Abstract: The disclosed technology generally relates to detecting and identifying objects in digital images, and more particularly to detecting, identifying and/or tracking moving objects in video images using an artificial intelligence neural network configured for deep learning. In one aspect, a method comprises capturing a video input from a scene comprising one or more candidate moving objects using a video image-capturing device, where the video input comprises at least two temporally spaced image frames captured from the scene. The method additionally includes transforming the video input into one or more image pattern layers, where each of the image pattern layers comprises a pattern representing one of the candidate moving objects. The method additionally includes determining a probability of match between each of the image pattern layers and a stored image in a big data library.

Abstract: Rate control techniques are provided for encoding an input video sequence into a compressed coded bitstream with multiple coding passes. The final coding pass may comprise final splices with non-overlapping frames that do not extend into neighboring final splices. A final splice in the final coding pass may correspond to at least one non-final splice in a non-final coding pass. A non-final splice may have overlapping frames that extend into neighboring final splices in the final coding pass. The overlapping frames in the non-final splice may be used to derive complexity information about the neighboring final splices. The complexity information about the neighboring final splices, as derived from the overlapping frames, may be used to allocate or improve rate control related budgets in encoding the final splice into the compressed coded bitstream in the final coding pass.

Abstract: A method for decoding image, includes generating a residual block; reconstructing an intra prediction mode group indicator and a prediction mode index of a current block, wherein the indicator indicates whether the intra prediction mode of the current block belongs to which intra prediction mode group among a first or second group, and the prediction mode index specifies intra prediction mode in an intra prediction mode group indicated by the indicator; constructing the first group including three intra prediction modes using valid intra prediction modes of left and top blocks of the current block; determining the intra prediction mode corresponding to the prediction mode index in the first group as the intra prediction mode of the current block when the indicator indicates the first group; generating a prediction block based on the determined intra prediction mode of the current block; generating a reconstructed block using the residual block and the prediction block; and performing deblocking filtering on a

Abstract: A panoramic video system obtains a digital video image having a width w1 pixels and a height of h pixels. A plurality of digital still images is also obtained, each still image comprising a ray having a width w2 comprising pixels, 1?w2?50, a height substantially equal to h. A storage device records or a display device displays the digital video image and plurality of digital still images. The set of n rays on either or both sides of the first horizontal angle is obtained from the plurality of still images to form one or two still background images, adjacent horizontally to the digital video image, each of the background image(s) having a width of n·w2 and a height substantially equal to h, and each ray forming one vertical stripe of the still background image.

Abstract: The present invention relates to a video encoding/decoding method and apparatus, and more particularly, to a method and apparatus for generating a reference image for a multiview video. The video encoding method includes, in the presence of a second image having a different view from a first image having a first view, transforming the second image to have the first view, generating a reference image by adding the second image to a side of the first image, and storing the reference image in a reference picture list.

Abstract: A method of rectifying stereo images includes providing a plurality of pairs of sets of keypoints extracted from a pair of current stereo images and from a pair of previous stereo images wherein each pair of stereo images includes a left image and a right image respectively obtained from a left camera and a right camera; providing a plurality of pairs of sets of next-choice-match points extracted from the pair of current stereo images and the pair of previous stereo images; finding one or more anchor points in a left previous image; finding a right linking point which is the corresponding keypoint in the right previous image, and a left linking point which is the corresponding keypoint in the left current image; finding a closing point; and calculating a cost from the right linking point, the left linking point, and the closing point.

Abstract: A tree examining system is provided which employs a thermographic image which electronically differentiates individual areas of a total area of a tree by areas according to temperature differential. Pixels in individual areas determined as abnormal are assigned a color which differentiates them from adjacent areas determined as normal. A displayable colorized image thereby produced provides a map viewable by a user to cut or trim the tree to remove the abnormal areas shown colorized therein.

Abstract: A device for nullifying images captured by drones utilizing a high power light to overexpose said images. One or more cameras installed in the device capture images of the area surrounding the device. Computer vision and deep learning are utilized to identify drones in the captured images. If a drone is identified, the location thereof is estimated. A high power light installed in the device is directed at the estimated location of the drone to overexpose any images that are being taken of the area around the device by the drone.

Abstract: There is disclosed a system for viewing augmented, virtual, mixed, and combined reality environments. The system calculates the position and orientation of a movable display within physical space and translated those into an appropriate rendering of the content on the movable display also taking into account a head location for a viewer. The motion and location data for the movable display device and a viewer's head may be generated using trackers of various types.

Abstract: According to one embodiment, an image processing apparatus comprises a screen, a camera, a distance sensor, and a processor. The screen has one surface where an object is put. The camera photographs an image from the other surface of the screen. The distance sensor measures a distance from the one surface of the screen and generates distance information indicating the distance. The processor extracts a region of the object from the photographed image, specifies the object from the image of the region of the object, acquires an outer dimension preset to the specified object, calculates a length, from the screen, of the object put on the one surface of the screen based on the distance information, and outputs an alert when the outer dimension does not match the length of the object.

