Abstract: A redundant tracking system comprising multiple redundant tracking sub-systems, enabling seamless transitions between such tracking sub-systems, provides a solution to this problem by merging multiple tracking approaches into a single tracking system. This system is able to combine tracking objects with six degrees of freedom (6DoF) and 3DoF through combining and transitioning between multiple tracking systems based on the availability of tracking indicia tracked by the tracking systems. Thus, as the indicia tracked by any one tracking system becomes unavailable, the redundant tracking system seamlessly switches between tracking in 6DoF and 3DoF thereby providing the user with an uninterrupted experience.

Abstract: In one respect, there is provided a system that may include a processor and a memory. The memory may be configured to store instructions that results in operations when executed by the processor. The operations may include: training a machine learning model by at least processing a training set with the machine learning model, the training set including at least one synthetic image that is generated by applying one or more modifications to a non-synthetic image; determining, based at least on a result of the processing of the mixed training set, that the machine learning model is unable to classify images having a specific modification; and training the machine learning model with additional training data that includes one or more additional synthetic images having the specific modification. Related methods and articles of manufacture are also disclosed.

Abstract: The present application discloses a method for assessing aesthetic quality of a natural image based on multi-task deep learning. Said method includes: step 1: automatically learning aesthetic and semantic characteristics of the natural image based on multi-task deep learning; step 2: performing aesthetic categorization and semantic recognition to the results of automatic learning based on multi-task deep learning, thereby realizing assessment of aesthetic quality of the natural image. The present application uses semantic information to assist learning of expressions of aesthetic characteristics so as to assess aesthetic quality more effectively, besides, the present application designs various multi-task deep learning network structures so as to effectively use the aesthetic and semantic information for obtaining highly accurate image aesthetic categorization.

Abstract: A generation device according to the subject application includes an acquiring unit and a generating unit. The acquiring unit acquires multiple pieces of vector information that correspond to each search target during the proximity search. The generating unit generates multiple pieces of centroid information, in which each piece of vector information is related in accordance with a distance for vector quantization, based on multiple pieces of vector information acquired by the acquiring unit such that a difference in the number of pieces of vector information related to each piece of centroid information falls within a predetermined value.

Abstract: A method for controlling unlocking and related products are provided. An electronic device includes at least one processor and a computer readable storage coupled to the at least one processor. The computer readable storage stores at least one computer executable instruction thereon, which when executed by the at least one processor, cause the at least one processor to carry out actions, including: obtaining a face image; carrying out a group of rough features and a group of fine features from the face image; carrying out a verification operation for the group of rough features and the group of the fine features; carrying out a next unlocking process when verification of the group of rough features and the group of fine features is passed.

Abstract: A degree-of-saliency map generated from an input image is divided into a plurality of partial areas, and the degree of nonuniformity is calculated from the distribution characteristics of the degrees of saliency of the partial areas. Whether a main subject is present in the input image is judged based on the calculated degree of nonuniformity.

Abstract: Techniques are disclosed for removing text from an image of a form document. The text is removed by determining a spectral domain representation of the image and applying a filter to remove the high-frequency components which correspond to the text in the form. An image is reconstructed from the filtered spectral domain representation which maintains the low-frequency components, while deemphasizing or removing the high-frequency components. A shape of the filter applied to the spectral domain representation is determined based on a similarity measure between the image of the form and the reconstructed image.

Abstract: Method for a communication device switching among a plurality of sub-bands in a wireless network. The method comprises measuring a signal quality for an operation sub-band, comparing the signal quality with a signal quality threshold, determining a switch action according to the comparison of the signal quality with the signal quality threshold, and switching from the operation sub-band according to the determined switch action. The communication device and access node thereof are also disclosed.

Abstract: Implementations relate to generating theme-based videos. In some implementations, a computer-implemented method to automatically generate a theme-based video includes obtaining image criteria for a theme from a theme definition, querying an image database to obtain a plurality of images that meet the image criteria, determining that a count of the plurality of images satisfies a threshold based on the theme definition, and generating the theme-based video that includes one or more of the plurality of images.

