Abstract: An image coding method of coding an image on a per coding unit basis, the method comprising: applying a frequency transform to luminance data and chrominance data of transform units in the coding unit including predetermined blocks each corresponding to one or more of the transform units; and coding the luminance data and the chrominance data to which the frequency transform has been applied to generate a bitstream in which the luminance data and the chrominance data are grouped on a per predetermined block basis.

Abstract: A system and method are provided for discriminating and tracking a coherently structured feature dynamically defined in evolving manner within a migratory medium. A data set is captured for a plurality of physical points defined within a multi-dimensional physical space, in terms of a plurality of scalar parameter values. At least one pre-selected target feature type is established, as is a scalar field predefined by at least one of the scalar parameters. A sparse set of key points is selectively generated a within the scalar field. Each key point is associated with one of the physical points and descriptive information adaptively determined therefor from the data content within a neighborhood of the physical point coinciding therewith. At least one predetermined feature-based operation is executed responsive to the descriptive information of the key points as a surrogate for execution generally on the data set.

Abstract: Techniques for locating features of interest in warped images are described. For example, a destination image is generated from a source image based on warping a portion of the source image according to a backward map. Each of the two images is associated with a mesh having a tile-based geometry. The backward map maps points from the destination mesh to pixels in the source image. Triangles are formed from the tiles of the destination mesh. A forward map is generated based on the triangles and the backward map. The forward map maps points from the source mesh to pixels in the destination image. A point from the source mesh is identified that corresponds to a pixel of a feature of interest. The forward map is applied to this point to locate a pixel in the destination image. This pixel corresponds to the location of the feature in the destination image.

Abstract: A method and system are provided for optical character recognition (OCR) of multi-language content. The method includes extracting a text portion from an image received from a user-computing device. The text portion comprises a plurality of keywords associated with a plurality of languages. The method further includes segmenting the plurality of keywords into a plurality of layers. Each layer of the plurality of layers comprises one or more keywords which are associated with a language. The method further comprise generating an OCR output of each of the plurality of layers based on the language associated with the one or more keywords in each of the plurality of layers. The method further comprises generating an electronic document of the received image based on the generated OCR output of each of the plurality of layers. The method further includes transmitting the generated electronic document to the user-computing device.

Abstract: A neural network system that includes: multiple subnetworks that includes: a first subnetwork including multiple first modules, each first module including: a pass-through convolutional layer configured to process the subnetwork input for the first subnetwork to generate a pass-through output; an average pooling stack of neural network layers that collectively processes the subnetwork input for the first subnetwork to generate an average pooling output; a first stack of convolutional neural network layers configured to collectively process the subnetwork input for the first subnetwork to generate a first stack output; a second stack of convolutional neural network layers that are configured to collectively process the subnetwork input for the first subnetwork to generate a second stack output; and a concatenation layer configured to concatenate the pass-through output, the average pooling output, the first stack output, and the second stack output to generate a first module output for the first module.

Abstract: Systems and methods acquire and/or generate multiple different images of the same biometric identity, identify specific instances of biometric features in each of the different images, and merge the identified specific instances of biometric features into a data record that provides a digital representation of the biometric identity. Examples of biometric identities include fingerprints, handprints, palm prints, and thumbprints. In one embodiment, a counter is associated with each specific instance of a biometric feature found in the multiple images. Specific instances of biometric features found most frequently have high counts and are indicative of true identifications; those with low counts are indicative of false identifications. A threshold distinguishes between true and false identifications. Those specific instances with counts below the threshold are excluded when the digital representation of the biometric identity is generated.

Abstract: A method of one-shot joint instance and pose recognition in an artificial neural network includes receiving a first instance of a reference object from a reference image. The reference object has a first identity and a first pose in the first instance. The method also includes generating a first orbit of the reference object comprising additional poses including a second pose for the reference object. The method further includes recognizing a second instance of an example object from an example image. The example object has the first identity and the second pose in the second instance.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for compressing images using neural networks. One of the methods includes receiving an image; processing the image using an encoder neural network, wherein the encoder neural network is configured to receive the image and to process the image to generate an output defining values of a first number of latent variables that each represent a feature of the image; generating a compressed representation of the image using the output defining the values of the first number of latent variables; and providing the compressed representation of the image for use in generating a reconstruction of the image.

Abstract: An example radial histogram matching system may generate a target radial histogram by identifying pixels in an input digital image that are determined to be on, traversing each of the pixels in the input digital image one time to determine how many of the pixels that are turned on are contained in each sector of a circle and assigning to each elements of the target radial histogram the number of on pixels in that element. The system may also compare the target radial histogram to each of an initial and a sequence of rotated radial histograms to determine a match score for each of the comparisons, and identify an offset rotation between a baseline pattern digital image and the input digital image based on the match scores.

Abstract: Methods for annotating objects within image frames are disclosed. Information is obtained that represents a camera pose relative to a scene. The camera pose includes a position and a location of the camera relative to the scene. Data is obtained that represents multiple images, including a first image and a plurality of other images, being captured from different angles by the camera relative to the scene. A 3D pose of the object of interest is identified with respect to the camera pose in at least the first image. A 3D bounding region for the object of interest in the first image is defined, which indicates a volume that includes the object of interest. A location and orientation of the object of interest is determined in the other images based on the defined 3D bounding region of the object of interest and the camera pose in the other images.

