Abstract: Methods and systems for determining a position of inspection data with respect to a stored high resolution die image are provided. One method includes aligning data acquired by an inspection system for alignment sites on a wafer with data for predetermined alignment sites. The predetermined alignment sites have a predetermined position in die image space of a stored high resolution die image for the wafer. The method also includes determining positions of the alignment sites in the die image space based on the predetermined positions of the predetermined alignment sites in the die image space. In addition, the method includes determining a position of inspection data acquired for the wafer by the inspection system in the die image space based on the positions of the alignment sites in the die image space.

Abstract: Systems and methods for generating information for use in a wafer inspection process are provided. One method includes acquiring output of an inspection system for die(s) located on wafer(s), combining the output for the die(s) based on within die positions of the output, determining, on a within die position basis, a statistical property of variation in values of characteristic(s) of the combined output, and assigning the within die positions to different groups based on the statistical properties determined for the within die positions. The method also includes storing information for the within die positions and the different groups to which the within die positions are assigned in a storage medium that is accessible to the inspection system for performing the wafer inspection process, which includes applying defect detection parameter(s) to additional output of the inspection system generated for a wafer based on the information thereby detecting defects on the wafer.

Abstract: Automatic object retrieval from input video is based on learned, complementary detectors created for each of a plurality of different motionlet clusters. The motionlet clusters are partitioned from a dataset of training vehicle images as a function of determining that vehicles within each of the scenes of the images in each cluster share similar two-dimensional motion direction attributes within their scenes. To train the complementary detectors, a first detector is trained on motion blobs of vehicle objects detected and collected within each of the training dataset vehicle images within the motionlet cluster via a background modeling process; a second detector is trained on each of the training dataset vehicle images within the motionlet cluster that have motion blobs of the vehicle objects but are misclassified by the first detector; and the training repeats until all of the training dataset vehicle images have been eliminated as false positives or correctly classified.

Abstract: Apparatus and method for performing accurate text recognition of non-simplistic images (e.g., images with clutter backgrounds, lighting variations, font variations, non-standard perspectives, and the like) may employ a machine-learning approach to identify a discriminative feature set selected from among features computed for a plurality of irregularly positioned, sized, and/or shaped (e.g., randomly selected) image sub-regions.

Abstract: To improve feature selection accuracy during a visual search, interest points within a query image are two-way matched to features in an affine transformed image or otherwise transformed version of the query image. A user device implements a method for selecting local descriptors in the visual search. The method includes: detecting a first set of interest points for the original image; computing an affine transform matrix; computing a new image as a transformation of the original image using the affine transform matrix; detecting a second set of interest points from the and new image; performing a two-way matching between the first set of interest points and the second set of interest points; sorting matching pairs according to a specified self-matching score (SMS); assigning an infinite value to SMS of unmatched interest points from the original image; selecting the interest points based on SMS. Significant performance gains reduce false positive matches.

Abstract: An object detector includes a bottom-up object hypotheses generation unit; a top-down object search with supervised descent unit; and an object re-localization unit with a localization model.

Abstract: An image processing apparatus including: a processor; and a memory storing instructions that, when executed by the processor, cause the apparatus to perform: acquiring target image data and template image data; specifying a partial image in a target image based on a result of detecting a face area; and compositing the specified partial image in a specific area in a template image, wherein, when a single face area is detected, a first partial image including the single face area and having a size determined based on a size of the single face area is specified, and, when a plurality of face areas are detected, a second partial image including at least one of the face areas, having a shape homothetic to the specific area and having a maximum size within the target image is specified.

Abstract: Techniques described here use variations in the sensor to generate an identifier for the sensor. Each sensor may be comprised of sub-sensing units, called pixels that may demonstrate variation in their sensing capability from one pixel to another. Embodiments of the invention, describe a method for using the relative variance of each pixel (relative to the whole sensor or/and a portion of the sensor) in generating an identifier for the sensor. In one embodiment, the method may obtain information associated with a plurality of pixels from a sensor, detect variations in the information associated for each of the pixels from a subset of the plurality of pixels and generate an identifier for the sensor using the detected variations in the information associated with each of the pixels from the subset of plurality of pixels.

Abstract: A system estimates text orientation in images captured using a handheld camera prior detecting text in the image. Text orientation is estimated based on edges detected within the image, and the image is rotated based on the estimated orientation. Text detection and processing is then performed on the rotated image. Non-text features along a periphery of the image may be sampled to assure that clutter will not undermine the estimation of orientation.