Abstract: For displaying a subject composition, a method determines a subject number, a subject age category, and a subject size for one or more subjects of an image. The method further calculates a subject distribution that indicates a position for a face of each subject. In addition, the method determines a subject pose for each subject that indicates a body position for the subject. The method further displays the subject composition comprising the subject distribution and the subject poses overlaying a display image.

Abstract: Systems and methods for correlating visual content in 2D and 3D scenes of a geographical region can include at least one data source for providing geographical data, a display device of an aircraft, and a processor. The display device can display a 2D view and a 3D view representing top-down views of the geographical region using data received from the data source(s). The processor(s) can identify a first image point within the 2D view referenced by a first cursor of the 2D view. The processor(s) can determine, using information received from the at least one data source, a geographical location corresponding to the first image point. The processor(s) can determine a second image point within the 3D view corresponding to the determined geographical location. The processor(s) can position a second visual indicator of the 3D view to reference the determined second image point within the 3D view.

Abstract: A method for aerial image processing preferably includes receiving ground control point (GCP) information, receiving a set of images, determining spatial information associated with the images, and determining GCP image locations; and can additionally or alternatively include determining aerial map information, training an automated GCP detector, and/or any other suitable elements. A system for aerial image processing preferably includes one or more computing devices, and can additionally or alternatively include one or more GCPs, aerial vehicles, and/or any other suitable elements.

Abstract: A method for operating a robotic system includes determining a discretized object model based on source sensor data; comparing the discretized object model to a packing plan or to master data; determining a discretized platform model based on destination sensor data; determining height measures based on the destination sensor data; comparing the discretized platform model and/or the height measures to an expected platform model and/or expected height measures; and determining one or more errors by (i) determining at least one source matching error by identifying one or more disparities between (a) the discretized object model and (b) the packing plan or the master data or (ii) determining at least one destination matching error by identifying one or more disparities between (a) the discretized platform model or the height measures and (b) the expected platform model or the expected height measures, respectively.

Abstract: A display processing device includes a moving image generation unit, a display control unit, and an input detection unit. The moving image generation unit generates a combined moving image that is a moving image combined from a plurality of moving images. The input detection unit detects an input operation performed on the input device that receives the input operation for inputting at least one of a position on the combined moving image displayed on a display device and a direction based on the combined moving image. The display control unit allows the display device to display a moving image that is based on, among the plurality of moving images constituting the combined moving image, one or more moving images according to the input detected by the input detection unit, and that has a display mode different from at least the combined moving image.

Abstract: The present disclosure provides systems and methods for sorting a cell. The system may comprise a flow channel configured to transport a cell through the channel. The system may comprise an imaging device configured to capture an image of the cell from a plurality of different angles as the cell is transported through the flow channel. The system may comprise a processor configured to analyze the image using a deep learning algorithm to enable sorting of the cell.

Abstract: An image processing apparatus includes an acquiring unit and a determining unit. For example, the acquiring unit acquires an imaging image captured by a camera mounted on a vehicle and registers the acquired imaging image in drive recorder information. When a vanishing point is detected from the imaging image, the determining unit determines that the imaging direction of the imaging device is normal such that the front of the vehicle is normally imaged. When the vanishing point is not detected from the acquired the imaging image, the determining unit determines that the imaging direction of the imaging device is abnormal such that the front of the vehicle is not normally imaged.

Abstract: An electronic device comprising: a display configured to output image data; and at least one processor configured to: detect an input; select a portion of the image data based on the input; and generate a preview of the image data based on the selected portion.

Abstract: An image processing apparatus includes an image obtaining unit configured to obtain a time-sequential image obtained in order of a start image, a plurality of intermediate images, and an end image, an image selecting unit configured to select a plurality of combinations of images including the start image, at least one of the plurality of intermediate images, and the end image, and a conformable image determining unit configured to determine a conformable combination from among the plurality of combinations based on image quality of the images included in each of the plurality of combinations and a similarity between the images included in each of the plurality of combinations.