Abstract: Methods and systems are provided for assessing the presence of functionally significant stenosis in one or more coronary arteries, further known as a severity of vessel obstruction. The methods and systems can implement a prediction phase that comprises segmenting at least a portion of a contrast enhanced volume image data set into data segments corresponding to wall regions of the target organ, and analyzing the data segments to extract features that are indicative of an amount of perfusion experiences by wall regions of the target organ. The methods and systems can obtain a feature-perfusion classification (FPC) model derived from a training set of perfused organs, classify the data segments based on the features extracted and based on the FPC model, and provide, as an output, a prediction indicative of a severity of vessel obstruction based on the classification of the features.

Abstract: Provided is a method for salient object segmentation of an image by aggregating a multi-linear exemplar regressors, including: analyzing and summarizing visual attributes and features of a salient object and a non-salient object using background prior and constructing a quadratic optimization problem, calculating an initial saliency probability map, selecting a most trusted foreground and a background seed point, performing manifold preserving foreground propagation, generating a final foreground probability map; generating a candidate object set for the image via an objectness adopting proposal, using a shape feature, a foregroundness and an attention feature to characterize each candidate object, training the linear exemplar regressors for each training image to characterize a particular saliency pattern of the image; aggregating a plurality of linear exemplar regressors, calculating saliency values for the candidate object set of a test image, and forming an image salient object segmentation model capable

Abstract: The present disclosure relates to an image searching system and method for searching by using images of objects. Images of objects which it is desired to search are input via a PC-based terminal equipped with a webcam or a smart phone and then the input object images are analyzed using image recognition technology based on feature point extraction such that types of code contained in the object images are categorized and a decoder matching the object type is automatically driven such that the object is read, and thus a search function is provided which allows integrated recognition of all objects regardless of the type of object and which is convenient for the user.

Abstract: A method is provided for receiving from a user a first search query for media files from a collection of media files. The method includes providing, in response to the first search query, a first search result to the user, the first search result including responsive media files associated with the first search query, detecting a user selection of a responsive media file based on an interaction between the user and the responsive media file and selecting multiple similar media files having a visual similarity with the responsive media file. The method also includes generating a refined query based on a caption associated with a refined cluster of media files, the refined cluster of media files being proximal to the similar media file, and displaying, to the user and based on the refined query, a refined search result comprising refined media files from the refined query.

Abstract: An efficient method and apparatus for identifying one or more targets in an image is presented. A matching operation is performed that compares the image against multiple templates jointly. The multiple templates are sorted into multiple clusters. The joint template matching operation achieves an improvement in performance by focusing on a subset of the clusters. For each grid location, clusters whose reference templates have low matching scores with the image content of the grid location are bypassed or excluded from the matching operation. This saving in computation load is made possible by the clustering of the templates, which can be performed offline and does not affect the performance of the joint template matching operation.

Abstract: A system configured to improve the operations associated with generating virtual representations on limited resources of a mobile device. In some cases, the system may utilize viewpoint bundles that include collection of image data with an associated pose in relative physical proximity to each other to render a virtual scene. In other cases, the system may utilize 2.5D manifolds including 2D image data and a weighted depth value to render the 3D environment.

Abstract: A face recognition based key generation apparatus controls a key generation model that is formed of a CNN and an RNN to be learned to generate a desired key having a consistent value by using sample facial images of a key owner and a PIN of the key owner as inputs, and the key generation model receives a facial image of the key owner and the PIN of the key owner, as inputs at a desired key generation time, and generates a key.

Abstract: According to an embodiment, an image decoding method includes: determining at least one prediction unit included in a current coding unit that is one of at least one coding unit splitting the image; generating a merge candidate list related to a current prediction unit included in the at least one prediction unit; determining a prediction mode to be performed in the prediction unit based on the merge candidate list; and performing prediction according to the determined prediction mode, wherein the generating of the merge candidate list further comprises determining whether to add a view synthesis prediction merging candidate to the merge candidate list based on whether a neighboring prediction unit adjacent to the current prediction unit has been encoded using the view synthesis prediction merging candidate.

Abstract: One embodiment provides a method, including: providing, on an input and display device, an input field overlay application occupying an initial area of the input and display device, the input field overlay application overlaying an input field of an underlying application; accepting, using the input and display device, one or more handwriting input strokes provided to the input field overlay application; and resizing, on the input and display device, the initial area. Other embodiments are described and claimed.

Abstract: Disclosed is a system for image dehazing by modifying a lower bound of a transmission rate and a method therefor, whereby a clear image is obtained by removing components, such as fog or haze, from an image having low image quality due to fog or haze, the system configured such that a lower bound of a transmission rate representing how many rate of haze is mixed is calculated for each pixel, and an initial transmission rate is obtained by an exponentiation operation with the lower bound of transmission rate. By transmission rate correction that reserves an edge showing a large change relative to the initial transmission rate and processes a smooth area showing a small change with a low-pass filter, a final transmission rate is obtained such that a clear image is obtained by removing haze components.