Abstract: An image processing apparatus includes a corresponding region connecting unit that connects regions that depict the same target between a series of images captured in time series, thereby sets at least one connected region; a connected region feature data calculation unit that calculates feature data of the connected region; a digest index value calculation unit that calculates a digest index value corresponding to a degree at which the target depicted in the series of images is aggregated in each image of the series of images, based on the feature data; and a digest image detector that detects a digest image based on the digest index value.

Abstract: Disclosed are a system, apparatus, and method for detecting objects. Input image frames may be received and distinct regions created within each frame. Descriptors may be extracted from the regions according to their associated probability. Extracted descriptors may be matched to reference descriptors. Votes or confidence is cast for particular regions according to region properties. The region properties may be determined from center voting methods based on vector intersection to other vectors or intersections with a region. The probability of selecting particular regions can increase with each vote or increase in confidence for a region. In response to updating probabilities, additional regions may be selected and additional descriptors may be extracted. Additional voting iterations can update the probability of selecting a next region. An object pose may be estimated in response to meeting one or more thresholds.

Abstract: A computer-implemented apparatus and method for performing patent portfolio analysis. The patent portfolio analysis apparatus and method clusters a group of patents based upon one or more techniques. The clustering techniques include linguistic clustering techniques (e.g., eigenvector analysis), claim meaning, and patent classification techniques. Different aspects of the clusters are analyzed, including financial, claim breadth, and assignee patent comparisons. Moreover, patents and/or their clusters are linked to the Internet in order to determine what products might be covered by the claims of the patents or whether materials on the Internet might render patent claims invalid.

Abstract: A system and method for calibrating a dynamic digital imaging system for the detection of defects in a moving product stream. The system has an elevated platform above a conveying unit for receiving a reference color tile. The elevated platform allows for the passage of products to be inspected on the below conveying unit surface such that calibration and re-calibration processes during image capturing may be accomplished on a continuous basis without interruption of the product stream.

Abstract: Provided are an apparatus for processing a medical image, and a method of processing a medical image by using the apparatus. The apparatus includes: an image obtainer that receives a first image obtained by photographing an object; and an image processor that calculates a scaling factor for adjusting a magnification of the first image based on statistic information corresponding to the object and generates a second image to be displayed by normalizing a size of the object by applying the calculated scaling factor to the first image.

Abstract: The method and apparatus segment parts from a main target image, represent each part as a vertex in a spanning tree, use a detector to generate a confidence map of a location of each part in a succeeding video frame and apply scale change to detector sliding windows centered about each pixel in the part location image. In the succeeding video frame, the target location is sampled and a tracking probability is generated for each part bounding box, with the tracking probability having the maximum value being selected as the location of the target in the succeeding video frame.

Abstract: An analyzing unit sets a threshold on the basis of a target compression ratio of a frame. The compression unit compresses a block having an attribute value larger than or equal to a threshold by applying a first compression format and a block having an attribute value smaller than the threshold by applying a second compression format having a higher compression ratio than the first compression format.

Abstract: A method for parsing the text of a complex graphical image comprises obtaining a series of blocks of text from a complex graphical image. Those blocks of text are used to generate location scores, size scores and length scores. Each of the scores is weighted and linearly summed. The highest resulting sum is identified as the most likely to be a desired text block.

Abstract: The present disclosure relates to systems and methods for classifying videos based on video content. For a given video file including a plurality of frames, a subset of frames is extracted for processing. Frames that are too dark, blurry, or otherwise poor classification candidates are discarded from the subset. Generally, material classification scores that describe type of material content likely included in each frame are calculated for the remaining frames in the subset. The material classification scores are used to generate material arrangement vectors that represent the spatial arrangement of material content in each frame. The material arrangement vectors are subsequently classified to generate a scene classification score vector for each frame. The scene classification results are averaged (or otherwise processed) across all frames in the subset to associate the video file with one or more predefined scene categories related to overall types of scene content of the video file.

