Abstract: In one embodiment, there is provided a method for an Optical Character Recognition (OCR) system. The method comprises: recognizing an input character based on a plurality of classifiers, wherein each classifier generates an output by comparing the input character with a plurality of trained patterns; grouping the plurality of classifiers based on a classifier grouping criterion; and combining the output of each of the plurality of classifiers based on the grouping.

Abstract: A method, system and computer program product for detecting presence of an object in an image are disclosed. According to an embodiment, a method for detecting a presence of an object in an image comprises: receiving multiple training image samples; determining a set of adaptive features for each training image sample, the set of adaptive features matching the local structure of each training image sample; integrating the sets of adaptive features of the multiple training image samples to generate an adaptive feature pool; determining a general feature based on the adaptive feature pool; and examining the image using a classifier determined based on the general feature to detect the presence of the object.

Abstract: A classifier training system trains unified classifiers for categorizing videos representing different categories of a category graph. The unified classifiers unify the outputs of a number of separate initial classifiers trained from disparate subsets of a training set of media items. The training process divides the training set into a number of bags, and applies a boosting algorithm to the bags, thus enhancing the accuracy of the unified classifiers.

Abstract: A method for a computer system for inspecting an animation sequence having a series of specified poses of an animated character includes receiving a mathematical performance style model associated with the animated character, receiving the series of specified poses of the animated character, each specified pose comprising respective values for the plurality of animation variables, determining an associated quality factor for each specified pose in response to the respective values for the plurality of animation variables and in response to the mathematical performance style model associated with the animated character, and providing feedback to a user in response to an associated quality factor for at least one specified pose.

Abstract: A method for assessing events detected by a surveillance system includes assessing the likelihood that the events correspond to events being monitored from feedback in response to a condition set by a user. Classifiers are created for the events from the feedback. The classifiers are applied to allow the surveillance system improve its accuracy when processing new video data.

Abstract: An information processing apparatus includes: a recognition section which recognizes the shape of an object being in contact with an operation screen of an operating section; a pressure detecting section which detects the pressure of the object on the operation screen; a threshold value setting section which sets a threshold value of the pressure, which is a value for determining a pressure operation on the operation screen, on the basis of the shape of the object recognized by the recognition section; and a determination section which determines whether or not a pressure operation has been performed on the operation screen on the basis of the pressure detected by the pressure detecting section and the threshold value set by the threshold value setting section.

Abstract: An approach is described for automatically tagging a single image or multiple images. The approach, in one example embodiment, is based on a graph-based framework that exploits both visual similarity between images and tag correlation within individual images. The problem is formulated in the context of semi-supervised learning, where a graph modeled as a Gaussian Markov Random Field (MRF) is solved by minimizing an objective function (the image tag score function) using an iterative approach. The iterative approach, in one embodiment, comprises: (1) fixing tags and propagating image tag likelihood values from labeled images to unlabeled images, and (2) fixing images and propagating image tag likelihood based on tag correlation.

Abstract: Embodiments include a scene classification system and method. In one embodiment, a method includes forming a first plurality of image features from an input image, processing the first plurality of image features in the first scene classifier.

Abstract: A background model learning system for lighting change adaptation for video surveillance is provided. The system includes a background model estimation unit that estimates a background model for a scene of interest; a foreground map construction unit that constructs a reference foreground map of the time instance; and a lighting change processing unit that revises the reference foreground map by reducing false foreground regions resulting from lighting changes. The revised foreground map is then sent back to both the background model estimation unit and the lighting change processing unit as feedbacks for model learning rate tuning in background model estimation and map integration in lighting change processing, respectively, for the next time instance.

Abstract: Systems and methods receive, from a building, an inventory identifying a plurality of electronic devices associated with a building automation system. The systems and methods compare the inventory to an additional inventory associated with an additional building automation system of an additional building to determine that the inventory is the same as or similar to the additional inventory, wherein the additional inventory identifies a plurality of additional electronic devices. The systems and methods further identify an automation scene that coordinates operation of at least a portion of the plurality of additional electronic devices and provide the automation scene to the building for implementation in the building automation system.

Abstract: An apparatus includes: an input configured to receive information indicative of sensed light locations; memory coupled to the input and storing indicia of receptive fields forming a mosaic, each of the receptive fields corresponding to an electrode, the mosaic including first and receptive fields having first and second shapes that are different, the memory further storing instructions; a processor coupled to the input and the memory and configured to read and execute the instructions to: analyze the information indicative of sensed light locations; determine, for each of respective ones of the sensed light locations, one or more receptive fields that include the corresponding sensed light location; and produce excitation indicia; the apparatus further including an output coupled to the processor and configured to be coupled to a retinal implant and to convey the excitation indicia toward the retinal implant.

