Abstract: Points are grouped from a point cloud using Concavity-induced Distance (CID). First, a set of seed points is sampled from the input point cloud. Then a grouping of points is conducted based on a nearest neighbor search computed using CID. These two steps enable two novel solutions in point cloud segmentation and scene abstraction. In one embodiment, CID is determined between two points residing within an object surface. In a second embodiment, a CID is determined between groups of points.

Abstract: Generating images includes generating encoded data by encoding input data into a latent space. The encoded data is decoded through a first decoder having first decoder layers by processing the encoded data through one or more of the first decoder layers. The encoded data is decoded through a second decoder having second decoder layers by processing the encoded data through one or more of the second decoder layers. An updated feature map is generated by replacing at least a portion of a feature map output from a selected layer of the first decoder layers with at least a portion of a feature map output from a selected layer of the second decoder layers. An image is generated by further decoding the updated feature map through one or more additional layers of the first decoder layers.

Abstract: A 3D medical image recognition method and apparatus, a device, a non-transitory computer-readable storage medium, and a computer program product are provided, which relate to the field of artificial intelligence. View rearrangement processing is performed in an ith-round feature extraction process on an (i?1)th-round 3D medical image feature to obtain 2D image features. The (i?1)th-round 3D medical image feature is obtained by performing (i?1)th-round feature extraction on a 3D medical image. Different 2D image features are features of the (i?1)th-round 3D medical image feature in different views. Semantic feature extraction processing is performed on each 2D image feature in different views. Feature fusion processing is performed on the image semantic features to obtain an ith-round 3D medical image feature. Image recognition processing is performed on an Ith-round 3D medical image feature obtained through Ith-round feature extraction to obtain image recognition of the 3D medical image.

Abstract: Systems and methods for super-resolution (SR) are described. The SR method includes determining a direction type of an input image based on a gradient of the input image, obtaining a first intermediate image based on the direction type using a look-up table (LUT) that stores an SR operation result of a pixel value combination corresponding to a kernel set mapped to the determined direction type, wherein the first intermediate image is an SR image corresponding to the input image, applying SR interpolation to the input image to obtain a baseline image, and obtaining an output image based on the first intermediate image and the baseline image.

Abstract: A computer-implemented method includes receiving a focal image stack. The focal image stack includes multiple images of a measurement object. Each of the images captures a region of a surface of the measurement object with a defined focal plane position in a depth direction. The defined focal plane positions of the plurality of images are different from each other. The method includes generating an initial image with an extended depth of field in the depth direction based on the focal image stack. The method includes generating a corrected image by correcting a set of imaging errors in the initial image. The set of imaging errors includes a distortion error.

Abstract: An image processing apparatus, according to one embodiment of the present invention, comprises: a receiver for receiving a first image and a second image; a first processor for outputting a third image by using at least one of the first image and the second image; a signal line bypassing at least one of the first image and the second image; a switch for connecting a signal output from the receiver to the first processor or the signal line; and a transmitter for transmitting at least two of the first to third images.

Abstract: A false video detection method is provided. The method includes: retrieving N video segments in a target video, wherein each video segment includes M image frames; performing a feature retrieving procedure on the N×M image frames to obtain feature maps corresponding to the N×M image frames; inputting the feature maps into a time-attention model to obtain concentrated feature; multiplying the concentrated feature with the feature maps to obtain first enhanced feature maps, adding the first enhanced feature maps with the feature maps to obtain second enhanced feature maps, and inputting the second enhanced feature maps into a fully connected layer to obtain a determination result outputted by the fully connected layer, wherein the determination result indicates that the target video is true or false.

Abstract: The present invention provides an image processing apparatus that inspects an image formed on a printing medium. The image processing apparatus reads an image formed on a printing medium and obtains a target image to be inspected, and corrects a position of control point among control points disposed on the target image, at or near an edge of the printing medium, and performs alignment using the corrected position of the control point so as to align the target image to a reference image using a non-rigid-body alignment, and inspects for a defect in the target image on a basis of the aligned target image and the reference image. In the correcting, a position of a control point at or near the edge is corrected on a basis of an interval between adjacent control points at or near the edge.

Abstract: A method of image correlation includes: obtaining a first image frame with a set of distinct features; obtaining a second image frame at least partially overlapping the first image frame; carrying out a cross-correlation operation of the first and second image frames to obtain a cross-correlation result indicating correlation values for different offset positions of the second image frame with respect to the first image frame, the highest correlation value with its associated offset position in the cross-correlation result indicating the most likely offset position of the second image frame with respect to the first image frame; and carrying out a reaggregation operation of the cross-correlation result to generate a distinct global maximum correlation value from a plurality of non-distinct correlation values in an aggregated cross-correlation result. The reaggregation operation involves carrying out a convolution of the cross-correlation result with a 1×2, 2×1, or 2×2 array of 1's.

Abstract: An image processing apparatus, according to one embodiment of the present invention, comprises: a receiver for receiving a first image and a second image; a first processor for outputting a third image by using at least one of the first image and the second image; a signal line bypassing at least one of the first image and the second image; a switch for connecting a signal output from the receiver to the first processor or the signal line; and a transmitter for transmitting at least two of the first to third images.