Abstract: An image analysis method includes acquiring fluorescent images of frames in time-series. Each fluorescent image comprises pixels in which pixel data are acquired in the time-series. The method further includes setting analysis areas to the acquired fluorescent images, calculating a classification value of the analysis areas, classifying the images into one or more groups on the basis of the classification value, calculating an average image of the analysis area every group, subtracting the average image from each image of the analysis area every group to calculate a new image of the analysis area, and calculating a correlation value on the basis of the new images every group.

Abstract: An image analysis method includes acquiring fluorescent images of frames in time-series. Each fluorescent image comprises pixels in which pixel data are acquired in the time-series. The method further includes setting analysis areas to the fluorescent images, selecting the fluorescent images of two or more frames to be used in analysis, extracting data pairs each comprising two pixels in which acquisition time intervals are the same in the analysis area of each of the selected fluorescent images, and performing product sum calculation of each of the data pairs for all of the selected images to calculate a correlation value.

Abstract: Techniques are provided for segmenting an input by cut point classification and training a cut classifier. A method may include receiving, by a computerized text recognition system, an input in a script. A heuristic may be applied to the input to insert multiple cut points. For each of the cut points, a probability may be generated and the probability may indicate a likelihood that the cut point is correct. Multiple segments of the input may be selected, and the segments may be defined by cut points having a probability over a threshold. Next, the segments of the input may be provided to a character recognizer. Additionally, a method may include training a cut classifier using a machine learning technique, based on multiple text training examples, to determine the correctness of a cut point in an input.

Abstract: Methods and systems for achieving accurate segmentation of characters with respect to a license plate image utilizing a reinforcement learning approach. A vehicle image can be captured by an image capturing unit and processed utilizing an ALPR (Automatic License Plate Recognition) unit. The reinforcement learning (RL) approach can be configured to initialize a segmentation agent with a starting location. A proper segmentation path (cuts) from top to bottom and from a darker to lighter area in a cropped license plate image can be identified by the segmentation agent during a training phase. Rewards can be provided based on a number of good and bad moves. The association between a current state and a sensory input with a preferred action can be learned by the segmentation agent at the end of the training phase.

Abstract: A structured query may specify attribute values for attributes. An estimate of the number of items that will match the structured query if it is applied to a structured database is determined. If the estimated number of items is below a threshold, the structured query may be relaxed to form new candidate structured queries. The number of candidate queries may be determined based on a desired running time. Each of the candidate structured queries may be determined by changing one or more attribute values of the attributes of the structured query. Estimates of the number of items each of the candidate structured queries will match is determined, and the candidate structured query that has the highest matching estimation is used to query the database. The matching results may be output.

Abstract: A method and system for identifying one or more features represented in a plurality of sensor acquired data sets is described. The method and apparatus is particularly useful in automatic license plate recognition applications, where the sensor acquired data sets are data obtained from one or more digital cameras. This is achieved by determining a first probability of the identity of the one or more features (e.g., alphanumeric characters) from a first one of the data sets; determining a second probability of the identity of the one or more features from a second one of the data sets; and, using data fusion techniques, fusing the determined first and second probabilities to provide a fused probability. This fused probability is used to identify the one or more features from data sets.

Abstract: Provided are an age estimation apparatus, an age estimation method, and an age estimation program capable of obtaining a recognition result closely matching the result perceived by human. An age estimation apparatus 10 for estimating an age of a person on image data includes a dimension compressor 11 for applying dimension compression to the image data to output low dimensional data; and an identification device 12 for estimating an age of a person on the basis of a learning result using a feature amount contained in the low dimensional data, wherein a parameter used for the dimension compression by the dimension compressor 11 and the feature amount used for age estimation by the identification device 12 are set on the basis of a result of an evaluation of a generalization capability using a weighting function that shows a degree of seriousness of an age estimation error for every age, and learning of the identification device 12 is performed on the basis of the weighting function.

Abstract: An apparatus for providing pattern detection may include a processor. The processor may be configured to iteratively test different models and corresponding scales for each of the models. The models may be employed for modeling parameters corresponding to a visually detected data. The processor may be further configured to evaluate each of the models over a plurality of iterations based on a function evaluation of each of the models, select one of the models based on the function evaluation of the selected one of the models, and utilize the selected one of the models for fitting the data.