Abstract: An image processing apparatus obtains a first image, cuts out at least a portion of the first image to generate a second image that is distortion-corrected, receives designation of a first area in which image quality is to be different in the first image, calculates a second area in the second image, the second area corresponding to the first area designated in the first image, and sets, as an area associated with a function that influences encoding processing for encoding the second image, an area that includes the second area and is based on a processing unit of the encoding processing for encoding the second image, the processing unit being an area in the second image.

Abstract: Methods and image-guided surgical navigation systems for updating an existing landmark registration of one or more landmark features in a common coordinate space are provided. The image-guided surgical navigation system includes a processor, an imaging device coupled to the processor, and a memory coupled to the processor. The image-guided surgical navigation system may be configured to capture a planar image of a region of interest, where the planar view image includes illustration of the one or more landmark features; generate a depth map from the planar view image; based on the depth map, identify a current location of the one or more landmark features in the common coordinate space; and transform the existing landmark registration to the current location of the one or more landmark features in the common coordinate space.

Abstract: Embodiments of the present invention disclose a method and device for constructing a spatial index of massive point cloud data. The method includes: partitioning acquired original point cloud data to obtain a plurality of point cloud data blocks; constructing an octree index of a current point cloud data block for each point cloud data block; and merging the octree indexes of the plurality of point cloud data blocks to obtain a spatial index structure of the original point cloud data.

Abstract: In a calibration routine, a first set of X-ray recordings of at least one closed first reference package without content is produced, and a mass calibration signature is derived therefrom. A second set of X-ray recordings of at least one closed second reference package having a reference content is produced, and a reference signature is derived therefrom. From the reference signature and the mass calibration signature, a reference measurement value is derived via subtraction. The reference mass of the reference content is ascertained by weighing and assigned to the reference measurement value. In ongoing measurement operation, at least one set of measuring X-ray recordings of closed foil packages each having a content is produced and a measurement signature is derived therefrom. Herefrom, and from the mass calibration signature, measurement values for the individual closed foil packages are derived via subtraction, from which the masses of the contents are quantitatively determined.

Abstract: A device for determining information for video data includes one or more processors implemented in circuitry that are configured to determine one or more most-interested regions of a plurality of regions of an image of video data from data representative of the one or more most-interested regions. The one or more processors are further configured to generate a request specifying the one or more most-interested regions using the data representative of the one or more most-interested regions and output the request to a server device.

Abstract: The present disclosure provides a picture recognition method and apparatus, a computer device and a computer-readable medium. The method comprises: obtaining a to-be-recognized picture; according to the to-be-recognized picture and a pre-trained picture recognition model, acquiring a predicted class tag of the to-be-recognized picture; the pre-trained picture recognition model employs a convolutional neural network model, and the convolutional neural network model comprises at least one layer of kernel pooling layer upgrading a dimensionality of a channel from one dimension to multiple dimensions; recognizing a class of the to-be-recognized picture according to the predicted class tag of the to-be-recognized picture. By recognizing the to-be-recognized picture by using the picture recognition model having the kernel pooling layer, the present disclosure can achieve recognition of picture classes with finer granularities and effectively improve the picture recognition accuracy and efficiency.

Abstract: According to one embodiment, an information processing device includes a memory and a processor. The processor is configured to acquire an image, determine a likelihood of presence of an objected within the image, display information on a position of the object within the image based on the likelihood, identify the object within the image, and generate a recognition result based on the identified object.

Abstract: Various devices, arrangements and methods for managing communications using a head mounted display device are described. In one aspect, tracking data is generated at least in part by one or more sensors in a head mounted display (HMD) device. The tracking data indicates one or more facial movements of a user wearing the HMD device. A patch image is obtained based on the tracking data. The patch image is merged with a facial image. Various embodiments relate to the HMD device and other methods for generating and using the patch and facial images.

Abstract: A redundant tracking system comprising multiple redundant tracking sub-systems, enabling seamless transitions between such tracking sub-systems, provides a solution to this problem by merging multiple tracking approaches into a single tracking system. This system is able to combine tracking objects with six degrees of freedom (6DoF) and 3DoF through combining and transitioning between multiple tracking systems based on the availability of tracking indicia tracked by the tracking systems. Thus, as the indicia tracked by any one tracking system becomes unavailable, the redundant tracking system seamlessly switches between tracking in 6DoF and 3DoF thereby providing the user with an uninterrupted experience.