Abstract: Disclosed are methods and apparatus for automatic optoelectronic detection and inspection of objects, based on capturing digital images of a two-dimensional field of view in which an object to be detected or inspected may be located, analyzing the images, and making and reporting decisions on the status of the object. Decisions are based on evidence obtained from a plurality of images for which the object is located in the field of view, generally corresponding to a plurality of viewing perspectives. Evidence that an object is located in the field of view is used for detection, and evidence that the object satisfies appropriate inspection criteria is used for inspection. Methods and apparatus are disclosed for capturing and analyzing images at high speed so that multiple viewing perspectives can be obtained for objects in continuous motion.

Abstract: A method of detecting a predefined set of characteristic points of a face from an image of the face includes a step of making the shape and/or the texture of a hierarchy of statistical models of face parts converge over real data supplied by the image of the face.

Abstract: Methods and system employing the same for finding repeated structure for data extraction from document images are provided. A reference record and one or more reference fields thereof are identified from a document image. One or more candidate fields are generated for each of the reference fields. One or more best candidate records from the candidate fields are selected using a probabilistic model and an optimal record set is determined from the best candidate records.

Abstract: Techniques are disclosed for visually conveying an event map. The event map may represent information learned by a surveillance system. A request may be received to view the event map for a specified scene. The event map may be generated, including a background model of the specified scene and at least one cluster providing a statistical distribution of an event in the specified scene. Each statistical distribution may be derived from data streams generated from a sequence of video frames depicting the specified scene captured by a video camera. Each event may be observed to occur at a location in the specified scene corresponding to a location of the respective cluster in the event map. The event map may be configured to allow a user to view and/or modify properties associated with each cluster. For example, the user may label a cluster and set events matching the cluster to always (or never) generate an alert.

Abstract: Systems and methods for automated pattern recognition and object detection. The method can be rapidly developed and improved using a minimal number of algorithms for the data content to fully discriminate details in the data, while reducing the need for human analysis. The system includes a data analysis system that recognizes patterns and detects objects in data without requiring adaptation of the system to a particular application, environment, or data content. The system evaluates the data in its native form independent of the form of presentation or the form of the post-processed data.

Abstract: Embodiments of the present invention provide a method and a system for analyzing and learning behavior based on an acquired stream of video frames. Objects depicted in the stream are determined based on an analysis of the video frames. Each object may have a corresponding search model used to track an object's motion frame-to-frame. Classes of the objects are determined and semantic representations of the objects are generated. The semantic representations are used to determine objects' behaviors and to learn about behaviors occurring in an environment depicted by the acquired video streams. This way, the system learns rapidly and in real-time normal and abnormal behaviors for any environment by analyzing movements or activities or absence of such in the environment and identifies and predicts abnormal and suspicious behavior based on what has been learned.

Abstract: Via intuitive interactions with a user, robots may be trained to perform tasks such as visually detecting and identifying physical objects and/or manipulating objects. In some embodiments, training is facilitated by the robot's simulation of task-execution using augmented-reality techniques.

Abstract: An automatic handwriting morphing and modification system and method for digitally altering the handwriting of a user while maintaining the overall appearance and style of the user's handwriting. Embodiments of the system and method do not substitute or replace characters or words but instead morph and modify the user's handwritten strokes to retain a visual correlation between the original user's handwriting and the morphed and modified version of the user's handwriting. Embodiments of the system and method input the user's handwriting and a set of morph rules that determine what the handwritten strokes of the user can look more like after processing. Morphs, which are a specific type or appearance of a handwritten stroke, are selected based on the target handwriting. The selected morphs are applied using geometric tuning, semantic tuning, or both. The result is a morphed and modified version of the user's handwriting.

Abstract: Techniques for creating a training technique for an individual are provided. The techniques include obtaining video of one or more events and information from a transaction log that corresponds to the one or more events, wherein the one or more events relate to one or more actions of an individual, classifying the one or more events into one or more event categories, comparing the one or more classified events with an enterprise best practices model to determine a degree of compliance, examining the one or more classified events to correct one or more misclassifications, if any, and revise the one or more event categories with the one or more corrected misclassifications, if any, and using the degree of compliance to create a training technique for the individual.