Abstract: An apparatus includes a first acquisition unit configured to acquire main object information specifying a main object in generation of a layout image, a second acquisition unit configured to acquire object correlation information specifying an object having a correlation with the main object, an extraction unit configured to extract at least one image including the main object and at least one image including the object having the correlation with the main object from a plurality of images based on the acquired main object information and the acquired object correlation information acquired, and a generation unit configured to generate, using a layout template, a layout image in which the at least one image extracted by the extraction unit and including the main object and the at least one image extracted by the extraction unit and including the object having the correlation with the main object are laid out therein.

Abstract: Integrating features is disclosed, including: determining a value associated with a temporal feature for a point; determining a value associated with a spatial feature associated with the temporal feature; including the value associated with a spatial feature and the value associated with the temporal feature into a feature vector; and using the feature vector to decode for a character. Determining a transform is also disclosed, including: determining, for a point associated with a sequence of points, a set of points including: the point, a first subset of points of the sequence preceding a sequence position associated with the point, and a second subset of points following the sequence position associated with the point; and determining the transform associated with the point based at least in part on the set of points.

Abstract: A hierarchy machine may be configured as a clustering machine that utilizes local feature embedding to organize visual patterns into nodes that each represent one or more visual patterns. These nodes may be arranged as a hierarchy in which a node may have a parent-child relationship with one or more other nodes. The hierarchy machine may implement a node splitting and tree-learning algorithm that includes hard-splitting of nodes and soft-assignment of nodes to perform error-bounded splitting of nodes into clusters. This may enable the hierarchy machine, which may form all or part of a visual pattern recognition system, to perform large-scale visual pattern recognition, such as font recognition or facial recognition, based on a learned error-bounded tree of visual patterns.

Abstract: A system includes preparing respective proof reading tools for performing carpet proof reading and side-by-side proof reading of text data, recording a log of time to perform proof reading operations by using the first and second proof reading tools. The method further includes estimating, based on times stored in a log, times to perform proof reading of a character using 1) the first proof reading tool followed by using the second proof reading tool, and 2) the second proof reading tool. The method further includes determining for each character value, based on the estimated times, to use the first proof reading tool along with using the second proof reading tool and determining, or to use the second proof reading tool without using the first proof reading tool.

Abstract: Techniques for separating shareable images from non-shareable images. In various implementations, image metadata and feature analysis may be used to evaluate the “shareability” of a photograph associated with a particular user. In some implementations, single photos may be determined to be shareable. In another implementation, an event associated with multiple photos may be determined to be shareable. In some implementations, a photo may be determined to be shareable with a single recipient. In another implementation, a photo may be determined to be shareable with multiple recipients. In yet another implementation, these techniques may be assisted by supervised machine learning. In still yet another implementation, photos determined to be shareable may be suggested to a user for sharing, or automatically shared, per an opt-in feature.

Abstract: A method includes preparing respective proof reading tools for performing carpet proof reading and side-by-side proof reading of text data, recording a log of time to perform proof reading operations by using the first and second proof reading tools. The method further includes estimating, based on times stored in a log, times to perform proof reading of a character using 1) the first proof reading tool followed by using the second proof reading tool, and 2) the second proof reading tool. The method further includes determining for each character value, based on the estimated times, to use the first proof reading tool along with using the second proof reading tool or to use the second proof reading tool without using the first proof reading tool.

Abstract: In various embodiments, methods and apparatus are provided for automated selection of features of cells useful for classifying cell phenotype. The methods include determining a signal-to-noise ratio (S/N) for each of a plurality of pairs of features, rather than S/N for individual features. The approach is capable of quickly identifying a small set of features of imaged cells that are most relevant for classification of a desired cell phenotype from among a very large number of features. The small group of relevant features can then be used to more efficiently and more accurately classify phenotype of unidentified cells.