Abstract: A system configured to improve the operations associated with generating virtual representations of physical environments to recognize the physical environments and/or relocalize within the virtual representations in a substantially real time system. In some cases, the system may use a first pre-training phase of descriptors and/or split nodes of regression forests using features common across various scenes to learn general image appearance, and a second training phase of descriptors and/or leaf nodes of regression forests to learn scene specific features. The system may align the features using an orientation vector, correct for camera perspective and lens distortion of the features as well as learn robust illumination invariant features from real and synthetic data.

Abstract: Methods and systems for videoconferencing include generating work quality metrics based on emotion recognition of an individual such as a call center agent. The work quality metrics allow for workforce optimization. One example method includes the steps of receiving a video including a sequence of images, detecting an individual in one or more of the images, locating feature reference points of the individual, aligning a virtual face mesh to the individual in one or more of the images based at least in part on the feature reference points, dynamically determining over the sequence of images at least one deformation of the virtual face mesh, determining that the at least one deformation refers to at least one facial emotion selected from a plurality of reference facial emotions, and generating quality metrics including at least one work quality parameter associated with the individual based on the at least one facial emotion.

Abstract: There is provided a method and an apparatus for updating a background model used for background subtraction of an image. An image is received and each pixel in the image is classified as being foreground or background by performing background subtraction using a background model which comprises a collection of background samples for each pixel in the image. For each pixel in the image being classified as background, the background model is updated by selecting a background sample in the collection of background samples associated with the pixel which fulfills a similarity condition, and replacing the background sample which fulfills the similarity condition with image data of the pixel.

Abstract: A method of determining a summation of pixel characteristics for a rectangular region of a digital image includes determining if a base address for a data element in an integral image buffer is aligned for an SIMD operation by a processor embedded in an electronic assembly configured to perform Haar-like feature calculations. The data element represents a corner of the rectangular region of an integral image. The integral image is a representation of the digital image. The integral image is formed by data elements stored in the integral image buffer. The data element is loaded from the integral image buffer to the processor when the base address is aligned for the SIMD operation. An offset data element of an offset integral image is loaded from an offset integral buffer when the base address is non-aligned for the SIMD operation. The offset data element represents the corner of the rectangular region.

Abstract: The present invention relates to a face recognition system and a face recognition method, mainly comprising a lighting preprocessing module, a feature generation module, a facial feature library module, and a feature matching and recognition module. In the face recognition method, the inputted face image is first light-difference optimized by means of the lighting preprocessing module, then a feature vector of the face image is generated by means of the feature generation module; then, the feature vector to be recognized is matched with all of the feature vectors in the feature library and computed to arrive at an identity result corresponding to the feature vector to be recognized. In the technical solutions of the present invention, the identity of a face image to be recognized may be determined in real time for a system whose computing resources are limited, using a short period of time and having high accuracy.

Abstract: Plural pieces of reference data are stored for plural pixel arrangement patterns. Each piece of reference data indicates which of first-type edge pixels needs to be subjected to determination to determine whether an outside neighboring first-type pixel adjacent to the first-type edge pixel exists. A piece of reference data corresponding to a target pixel block is specified as target reference data. A reference-data-based judgment is executed on the target pixel block to perform the determination onto those first-type edge pixels that are indicated by the target reference data as need to be subjected to the determination. A correction is executed on the target pixel block to correct the type of at least one first-type pixel contained in the target pixel block into the second type in a case where a result of the reference-data-based judgment indicates that no outside neighboring first-type pixel exists with respect to the target pixel block.

Abstract: An ROI in a first image and a second image are synthesized by synthesizing first wavelet coefficient data corresponding to the first image and second wavelet coefficient data corresponding to the second image. An ROI coefficient and a non-ROI coefficient in the first wavelet coefficient data are determined on the basis of a mask for the first wavelet coefficient data, and the ROI coefficient and a wavelet coefficient in the second wavelet coefficient data, which is corresponding to the non-ROI coefficient, are synthesized. When the above coefficient synthesis is performed after a decomposition level of the first wavelet coefficient data is increased, the first wavelet coefficient data is adjusted on the basis of auxiliary wavelet coefficient data corresponding to an auxiliary image which can be equated with the first image before increasing the decomposition level.