Abstract: Techniques are described herein for generating and displaying a confusion matrix wherein a data item belonging to one or more actual classes is predicted into a class. The classes in which the data item may be predicted (the “predicted classes”) are ranked according to a score that in one embodiment indicates the confidence of the prediction. If the data item is predicted into a class that is one of the top K ranked predicted classes, then the prediction is considered accurate and an entry is created in a cell of a confusion matrix indicating the accurate prediction. If the data item is not predicted into a class that is not one of the top K ranked predicted classes, then the prediction is considered inaccurate and an entry is created in a cell of a confusion matrix indicating the inaccurate prediction.

Abstract: An eigenprojection matrix (#14) is generated by a projection computation (#12) using a local relationship from a studying image group (#10) including a pair of a high quality image and a low quality image to create a projection core tensor (#16) defining a correspondence between the low quality image and an intermediate eigenspace and a correspondence between the high quality image and the intermediate eigenspace. A first sub-core tensor is created (#24) from the projection core tensor based on a first setting, and an inputted low quality image (#20) is projected (#30) based on the eigenprojection matrix and the first sub-core tensor to calculate a coefficient vector in the intermediate eigenspace. The coefficient vector is projected (#34) based on a second sub-core tensor (#26) created by a second setting from the projection core tensor and based on the eigenprojection matrix to obtain a high quality image (#36).

Abstract: Techniques are disclosed relating to automatically adjusting images. In one embodiment, an image may be automatically adjusted based on a regression model trained with a database of raw and adjusted images. In one embodiment, an image may be automatically adjusted based on a model trained by both a database of raw and adjusted images and a small set of images adjusted by a different user. In one embodiment, an image may be automatically adjusted based on a model trained by a database of raw and adjusted images and predicted differences between a user's adjustment to a small set of images and a predicted adjustment based on the database of raw and adjusted images.

Abstract: Systems and methods are provided for control of a personal computing device based on user face detection and recognition techniques.

Abstract: Techniques are disclosed relating to modifying an automatically predicted adjustment. In one embodiment, the automatically predicted adjustment may be adjusted, for example, based on a rule. The automatically predicted adjustment may be based on a machine learning prediction. A new image may be globally adjusted based on the modified automatically predicted adjustment.

Abstract: An image-based biomarker is generated using image features obtained through object-oriented image analysis of medical images. The values of a first subset of image features are measured and weighted. The weighted values of the image features are summed to calculate the magnitude of a first image-based biomarker. The magnitude of the biomarker for each patient is correlated with a clinical endpoint, such as a survival time, that was observed for the patient whose medical images were analyzed. The correlation is displayed on a graphical user interface as a scatter plot. A second subset of image features is selected that belong to a second image-based biomarker such that the magnitudes of the second image-based biomarker for the patients better correlate with the clinical endpoints observed for those patients. The second biomarker can then be used to predict the clinical endpoint of other patients whose clinical endpoints have not yet been observed.

Abstract: A method and apparatus that classify an image. The method extracts a feature vector from the image, wherein the feature vector includes a plurality of first features. The extracting of each of the first features includes: acquiring a difference between sums or mean values of pixels of the plurality of first areas in the corresponding combination to obtain a first difference vector in the direction of the first axis, and obtaining a second difference vector in the direction of the second axis. A first projection difference vector is acquired with a second projection difference vector. A sum of magnitudes of the first projection difference vector and the second projection difference vector as the first feature is obtained; and the image according to the extracted feature vector is classified.

Abstract: Described herein is a technology for facilitating deformable model-based segmentation of image data. In one implementation, the technology includes receiving training image data (202) and automatically constructing a hierarchical structure (204) based on the training image data. At least one spatially adaptive boundary detector is learned based on a node of the hierarchical structure (206).

Abstract: In an image processing device and method, program, and recording medium of the present invention, high frequency components of a low quality image and a high quality image included in a studying image set are extracted, and an eigenprojection matrix and a projection core tensor of the high frequency components are generated in a studying step. In a restoration step, a first sub-core tensor and a second sub-core tensor are generated based on the eigenprojection matrix and the projection core tensor of the high frequency components, and a tensor projection process is applied to the high frequency components of an input image to generate a high quality image of the high frequency components. The high quality image of the high frequency components is added to an enlarged image obtained by enlarging the input image to the same size as an output image.