Abstract: A method of detecting a face in an image includes performing face detection within a first window of the image at a first location. A confidence level is obtained from the face detection indicating a probability of the image including a face at or in the vicinity of the first location. Face detection is then performed within a second window at a second location, wherein the second location is determined based on the confidence level.

Abstract: Methods are disclosed for classifying different parts of a sample into respective classes based on an image stack that includes one or more images.

Abstract: One or more apparatus and method for adaptively detecting motion instability in video. In embodiments, video stabilization is predicated on adaptive detection of motion instability. Adaptive motion instability detection may entail determining an initial motion instability state associated with a plurality of video frames. Subsequent transitions of the instability state may be detected by comparing a first level of instability associated with a first plurality of the frames to a second level of instability associated with a second plurality of the frames. Image stabilization of received video frames may be toggled first based on the initial instability state, and thereafter based on detected changes in the instability state. Output video frames, which may be stabilized or non-stabilized, may then be stored to a memory. In certain embodiments, video motion instability is scored based on a probability distribution of video frame motion jitter values.

Abstract: Systems, apparatus and methods are described related to accelerated object detection filter using a video estimation module.

Abstract: An image recognition device that improves the accuracy of generic object recognition compared with conventional technologies by reducing the influence of the position, size, background clutter and the like of an object that is targeted to be recognized in the input image by the generic object recognition. The image recognition device performs a generic object recognition and includes: a segmenting unit configured to segment an input image into a plurality of regions in accordance with meanings extracted from content of the input image; a generating unit configured to compute feature data for each of the plurality of regions and generate feature data of the input image reflecting the computed feature data; and a checking unit configured to check whether or not a recognition-target object is present in the input image in accordance with the feature data of the input image.

Abstract: A system and method for estimating a set of landmarks for a large image ensemble employs only a small number of manually labeled images from the ensemble and avoids labor-intensive and error-prone object detection, tracking and alignment learning task limitations associated with manual image labeling techniques. A semi-supervised least squares congealing approach is employed to minimize an objective function defined on both labeled and unlabeled images. A shape model is learned on-line to constrain the landmark configuration. A partitioning strategy allows coarse-to-fine landmark estimation.

Abstract: An image editing application (or a blur classification module thereof) may automatically estimate a coherent defocus blur map from a single input image. The application may represent the blur spectrum as a differentiable function of radius r, and the optimal radius may be estimated by optimizing the likelihood function through a gradient descent algorithm. The application may generate the spectrum function over r through polynomial-based fitting. After fitting, the application may generate look-up tables to store values for the spectrum and for its first and second order derivatives, respectively. The use of these tables in the likelihood optimization process may significantly reduce the computational costs of a given blur estimation exercise. The application may minimize an energy function that includes a data term, a smoothness term, and a smoothness parameter that is adaptive to local image content. The output blur map may be used for image object depth estimation.

Abstract: In a technology of acquiring an object position by a visual cone intersection method, an object position estimation device which reduces estimation of incorrect position due to ghost image objects is provided, wherein an object candidate acquisition means estimates an object position by a visual cone intersection method and acquires a candidate object, an object candidate information storage means which stores information on overlap among the candidate objects for the visual cone of the object which is acquired based on the captured image, a degree of co-movement acquisition means acquires an index as a degree of co-movement which is calculated based on number of visual cones including other candidate objects which co-move with a movement of the candidate object and indicates interrelation among candidate objects for candidate objects, and a candidate object determination means identifies said candidate object as either an object or a non-object by a posteriori probability based on the acquired index and outp

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: An apparatus and method are provided for recognizing an emotion of an individual based on Action Units. The method includes receiving an input AU string including one or more AUs that represents a facial expression of an individual from an AU detector; matching the input AU string with each of a plurality of AU strings, wherein each of the plurality of AU strings includes a set of highly discriminative AUs, each representing an emotion; identifying an AU string from the plurality of AU strings that best matches the input AU string; and outputting an emotion label corresponding to the best matching AU string that indicates the emotion of the individual.