Abstract: Video information defining video content may be accessed. The video content may include video frames. The video frames may be grouped into similarity groups based on similarity of the video frames. The similarity groups may include a first similarity group including a first video frame and a second video frame. The first video frame may be similar to the second video frame. Representative video frames may be selected from the similarity groups. An individual representative video frame may correspond to an individual similarity group. Representative video frames may include a first representative video frame selected from and corresponding to the first similarity group. The video content may be analyzed based on the representative video frames.

Abstract: A system for and method for improved computer vision image processing and image segmentation and recognition is disclosed. The system and method utilizes a cost function approach for improved image segmentation and recognition. In particular, a global cost function is defined and then the global cost function is minimized. This global cost function takes into account two processing pipelines of data determined by the operation of two different processing technologies upon the target input image. Constraints are utilized to ensure data consistency across the model and between the data pipelines. The system and method are useful for processing low quality images containing alphanumeric data such as floorplan images.

Abstract: To precisely extract a static feature value from consecutive images taken in a dynamic environment that is crowded by many people. A feature value extraction apparatus includes: a consecutive-image acquisition unit configured to acquire consecutive images that are consecutively taken; a local feature value extraction unit configured to extract a local feature value at each feature point from the consecutive images; a feature value matching unit configured to perform matching between the consecutive input images for the local feature value extracted by the local feature value extraction unit; and an invariant feature value calculation unit configured to acquire, in the local feature values for which matching between a predetermined number of consecutive images has been obtained by the feature value matching unit, an average of the local feature values whose position changes between the consecutive images are equal to or less than a predetermined threshold value as invariant feature value.

Abstract: Embodiments of the present application relate to a method, apparatus, and system for detecting a specified image identifier. The method includes retrieving a target image to be detected from a predetermined area, binarizing the target image to be detected to obtain a target binary image to be detected, calibrating connected domains of the target binary image to be detected, successively retrieving image features of candidate connected domains, and comparing the image features corresponding to the candidate connected domains to image features of a standard specified identifier image, wherein the candidate connected domains are determined based at least in part on the calibration of the connected domains, and determining a candidate connected domain as the location of the specified identifier image based at least in part on the comparison of the image features corresponding to the candidate connected domains to image features of the standard specified identifier image.

Abstract: A tool shape measurement device includes: a contour detection unit that detects a tool contour from an image of a rotating tool that is taken; an axis direction calculation unit that calculates a tool axis direction that is an axis direction of the rotating tool on the basis of the tool contour; a tool diameter measurement unit that calculates an apparent tool diameter of the rotating tool on an imaging surface on the basis of a calibrated positional and postural relationship between an imaging device and the rotating tool, the tool axis direction, and the tool contour; and a tool diameter correction unit that calculates a distance between the imaging device and the rotating tool using the tool axis direction, and corrects the apparent tool diameter to an actual tool diameter by correcting distortion in the tool contour on the basis of the distance.

Abstract: There is provided a method to identify structure of a native PDF document. The method comprises: obtaining a native PDF document having a first line to start a table and a second line to end the table; detecting a value of a physical feature of the native PDF document, wherein the physical feature has a corresponding weighting factor; initiating a value to the weighting factor; assigning a first status for the first line and a second status for the second line based on (a) the physical feature and (b) the weighting factor; and identifying a location of the table on native PDF document from the first status and the second status, thus yielding an identified location of the table.

Abstract: An apparatus and methods for training and/or operating a robotic device to follow a trajectory. A robotic vehicle may utilize a camera and stores the sequence of images of a visual scene seen when following a trajectory during training in an ordered buffer. Motor commands associated with a given image may be stored. During autonomous operation, an acquired image may be compared with one or more images from the training buffer in order to determine the most likely match. An evaluation may be performed in order to determine if the image may correspond to a shifted (e.g., left/right) version of a stored image as previously observed. If the new image is shifted left, right turn command may be issued. If the new image is shifted right then left turn command may be issued.