Abstract: Foreground object image features are extracted from input video via application of a background subtraction mask, and optical flow image features from a region of the input video image data defined by the extracted foreground object image features. If estimated movement features indicate that the underlying object is in motion, a dominant moving direction of the underlying object is determined. If the dominant moving direction is parallel to an orientation of the second, crossed thoroughfare, an event alarm indicating that a static object is blocking travel on the crossing second thoroughfare is not generated. If the estimated movement features indicate that the underlying object is static, or that its determined dominant moving direction is not parallel to the second thoroughfare, an appearance of the foreground object region is determined and a static-ness timer run while the foreground object region comprises the extracted foreground object image features.

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: System and method for determining a classifier to discriminate between two classes—object or non-object. The classifier may be used by an object detection program to detect presence of a 3D object in a 2D image. The overall classifier is constructed of a sequence of classifiers, where each such classifier is based on a ratio of two graphical probability models. A discreet-valued variable representation at each node in a Bayesian network by a two-stage process of tree-structured vector quantization is discussed. The overall classifier may be part of an object detector program that is trained to automatically detect different types of 3D objects. Computationally efficient statistical methods to evaluate overall classifiers are disclosed. The Bayesian network-based classifier may also be used to determine if two observations belong to the same category.

Abstract: In accordance with one aspect of this invention, a pattern inspection apparatus includes an optical image acquisition unit configured to acquire optical images regarding dies of a target object to be inspected on which the dies having a same pattern formed therein is arranged; a sub-optical image division unit configured to divide an optical image of the optical images regarding a die of the dies positioned in a non-resolved pattern region into sub-optical images using non-resolved pattern region information capable of recognizing the non-resolved pattern region in which a non-resolved pattern that is not resolved is formed; a first comparison unit configured to compare the sub-optical images divided from the optical image of the same die regarding the non-resolved pattern region pixel by pixel; and a second comparison unit configured to compare optical images of the optical images regarding different dies of the dies pixel by pixel.

Abstract: Systems and methods are disclosed for detecting an object in an image by determining convolutional neural network responses on the image; mapping the responses back to their spatial locations in the image; and constructing features densely extract shift invariant activations of a convolutional neural network to produce dense features for the image.

Abstract: The overlay error of a target in a scribelane is measured. The overlay error of the required feature in the chip area may differ from this due to, for example, different responses to the exposure process. A model is used to simulate these differences and thus a more accurate measurement of the overlay error of the feature determined.

Abstract: A method is disclosed for correcting errors introduced by optical distortions or aberrations in the measured values of coordinates of targets of an array of targets, like for example structures on a wafer or a photolithography mask. The array of targets is placed into a field of view of an imaging system via which the coordinates are to be measured. The array of targets is placed at different relative positions with respect to the field of view, and for each relative position the coordinates of the targets relative to the array of targets are determined by measurements. From the results of these measurements an alignment function, giving a correction for optical errors as a function of the position in the field of view, is derived. The measured coordinates are corrected by the alignment function. The corrected coordinates can be used to identify registration errors of a mask writer.

Abstract: A method of generating a feature vector for an image is disclosed. Values are determined from a plurality of points in a region of the image, each of the values being determined using at least two of the plurality of points. A periodic sequence of the determined values is determined based on an order of the plurality of points. The periodic sequence is phase variant to a starting point of the ordered plurality of points, the order of the plurality of points being determined according to a predetermined rule. The feature vector for one of the points is generated from a frequency domain representation of the periodic sequence, the feature vector being invariant to rotation with respect to the plurality of points.

Abstract: A document registration apparatus includes a receiving unit that receives a request for registration of a registration candidate document from a new registrant, a word extracting unit that extracts a word from the registration candidate document, a registrant information acquiring unit that acquires information on the new registrant, an associating unit that associates the extracted word with a group to which the new registrant belongs, a first storage unit that stores history information, a second storage unit that stores an identifier of a previous registrant and a group to which the previous registrant has belonged, an extracting unit that extracts an identifier of a previous registrant who registered a word identical to the extracted word, and extracts a group to which the previous registrant has belonged, a registration permission determining unit that determines whether to allow registration, and a document registering unit that registers the registration candidate document.

Abstract: Systems and methods for applying a vector texture to free-form drawing, writing, etc., and more particularly, for rendering a vector texture to touch-based, free-form drawing, writing, etc.

Abstract: A method for performing three-dimensional (3D) localization requiring only a single camera including capturing images from only one camera; generating a cue combination from sparse features, dense stereo and object bounding boxes; correcting for scale in monocular structure from motion (SFM) using the cue combination for estimating a ground plane; and performing localization by combining SFM, ground plane and object bounding boxes to produce a 3D object localization.