Abstract: An image segmentation method includes generating a hierarchy of regions by unsupervised segmentation of an input image. Each region is described with a respective region feature vector representative of the region. Hierarchical structures are identified, each including a parent region and its respective child regions in the hierarchy. Each hierarchical structure is described with a respective hierarchical feature vector that is based on the region feature vectors of the respective parent and child regions. The hierarchical structures are classified according to a set of predefined classes with a hierarchical classifier component that is trained with hierarchical feature vectors of hierarchical structures of training images. The training images have semantic regions labeled according to the set of predefined classes. The input image is segmented into a plurality of semantic regions based on the classification of the hierarchical structures and optionally also on classification of the individual regions.

Abstract: An apparatus is provided for classifying targets into a known-object group and an unknown-object group. The apparatus includes a speech/image data storage unit configured to store a spoken sound of a name of an object and an image of the object; a unit configured to calculate a speech confidence level of a speech for the name of the object with reference to a spoken sound of a name of a known object; a unit configured to calculate an image confidence level of an image of an object with respect to an image of a known object; and a unit configured to compare an evaluation value, which is obtained by combining the speech confidence level and image confidence level, with a threshold value, and classify a target object into an object group determined according to whether the spoken sound of the name and the image are known or unknown.

Abstract: Embodiments of the present invention include systems and methods for identifying an object in an image. In embodiments, object identification includes using smooth encoding from a tree structure to generate a feature from a descriptor. In embodiments, the smooth encoding may be performed by, having identifying a leaf node for a descriptor, moving up the tree voting structure a number of levels from the identified leaf node to a branch node to identify leaf nodes dependent from the branch node; and then, by determining a sparse code under a condition that a distance between the descriptor and centroids of the leaf nodes dependent from the branch node weighted by the sparse code is minimized, wherein each element of the sparse code representing a weight corresponding to leaf nodes.

Abstract: The invention provides a method, system, and program product for detecting an object in a digital image. In one embodiment, the invention includes: deriving an initial object indication mask based on pixel-wise differences between a first digital image and a second digital image, at least one of which includes the object; performing an edge finding operation on both the first and second digital images, wherein the edge finding operation includes marking added edges; generating a plurality of straight linear runs of pixels across an image containing the object, wherein each of the plurality of straight linear runs starts and ends on an added edge and is contained within the initial object indication mask; and forming a final object indication mask by retaining only pixels that are part of at least one of the plurality of straight linear runs.

Abstract: Evaluating coded video by identifying spatial and temporal errors in the reconstructed video. Spatial errors may be identified by evaluating the severity of each individual error and pooling closely located errors to determine whether the collection of errors is noticeable. Temporal errors may be identified by evaluating the quality of each frame and then pooling consecutive frames to determine whether the collection of frames has errors that may be noticeable if the errors persist within the sequence of consecutive frames. The quality of the frames may be analyzed with any conventional analysis metric. In an embodiment, the quality of the frames may be analyzed using a row-column metric. The calculated analysis metrics may be normalized to accurately compare different metric calculations.

Abstract: Accurately estimating the image quality of video is still difficult, especially when there is no reference video available. A method for determining a quality measure for a video image comprises selecting a measuring point, such as a MB, determining a context area around the measuring point, calculating a variance of pixel values in the context area, calculating a variance of pixel values in the measuring point, calculating a relation between the two variances of pixel values, and averaging said relations for a plurality of measuring points, wherein a quality measure for a video image is obtained.

Abstract: A method for calculating a depth map from an original matrix image, comprising the steps of: calculating a first matrix image corresponding to the original matrix image with a low resolution and in which the depth of field is similar to that of the original matrix image, calculating a second matrix image corresponding to the original matrix image with a low resolution, comprising a number of pixels similar to that of the first matrix image and in which the depth of field is greater than that of the original matrix image, implementing a DFD type three-dimensional reconstruction algorithm from the first and the second matrix images, outputting the depth map.

