Abstract: For each frame of a video, a determination is made whether an image of a hand exists in the frame. When at least one frame of the video includes the image of the hand, locations of the hand in the frames of the video are tracked to obtain a tracking result. A verification is performed to determine whether the tracking result is valid in a current frame of the frames of the video. When the tracking result is valid in the current frame of the video, a location of the hand is tracked in a next frame. When the tracking result is not valid in the current frame, localized hand image detection is performed on the current frame.

Abstract: For each frame of a video, a determination is made whether an image of a hand exists in the frame. When at least one frame of the video includes the image of the hand, locations of the hand in the frames of the video are tracked to obtain a tracking result. A verification is performed to determine whether the tracking result is valid in a current frame of the frames of the video. When the tracking result is valid in the current frame of the video, a location of the hand is tracked in a next frame. When the tracking result is not valid in the current frame, localized hand image detection is performed on the current frame.

Abstract: Disclosed are systems, devices, and methods for determining pleura boundaries of a lung, an exemplary method comprising acquiring image data from an imaging device, generating a set of two-dimensional (2D) slice images based on the acquired image data, determining, by a processor, a seed voxel in a first slice image from the set of 2D slice images, applying, by the processor, a region growing process to the first slice image from the set of 2D slice images starting with the seed voxel using a threshold value, generating, by the processor, a set of binarized 2D slice images based on the region grown from the seed voxel, filtering out, by the processor, connected components of the lung in each slice image of the set of binarized 2D slice images, and identifying, by the processor, the pleural boundaries of the lung based on the set of binarized 2D slice images.

Abstract: The invention proposes an instantaneous search and comparison method for large-scale distributed palm vein micro-feature data, which consists of three parts: 1) feature extraction and calculation of palm vein micro-feature images; 2) building a feature database; 3) search and comparison. The technical solution provided by this invention, referring to the idea of GIS and web search method, is applied to the search and comparison of the palm vein micro-feature data, which enables instantaneous recognition on massive palm vein micro-feature data under large-scale, large-traffic and high-frequency application scenarios, and solves the technical problem that traditional palm vein recognition methods can not be applied to large-scale and large-traffic scenarios due to low speed thereof.

Abstract: A method for obtaining a probability in a 3D probability map, includes: obtaining at least one value of at least one parameter for each stop of a 3D moving window, wherein a first, second, third and fourth of the stops are partially overlapped, the first and second stops are shifted from each other by a distance equal to a first dimension of a computation voxel, the first and third stops are shifted from each other by a distance equal to a second dimension of the computation voxel, and the first and fourth stops are shifted from each other by a distance equal to a third dimension of the computation voxel; matching the at least one value to a classifier to obtain a first probability for each stop of the 3D moving window; and calculating a second probability for the computation voxel based on information associated with the first probabilities for the first through fourth stops.

Abstract: An exemplary volumetric reconstruction system accesses first and second color and depth data captured for a surface point on a surface of an object in a real-world capture space. The first color and depth data is captured by a first capture device positioned to have a first vantage point of the surface of the object, while the second color and depth data is captured for the surface point by a second capture device positioned to have a second vantage point of the surface of the object. Based on the first and second color and depth data, the volumetric reconstruction system determines a confidence field value for a voxel node corresponding to the surface point, and, based on that confidence field value, generates reconstructed color and depth data for a volumetric reconstruction of the surface of the object. Corresponding methods and systems are also disclosed.

Abstract: Systems and methods of real-time augmented visualization of anatomical features are disclosed herein. The systems and methods can use concurrently collected and/or preexisting data regarding an anatomical structure to present an image of the anatomical structure to an operator in virtual space as overlayed in real space. The systems and methods can include acquiring image data from an anatomical structure of a subject. Visual image data of the subject can then be received, including non-rigid deformation of the subject. The device pose of the ultrasound device in real space can be determined using device pose data. Image pose of the image data can then be determined in virtual space. Then, a perspective pose of the image of at least a portion of the anatomical structure in virtual space can be mapped to the subject in real space.

Abstract: For anomaly detection based on topogram predication from surface data, a sensor captures the outside surface of a patient. A generative adversarial network (GAN) generates a topogram representing an interior anatomy based on the outside surface of the patient. An X-ray image of the patient is acquired and compared to the generated topogram. By quantifying the difference between the real X-ray image and the predicted one, anatomical anomalies may be detected.

Abstract: The present disclosure provides systems and methods for end-to-end handwritten text recognition using neural networks. Most existing hybrid architectures involve high memory consumption and large number of computations to convert an offline handwritten text into a machine readable text with respective variations in conversion accuracy. The method combine a deep Convolutional Neural Network (CNN) with a RNN (Recurrent Neural Network) based encoder unit and decoder unit to map a handwritten text image to a sequence of characters corresponding to text present in the scanned handwritten text input image. The deep CNN is used to extract features from handwritten text image whereas the RNN based encoder unit and decoder unit is used to generate converted text as a set of characters. The disclosed method requires less memory consumption and less number of computations with better conversion accuracy over the existing hybrid architectures.

Abstract: The present disclosure relates to an x-ray imaging method using a variable imaging plane projection and to an x-ray imaging device applying the same. By applying a variable imaging plane projection using at least two sets of scan data for different heights obtained while varying the height of the x-ray generator from an imaging object, it is possible to solve problems caused due to a magnification effect generated in the x-ray imaging field based on a fan-beam-type or cone-beam-type x-ray generator, thereby providing more accurate x-ray image information.