Abstract: A state estimation system that utilizes long-range stereo visual odometry that can degrade to a monocular system at high-altitude, and integrates GPS, Barometer and IMU measurements. The system has two main parts: An EKF that is loosely fused and a long-range visual odometry part. For visual odometry, the system takes the EKF information for robust camera pose tracking, and the visual odometry outputs will be the measurement for EKF state update.

Abstract: A degree-of-saliency map generated from an input image is divided into a plurality of partial areas, and the degree of nonuniformity is calculated from the distribution characteristics of the degrees of saliency of the partial areas. Whether a main subject is present in the input image is judged based on the calculated degree of nonuniformity.

Abstract: The technology disclosed relates to creating user-defined interaction spaces and modalities in a three dimensional (3D) sensor space in response to control gestures. It also relates to controlling virtual cameras in the 3D sensor space using control gestures and manipulating controls of the virtual cameras through the control gestures. In particular, it relates to defining one or more spatial attributes of the interaction spaces and modalities in response to one or more gesture parameters of the control gesture. It also particularly relates to defining one or more visual parameters of a virtual camera in response to one or more gesture parameters of the control gesture.

Abstract: A computer-implemented method and system for seeding video programs with episodic interactive items improve computer performance by averting tracking loss. The system includes a scene map table that maintains a scene record for every scene in the video. The scene record contains boundary information for the scene and linking information to companion scenes that are visually similar. When a developer tags an episodic item in the video for interactivity, the scene record for the scene where the item is tagged is accessed in the scene map table. The boundary information from this scene record is used to track the item across the scene without tracking loss and the linking information from this scene record is used to access scene records for companion scenes, allowing the item to be tracked across the companion scenes without tracking loss.

Abstract: An apparatus for determining a distance to a target area includes an imaging system configured to provide at least two images of a target area. The images are associated with different imaging axes. A Lidar system including at least one laser is configured to direct an optical beam to the target and an optical detection system configured to receive a portion of the optical beam from the target area and establish a distance to the target area based on the received portion.

Abstract: Among other things, embodiments of the present disclosure can help to automatically generate images displaying activity-based information and distribute such images to other users, such as members of a social network. The image may be modified based on activity data associated with other users and/or from other activity sensors.

Abstract: A method for operating a computer graphic system, the method comprising: inputting a media content object (MCO) into a feature extractor comprising semantic abstraction levels; extracting feature maps from the MCO on each of the semantic layers; selecting at least a portion of the MCO to be analyzed; determining, based on the analysis of the feature maps from the portion of the MCO and the analysis of a previous state of a recognition unit, one or more feature maps selected from the feature maps of the semantic layers; determining a weight for each feature map; repeating the determining steps N times, each time processing, based on the analysis, each feature map by applying the corresponding weight; inputting the processed feature maps to the recognition unit; and analyzing a number of the processed feature maps until a prediction about the portion of the MCO is output.

Abstract: System and method for creating a machine vision application. A machine vision prototype comprising a plurality of machine vision steps specifying a machine vision image processing algorithm and associated parameters may be stored. The steps may be interpretable by an emulator to perform the specified image processing on an image by emulating or simulating execution of the steps on a hardware accelerator, e.g., a programmable hardware element or graphics processing unit. An emulator may emulate or simulate execution of the steps on the hardware accelerator, thereby generating image processing results, which may be displayed for validation of the emulating or simulating by a user. The prototype may be analyzed, and based on the analyzing, an estimate of resource usage or performance of the image processing algorithm for the hardware accelerator may be determined and displayed, and may be usable for target platform selection or modification of the image processing algorithm.

Abstract: A computer-implemented method for identifying features in an image. The method comprises fitting a plurality of second models to the image, the plurality of second models together modelling a region of interest, wherein each part of the region of interest is modelled by at least two of the plurality of second models; and identifying the features in the image based upon the fit of the plurality of second models.