Abstract: A system for predicting animal activity is disclosed. The system comprises an imaging device to capture an image of a predetermined area and an environmental data sensor to detect one or more environmental factors within the predetermined area. The environmental data sensor collects random environmental data at least once during a predetermined period. A trigger is in signal communication with the imaging device and the environmental data sensor. When the trigger is activated, the imaging device captures the image and the environmental data sensor collects triggered environmental data. The trigger is responsive to the presence of wildlife. A storage unit stores the random environmental data, the image, and the triggered environmental data. The random environmental data and the triggered environmental data are provided to a statistical regression to determine a statistical probability algorithm. The statistical probability algorithm calculates a predicted activity index for wildlife in the predetermined area.

Abstract: Events may be determined based on an image and context data associated with the image. An event type associated with the image may be determined based on a concept of the image. A list of events may be retrieved from an event database based on the context data. The retrieved list of events may then be ranked based on the determined event type and the context data. Through this event determination, a user may obtain information of one or more events happening at a specific location simply by capturing an image of that specific location, thereby saving the user from searching and browsing the Internet or brochure to locate the information of the one or more events at the specific location.

Abstract: Methods and apparatus are provided for color correction of images. One or more colors in an image obtained from a static video camera are corrected by obtaining one or more historical background models from one or more prior images obtained from the static video camera; obtaining a live background model and a live foreground model from one or more current images obtained from the static video camera; generating a reference image from the one or more historical background models; and processing the reference image, the live background model, and the live foreground model to generate a set of color corrected foreground objects in the image. The set of color corrected foreground objects is optionally processed to classify a color of at least one of the foreground objects.

Abstract: Provided are an age estimation apparatus, an age estimation method, and an age estimation program capable of reducing the labor of labeling the image data used for age estimation. An age estimation apparatus for estimating an age of a person on image data includes a dimension compression unit for applying dimension compression to the image data to output low dimensional data; a clustering unit for performing clustering of the low dimensional data outputted; a labeling unit for labeling representative data of each cluster among the low dimensional data clustered; and an identification unit for estimating an age of a person on the basis of a learning result using a feature amount contained in labeled low dimensional data and unlabeled low dimensional data.

Abstract: The invention provides methods for automated classification of a medical diagnostic image of a lung according to its deduced probability of relating to a lung of a patient who is suffering from a diffuse parenchymal lung disease such as chronic obstructive pulmonary disease (COPD), cystic fibrosis, or severe asthma, or to a class of patients characterized by the severity of such a condition, or to a class of patients characterized by a prognostic likelihood of developing such a condition or severity of condition.

Abstract: Quantification of metric or functional parameters often requires image segmentation. A crucial part of such method is the model of the surface characteristics of the object of interest (features), which drives the deformable surface towards the object boundary in the image. According to the present invention, sections of the mesh are assigned to different classes for different features. According to the present invention, the assignment of mesh sections to the classes is adapted by using actual feature information from the unseen image. Advantageously, this allows for an adaptation of the feature category to which the mesh section is assigned and thereby allows an improved segmentation of the object.

Abstract: A method of identifying an object captured in a video image in a multi-camera video surveillance system is disclosed. Sets of identifying information are stored in profiles, each profile being associated with one object. The disclosed method of identifying an object includes comparing identifying information extracted from images captured by the video surveillance system to one or more stored profiles. A confidence score is calculated for each comparison and used to determine a best match between the extracted set of identifying information and an object. In one embodiment, the method is used as part of a facial recognition system incorporated into a video surveillance system.

Abstract: An image processing apparatus includes a photographing unit configured to generate a plurality of images by photographing a plurality of times a range which can be photographed, an object detection unit configured to detect a specified object from each of the plurality of images generated by the photographing unit, a position determination unit configured to determine an existing position of the specified object based on a detection result of the specified object in the plurality of images, and a range determination unit configured to determine a photographing range to be photographed within the range which can be photographed based on the existing position of the specified object determined by the position determination unit.

Abstract: Various embodiments of the invention provide systems and methods for classifying physical documents that have been converted to digital documents. Specifically, some embodiments are configured to classify digital documents that belong to a document classification whose representative members that lack structure or have varying structure, either of which makes automatic classification of such documents using conventional methods difficult. For example, certain systems and methods according to the invention can be used to classify physical real estate documents that have been converted to digital real estate documents, especially those that lack a discernable document structure.