Abstract: This invention relates to a hyperspectral imaging system for denoising and/or color unmixing multiple overlapping spectra in a low signal-to-noise regime with a fast analysis time. This system may be configured to carry out Hyper-Spectral Phasors (HySP) calculations to effectively analyze hyper-spectral time-lapse data. For example, this system may be configured to carry out Hyper-Spectral Phasors (HySP) calculations to effectively analyze five-dimensional (5D) hyper-spectral time-lapse data. Advantages of this imaging system may include: (a) fast computational speed, (b) the ease of phasor analysis, and (c) a denoising algorithm to obtain the minimally-acceptable signal-to-noise ratio (SNR). An unmixed color image of a target may be generated. These images may be used in diagnosis of a health condition, which may enhance a patient's clinical outcome and evolution of the patient's health.

Abstract: Frames of a video frame sequence capturing one or more skin regions of a body are provided to a first neural network. The first neural network generates respective appearance representations based on the frames. An appearance representation generated based on a particular frame is indicative of a spatial distribution of a physiological signal across the particular frame. Simultaneously with providing the frames to the first neural network, the frames are also provided to a second neural network. The second neural network determines the physiological signal based on the frames. Determining the physiological signal by the second neural network includes applying the appearance representations, generated by the first neural network, to outputs of one or more layers of the second neural network to emphasize regions, in the frames, that exhibit relatively stronger presence of the physiological signal and deemphasize regions, in the frames, that exhibit relatively weaker presence of physiological signal.

Abstract: The subject disclosure provides systems and methods for determination of Area of Interest (AOI) for different types of input slides. Slide thumbnails may be assigned into one of five different types, and separate algorithms for AOI detection executed depending on the slide type. Slide types include ThinPrep® slides, tissue micro-array (TMA) slides, control HER2 slides with 4 cores, smear slides, and a generic slide. The slide type may be assigned based on a user input. Customized AOI detection operations are provided for each slide type. If the user enters an incorrect slide type, operations include detecting the incorrect input and executing the appropriate method. The result of each AOI detection operations provides as its output a soft-weighted image having zero intensity values at pixels that are detected as not belonging to tissue, and higher intensity values assigned to pixels detected as likely belonging to tissue regions.

Abstract: An image processing module and related method. The module (IP) comprises—one or more input interfaces (IN) configured for receiving i) a first input image (II) acquired of an object by an imaging apparatus (IM) at a first geometrical configuration of the X-ray imaging apparatus and ii) a specification of a change from said first geometrical configuration to a second geometrical configuration of the imaging apparatus. An upsampler component (US) of the module (IP) computes a new image (I+) of the object (OB) by applying a geometrical transformation to said first input image. The geometrical transformation corresponds to said change in geometrical configuration of the imaging apparatus.

Abstract: A method for quantitatively measuring internal defects in an as-cast steel product includes optically scanning at least a portion of a surface of the steel product with a scanning device to create a digital image thereof, analyzing the digital image to calculate a quantitative value for an amount of internal defects therein, and normalizing the quantitative value to a rating according to a standardized scale.

Abstract: A clinical trial re-evaluation system is operable to perform at least one assessment function on a set of medical scans for each of a first subset of a set of patients of a failed clinical trial to generate automated assessment data for each of the first subset of the set of patients. The first subset of the set of patients corresponds to a subset of human assessment data determined to have failed to meet criteria of the clinical trial. Patient re-evaluation data is generated for each of the first subset of the set of patients by comparing the automated assessment data to the criteria. The patient re-evaluation data for a second subset of the first subset of the set of patients indicates the automated assessment data passes the criteria. Trial re-evaluation data is generated based on the patient re-evaluation data for transmission to a computing device for display.

Abstract: A head-mounted device (“HMD”) is configured to perform intrinsic and/or extrinsic calibration of the HMD's camera system by exploiting a displayed electronic image rendered on a separate display screen. A series of images are captured using one or more of the HMD's cameras. The displayed image is a known image that includes markers with known characteristics to the HMD. The known characteristics include known marker shapes and a number of coded or un-coded markers. Each image in the series captures the displayed image at a different angle or distance relative to another image in the series. The HMD then identifies, from within the series of images, two-dimensional image positions of the markers. The HMD uses the two-dimensional image positions and a determined three-dimensional position of the markers to perform a bundle adjustment used to subsequently determine a position and angular alignment of the separate display screen relative to the HMD.

Abstract: Differential modulation schemes encode a data channel within host signal or noisy environment in a manner that is robust, flexible to achieve perceptual quality constraints, and provides improved data capacity. Differential arrangements enable a decoder to suppress host signal or other background signal interference when detecting, synchronizing and extracting an encoded data channel. They also enable the incorporation of implicit or explicit synchronization components, which are either formed from the data signal or are complementary to it.

Abstract: Multiple mode star tracker methods and systems in which attitude information and image information is generated are provided. The multiple mode star tracker includes a detector having a plurality of pixels arranged in a focal plane array. The detector is operated to obtain multiple image frames from within a field of view containing a plurality of stars. For each of the image frames, the attitude of the detector and in turn the attitude of each pixel is determined. Based on the attitude quaternion of the individual pixels within a plurality of frames, image data from the plurality of frames is co-added or stacked to form a composite image. The co-addition of multiple frames of image data enables or facilitates the detection of dim objects by the multiple mode star tracker.

Abstract: A visual recognition system to process images includes a global sub-network including a convolutional layer and a first max pooling layer. A local sub-network is connected to receive data from the global sub-network, and includes at least two convolutional layers, each connected to a max pooling layer. A fusion network is connected to receive data from the local sub-network, and includes a plurality of fully connected layers that respectively determine local feature maps derived from images. A loss layer is connected to receive data from the fusion network, set filter parameters, and minimize ranking error.