Abstract: A tentative eigenprojection matrix (#12-a) is provisionally generated from a studying image group (#10) including a pair of a high quality image and a low quality image, and a tentative projection core tensor (#12-b) defining a correspondence between the low quality image and an intermediate eigenspace and a correspondence between the high quality image and the intermediate eigenspace is created. A first sub-core tensor is created from the tentative projection core tensor based on a first setting (#12-c), and the studying image group is projected based on the tentative eigenprojection matrix and the first sub-core tensor (#15-a) to calculate an intermediate eigenspace coefficient vector (#15-b).

Abstract: Described are methods and apparatuses, including computer program products, for determining model uniqueness with a quality metric of a model of an object in a machine vision application. Determining uniqueness involves receiving a training image, generating a model of an object based on the training image, generating a modified training image based on the training image, determining a set of poses that represent possible instances of the model in the modified training image, and computing a quality metric of the model based on an evaluation of the set of poses with respect to the modified training image.

Abstract: A technique for determining a characteristic of a face or certain other object within a scene captured in a digital image including acquiring an image and applying a linear texture model that is constructed based on a training data set and that includes a class of objects including a first subset of model components that exhibit a dependency on directional lighting variations and a second subset of model components which are independent of directional lighting variations. A fit of the model to the face or certain other object is obtained including adjusting one or more individual values of one or more of the model components of the linear texture model. Based on the obtained fit of the model to the face or certain other object in the scene, a characteristic of the face or certain other object is determined.

Abstract: In the image acquisition mode, desired past images are registered in an image table. A past image selected from the past images registered in the table is displayed as a reference image together with a live image in a predetermined form. Selecting another image registered in the image table at an arbitrary timing will display another past image upon replacing the reference image.

Abstract: A method for interpolation includes receiving an input image having a plurality of pixels. The edge direction proximate a first pixel of the input image is estimated using a first technique from a plurality of discrete potential directions. An edge direction is selected based upon the estimating the edge direction proximate the first pixel of the input image using a second technique. The pixels proximate the first pixel are interpolated based upon the selected edge direction. The pixels proximate the first pixel are interpolated based upon another technique. An output image is determined pixels having more pixels than the plurality of pixels.

Abstract: The Target Separation Algorithms (TSAs) are used to improve the results of Automated Target Recognition (ATR). The task of the TSAs is to separate two or more closely spaced targets in Regions of Interest (ROIs), to separate targets from objects like trees, buildings, etc., in a ROI, or to separate targets from clutter and shadows. The outputs of the TSA separations are inputs to ATR, which identify the type of target based on a template database. TSA may include eight algorithms. These algorithms may use average signal magnitude, support vector machines, rotating lines, and topological grids for target separation in ROI. TSA algorithms can be applied together or separately in different combinations depending on case complexity, required accuracy, and time of computation.

Abstract: A dictionary of atoms for coding data is learned by first selecting samples from a set of samples. Similar atoms in the dictionary are clustered, and if a cluster has multiple atoms, the atoms in that cluster are merged into a single atom. The samples can be acquired online.

Abstract: In a method for determining a parameter in an automatic study and data management system, data is gathered in a knowledge database, and a parameter is determined based the data gathered in the knowledge database. The data is correlated to at least one of a configuration and implementation of a previous clinical study. The parameter is usable for configuring a future clinical study.

Abstract: In a pattern identification method in which input data is classified into predetermined classes by sequentially executing a combination of a plurality of classification processes, at least one of the classification processes includes a mapping step of mapping the input data in an N (N?2) dimensional feature space as corresponding points, a determination step of determining whether or not to execute the next classification process based on the corresponding points, and selecting step of selecting a classification process to be executed next based on the corresponding points when it is determined in the determination step that the next classification process should be executed.

Abstract: The present invention discloses a system and method of transforming a sample of content data by utilizing known samples in a learning phase to best determine coefficients for a linear combination of non-linear filter functions and applying the coefficients to the content data in an operational phase.

Abstract: A multi-stage method of visual object detection is disclosed. The method was originally designed to detect humans in specific poses, but is applicable to generic detection of any object. A first stage comprises acts of searching for members of a predetermined general-class of objects (such as humans) in an image using a cognitive swarm, detecting members of the general-class of objects in the image, and selecting regions of the image containing detected members of the general-class of objects. A second stage comprises acts of searching for members of a predetermined specific-class of objects (such as humans in a certain pose) within the selected regions of the image using a cognitive swarm, detecting members of the specific-class of objects within the selected regions of the image, and outputting the locations of detected objects to an operator display and optionally to an automatic response system.

