrix is determined based on the two 1×N matrices. The device determines matching points in the real-time image based on coordinates of base points in the reference image and the mapping matrix, determines a matching region the real-time image based on the matching points, and determines an angle between the matching region and an X-axis of a coordinate system. If the angle does not equal zero, the device determines that the real-time is tilted, and corrects the real-time image to obtain a corrected image by taking a center of the real-time image as a turning pivot to rotate the real-time image until the angle equals zero.

Abstract: A method for extracting feature vectors from a palm image includes the steps of: determining a palm contour in the palm image, and labeling pixels on the palm contour as contour pixels in order along the palm contour; determining a rotation angle of the palm contour relative to a coordinate system; obtaining corrected contour pixels to offset the rotation angle; determining a plurality of feature points from the corrected contour pixels; obtaining a plurality of sub-images from the palm image with reference to the feature points, one of the sub-images corresponding to a palm center and another one of the sub-images corresponding to a corresponding palm finger; and determining the feature vectors with reference to the sub-images, each of the feature vectors corresponding to a corresponding one of the sub-images.

Abstract: Techniques for intelligent image search results summarization and browsing scheme are described. Images having visual attributes are evaluated for similarities based in part on their visual attributes. At least one preference score indicating a probability of an image to be selected into a summary is calculated for each image. Images are selected based on the similarity of the selected images to the other images and the preference scores of the selected images. A summary of the plurality of images is generated including the selected one individual image.

Abstract: A method and apparatus for estimating stereo misalignment using modified affine or perspective model. The method includes dividing a left frame and a right frame into blocks, comparing horizontal and vertical boundary signals in the left frame and the right frame, estimating the horizontal and the vertical motion vector for each block in a reference frame, selecting a reliable motion vectors from a set of motion vectors, dividing the selected block into smaller features, feeding the data to an affine or a perspective transformation model to solve for the model parameters, running the model parameters through a temporal filter, portioning the estimated misalignment parameters between the left frame and right frame, and modifying the left frame and the right frame to save some boundary space.