Abstract: A method of partitioning in video coding for JVET, comprising representing a JVET coding tree unit as a root node in a quadtree plus binary tree (QTBT) structure that can have quadtree or binary partitioning of the root node and quadtree or binary trees branching from each of the leaf nodes. The partitioning at any depth can use asymmetric binary partitioning to split a child node represented by a leaf node into two child coding units of unequal size, representing the two child coding units as leaf nodes in a binary tree branching from the parent leaf node and coding the child coding units represented by final leaf nodes of the binary tree with JVET. Disclosed is a generalized method of partitioning a block, either square or rectangular, which leads to more flexible block sizes with possible higher coding efficiency.

Abstract: A non-transitory storage medium storing an image transmission program that is executed on an in-vehicle device mounted in a vehicle having an imaging unit configured to image surroundings of the vehicle to acquire an image and a communication unit configured to perform communication with an information center, the image transmission program includes: an image acquisition step of causing the imaging unit to acquire an image relating to an intersection based on whether or not a mark target facing the intersection is prominent; and a transmission step of causing the communication unit to transmit the image acquired to the information center.

Abstract: Sub-block based entropy coding more efficiently encodes and decodes content by selecting the best coding scheme for sub-blocks of content. By comparing coding schemes for each sub-block of content and selecting the coding scheme which utilizes the fewest bits, the content is more efficiently encoded.

Abstract: The present disclosure relates to an image processing apparatus and an image processing method capable of suppressing a deterioration in image quality. The image processing apparatus includes a threshold value setting unit which sets a threshold value for identifying a feature of neighboring pixels of a current block in an intra prediction process in encoding of image data according to a bit depth of the image data and a filtering processing unit which performs a filtering process on the neighboring pixels by using a filter according to the feature of the neighboring pixels identified by using the threshold value set by the threshold value setting unit. The present disclosure may be applied to, for example, an image processing apparatus.

Abstract: A de-blocking method is applied to a reconstructed projection-based frame having a first projection face and a second projection face, and includes obtaining a first spherical neighboring block for a first block with a block edge to be de-blocking filtered, and selectively applying de-blocking to the block edge of the first block for at least updating a portion of pixels of the first block. There is image content discontinuity between a face boundary of the first projection face and a face boundary of the second projection face. The first block is a part of the first projection face, and the block edge of the first block is a part of the face boundary of the first projection face. A region on a sphere to which the first spherical neighboring block corresponds is adjacent to a region on the sphere from which the first projection face is obtained.

Abstract: A method of decoding JVET video that includes receiving a bitstream indicating how a coding tree unit was partitioned into coding units, and parsing said bitstream to generate at least one predictor based on an intra prediction mode signaled in the bitstream, the intra prediction mode selected from a plurality of intra prediction modes for calculating a prediction pixel P[x,y] at coordinate x,y for the coding unit. A number of intra prediction modes available for coding the coding unit are reduced by replacing two or more non-weighted intra prediction modes by a weighted intra prediction mode.

Abstract: Systems and methods related to augmented reality guidance for a room measurement process may include presenting various text, cues, and indications to guide the room measurement process. A user device including an imaging sensor and position and orientation sensors may present, via a display, the various text, cues, and indications during portions of the room measurement process, including identification of the floor, walls, and ceiling.

Abstract: Systems and methods related to a room measurement process may include determining a floor plane associated with a room, determining a plurality of wall planes associated with the room, and determining a ceiling plane associated with the room. The room measurements may be based on data received from an imaging sensor and position and orientation sensors associated with a user device. In addition, a three-dimensional geometric shell of the room may be generated based on the room measurements.

Abstract: Provided is an image decoding method including: determining at least one prediction block included in a current picture from among pictures forming an image; determining a first motion vector related to a current prediction block from among the at least one prediction block; determining a first reference block included in a first reference picture, based on the first motion vector; determining a second reference block included in a second reference picture, based on a location of the first reference block; and decoding the current picture by performing inter prediction on the current prediction block by using at least one of the first reference block and the second reference block, wherein the second reference picture includes a prediction value of the first reference picture. Also, provided are an image decoding apparatus performing the image decoding method, and an image encoding method and apparatus corresponding to the image decoding method and apparatus.

Abstract: An unmanned photographing device and method thereof is provided which includes establishing a wireless connection with an external electronic device using a communication device, receiving a first signal at a first three-dimensional (3D) position from the external electronic device through the wireless connection, the first signal comprising data associated with a first image comprising a first object, determining a second 3D position based on at least part of the data and the first 3D position, controlling the unmanned photographing device to fly to or near the second 3D position, tracking a second object corresponding to the first object using a camera, and capturing a second image comprising the second object at or near the second 3D position such that the second image corresponds to the first image.