Abstract: Methods, systems, and apparatus, including computer program products, for identifying images relevant to a query are disclosed. An image search subsystem selects images to reference in image search results that are responsive to a query based on an image relevance model that is trained for the query. An independent image relevance model is trained for each unique query that is identified by the image search subsystem. The image relevance models can be applied to images to order image search results obtained for the query. Each relevance model is trained based on content feature values of images that are identified as being relevant to the query (e.g., frequently selected from the image search results) and images that are identified as being relevant to another unique query. The trained model is applied to the content feature values of all known images to generate an image relevance score that can be used to order search results for the query.

Abstract: Machine-readable media, methods, apparatus and system for obtaining and processing image features are described. In some embodiments, a Gabor representation of an image may be obtained by using a Gabor filter. A region may be determined from the Gabor representation, wherein the region comprises a plurality of Gabor pixels of the Gabor representation; and, a sub-region may be determined from the region, wherein the sub-region comprises more than one of the plurality of Gabor pixels. Then, a Gabor feature may be calculated based upon a magnitude calculation related to the sub-region and the region.

Abstract: A technique for determining a characteristic of a face or certain other object within a scene captured in a digital image including acquiring an image and applying a linear texture model that is constructed based on a training data set and that includes a class of objects including a first subset of model components that exhibit a dependency on directional lighting variations and a second subset of model components which are independent of directional lighting variations. A fit of the model to the face or certain other object is obtained including adjusting one or more individual values of one or more of the model components of the linear texture model. Based on the obtained fit of the model to the face or certain other object in the scene, a characteristic of the face or certain other object is determined.

Abstract: A system and method for creating one of a plurality of test decks to qualify and test forms processing systems, including preparing a handprint snippet data base containing labeled handprint image snippets representing a unique hand, preparing a form description file and a data content file, selecting handprint snippets from the handprint snippet data base to formulate a form using the data content file, creating a form image using the selected snippets according to the form description file and printing the form image.

Abstract: Data set generation and data set presentation for image processing are described. The processing determines a location for each of one or more musical artifacts in the image and identifies a corresponding label for each of the musical artifacts, generating a training file that associates the identified labels and determined locations of the musical artifacts with the image, and presenting the training file to a neural network for training.

Abstract: A system for detecting faint perturbations of interest in an image is described. The system includes a memory comprising the image, and a processor. The processor is configured to estimate a local mean and variance for each of a plurality of pixels in the image, analyze a local region of the image, an image mean, and an image variance using a filter bank, thereby generating a plurality of response vectors, determine a likelihood of a local perturbation in the image using a probability distribution based on the plurality of response vectors, and make a classification decision of the local perturbation. The probability distribution is calculated, for each of the plurality of response vectors, based on a response of each of the plurality of response vectors to the estimated local mean and variance for each of the plurality of pixels. Methods and machine-readable media are also described.

Abstract: An image classification system trains an image classification model to classify images relative to text appearing with the images. Training images are iteratively selected and classified by the image classification model according to feature vectors of the training images. An independent model is trained for unique n-grams of text. The image classification system obtains text appearing with an image and parses the text into candidate labels for the image. The image classification system determines whether an image classification model has been trained for the candidate labels. When an image classification model corresponding to a candidate label has been trained, the image classification subsystem classifies the image relative to the candidate label. The image is labeled based on candidate labels for which the image is classified as a positive image.

Abstract: Embodiments of methods, apparatuses, devices and systems associated with one or more representative images are disclosed.

Abstract: The restoration of images by vector quantization utilizing visual patterns is disclosed. One disclosed embodiment comprises restoring detail in a transition region of an unrestored image, by first identifying the transition region and forming blurred visual pattern blocks. These blurred visual pattern blocks are compared to a pre-trained codebook, and a corresponding high-quality visual pattern blocks is obtained. The high-quality visual pattern block is then blended with the unrestored image to form a restored image.

Abstract: Techniques are disclosed relating to automatically adjusting images. In one embodiment, an image may be automatically adjusted based on a regression model trained with a database of raw and adjusted images. In one embodiment, an image may be automatically adjusted based on a model trained by both a database of raw and adjusted images and a small set of images adjusted by a different user. In one embodiment, an image may be automatically adjusted based on a model trained by a database of raw and adjusted images and predicted differences between a user's adjustment to a small set of images and a predicted adjustment based on the database of raw and adjusted images.