Abstract: Region-of-interest (ROI)-based image enhancement using a residual network, including: generating, based on an input image and a residual path of a residual network, a first output corresponding to a region-of-interest of the input image; generating, based on the input image and a skip path of the residual network, a second output; and generating an output image based on the first output and the second output.

Abstract: Embodiments of this application disclose an image processing method and apparatus, and a computer-readable storage medium. The method includes acquiring an image pair to be processed, and performing image feature extraction on an image to be processed in the image pair to be processed to obtain an image; extracting an association feature from the image feature, the association feature characterizing mutual information between images in the image pair to be processed; identifying a common view image of a common view region in the image to be processed according to the association feature, and calculating a scale difference between common view images; adjusting a size of the common view image based on the scale difference to obtain an adjusted common view image; and extracting at least one common view feature point from each of the adjusted common view images, and processing the image based on the common view feature point.

Abstract: A method of compensating for focus drift in a microscopy system includes receiving data representing a reference light signal reflected from a sample and received by an image sensor, identifying from the data a peak of the reference light signal, the peak having a maximum intensity value for the reference light signal, extracting a sub-region having a center from the data, the center being located at the peak, calculating a first harmonic of a Fourier transform of the extracted sub-region, estimating a peak position of the reference light signal based on a phase of the first harmonic, and moving the lens system of the microscopy system or the sample in the axial direction using the translation stage based on the estimated position. Also, a system compensating for focus drift in a microscopy system according to the above method.

Abstract: An apparatus for estimating a road edge includes a processor configured to estimate a trajectory of a vehicle, estimate the position of an edge of a road from images generated during travel of the vehicle by a camera, identify an undetected section when an image from which the edge of the road is not detected was generated, determine whether the road in the undetected section has a structure with a road edge, estimate the position of the edge of the road in the undetected section, based on the positions of the edge of the road in the sections in front of and behind the undetected section, when the road in the undetected section has a structure with an edge, and omit to estimate the position of the edge of the road in the undetected section when the road in the undetected section has a structure without an edge.

Abstract: In a feature data generation system, an edge pattern indicating a boundary of a target feature existing around a vehicle is extracted from position data of a large number of measurement points surrounding the vehicle measured using a LIDAR technology, on the basis of image data of surroundings of the vehicle obtained by imaging so as to reduce the amount of information, positions in a terrestrial reference frame are assigned to the extracted edge pattern, and a shape characteristic vector and a feature characteristic vector to be used for generating or updating a feature in a feature database, are generated from the edge pattern to which the positions in the terrestrial reference frame are assigned.

Abstract: Disclosed is an image processing apparatus including a receiving section that receives a captured image of a target object, an instruction section that gives a user an instruction on which direction to capture an image of the target object, an inferencing section that performs an inference of an imaging direction on the basis of the received captured image, and a verifying section that performs predetermined verification processing related to the target object on the basis of an imaging direction as the instructed direction and the imaging direction as a result of the inference.

Abstract: Systems and methods for vision test and uses thereof are disclosed. A method may be implemented on a mobile device having at least a processor, a camera and a display screen. The method may include capturing at least one image of a user using the camera of the mobile device; interactively guiding the user to a predetermined distance from the display screen of the mobile device based on the at least one image; presenting material on the display screen upon a determination that the user is at the predetermined distance from the display screen; and receiving input from the user in response to the material presented on the display screen. The material presented on the display screen may be for assessing at least one characteristic of the user's vision. Mobile devices and non-transitory machine-readable mediums having machine-executable instructions embodied thereon for assessing a user's vision also are disclosed.

Abstract: This disclosure enables various technologies for analyzing behaviors of objects or with respect to objects based on stereo imageries thereof. For example, such analysis may be useful in enforcement of certain actions by objects or with respect to objects, surveillance of objects or with respect to objects, or other situations involving analyzing behaviors of objects or with respect to objects.

Abstract: This application discloses a face image processing method performed by an electronic device. The method includes: acquiring a face image of a source face and a face template image of a template face; performing three-dimensional face modeling on the face image and the face template image to obtain a three-dimensional face image feature of the face image and a three-dimensional face template image feature of the face template image; fusing the three-dimensional face image feature and the three-dimensional face template image feature to obtain a three-dimensional fusion feature; performing face replacement feature extraction on the face image based on the face template image to obtain an initial face replacement feature; transforming the initial face replacement feature based on the three-dimensional fusion feature to obtain a target face replacement feature; and replacing the template face with the source face based on the target face replacement feature to obtain a target face image.

Abstract: A first frame of a video stream is obtained. The first frame is defined by a plurality of pixels associated with a set of color data. A determination is made that a pixel of the plurality of pixels comprises high-frequency information. Responsive to the determination that the pixel comprises high-frequency information, a pixel lock is generated for the pixel such that color data associated with the pixel is maintained during a color accumulation process for at least one of the first frame or a second frame of the video stream that is subsequent to the first frame.