Methods and apparatus of processing omnidirectional images are disclosed. According to one method, a current set of omnidirectional images converted from each spherical image in a 360-degree panoramic video sequence using a selected projection format is received, where the selected projection format belongs to a projection format group comprising a cubicface format, and the current set of omnidirectional images with the cubicface format consists of six cubic faces. If the selected projection format corresponds to the cubicface format, one or more mapping syntax elements to map the current set of omnidirectional images into a current cubemap image are signaled. The coded data are then provided in a bitstream including said one or more mapping syntax elements for the current set of omnidirectional images.

An electronic device includes a display circuit, a sensor circuit, a processor, and a memory. The memory stores instructions, which when executed by the processor, cause the electronic device to provide an omnidirectional image having a size larger than the screen through the display circuit, to provide the screen with a first area being a partial area of the omnidirectional image, to obtain information of a change in a direction, which the electronic device faces, through the sensor circuit when the omnidirectional image is provided, to provide the screen with a second area being another partial area of the omnidirectional image through the display circuit based on the obtained information, and to store a time point, at which the direction is changed, wherein the time point is correlated with a time point or a frame at which the omnidirectional image is provided.

Conversion between different projection formats of a 360-degree video may be performed in a uniform way. The geometric characteristics of the different projection formats may be considered when applying 3D-to-2D and 2D-to-3D mapping. Parameters reflective of the geometric characteristics of the different projection formats may be determined and used in the mapping and/or conversion. The parameters may include a normal vector that is perpendicular to a projection plane, a reference point in the projection plane, and/or unit vectors defined in the projection plane. An architecture with consolidated modules for handling the various projection formats may be provided.

A video processing method includes: decoding a part of a bitstream to generate a decoded frame, where the decoded frame is a projection-based frame that includes projection faces in a hemisphere cubemap projection layout; and remapping sample locations of the projection-based frame to locations on the sphere, where a sample location within the projection-based frame is converted into a local sample location within a projection face packed in the projection-based frame; in response to adjustment criteria being met, an adjusted local sample location within the projection face is generated by applying adjustment to one coordinate value of the local sample location within the projection face, and the adjusted local sample location within the projection face is remapped to a location on the sphere; and in response to the adjustment criteria not being met, the local sample location within the projection face is remapped to a location on the sphere.

An aerial navigation system comprises upright members mounted with anchor points at a substantially same height. Each anchor point is provided with an electric motor. A carrier device is coupled to the electric motors at corresponding ones of the anchor points using a set of first wires. The carrier device is operably moved by the electric motors in a horizontal plane co-planar with the anchor points. Further, a robotic device is suspended from the carrier device using a second wire. The robotic device is moveable within a volume defined between a ground surface, the plurality of upright members and the horizontal plane by at least one other electric motor mounted on the carrier device. Furthermore, a navigation control system synchronises operations of electric motors at the anchor points and the carrier device for moving the robotic device from a current location to a target location within the volume.

The present disclosure provides a three-dimensional object modeling method, an image processing method, and an image processing device. The three-dimensional object modeling method includes: a plane contour extraction step, in which for at least one panoramic image shot for each three-dimensional object, a plane contour of the at least one panoramic image in a three-dimensional space is extracted; a scale normalization step, in which a scale of the plane contour of the at least one panoramic image is normalized based on a camera position, to obtain normalized plane contours of respective panoramic images in the three-dimensional space; and a multi-object splicing step, in which based on the camera position, rotation and translation operations are performed on three-dimensional point coordinates of the normalized plane contours, so as to unify the three-dimensional point coordinates in the same coordinate system, to splice plane contours of respective three-dimensional objects into a multi-object plane contour.

An example system for generating stereoscopic light field panoramas includes at least one memory; and at least one processor to execute instructions to: access images of a scene captured by a plurality of cameras, the images including light field information associated with the scene; generate, based on first portions of the images, a first stereoscopic light field panorama for a first perspective associated with a left eye viewpoint of the scene, the left eye viewpoint associated with a left eye position on a first circle; and generate, based on second portions of the images, a second stereoscopic light field panorama for a second perspective associated with a right eye viewpoint of the scene, the right eye viewpoint associated with a right eye position on a second circle, the first circle being different than and concentric to the second circle.

Some implementations of the disclosure are directed to allowing interactive broadcast streamed video from games and other dynamic content. In accordance with some implementations, a content creator may publish a plurality of video surfaces of an environment for streaming to a plurality of client devices for video playback. The plurality of video surfaces may correspond, for example, to a cube map of a gaming environment captured from the perspective of a player. Upon receiving a stream including multiple video surfaces such as a cubemap, a media player of a viewer may generate a fully-rendered three-dimensional view of the environment.

Disclosed are an LSTM based personalization viewpoint estimation apparatus and method. According to an embodiment of the present disclosure, a viewpoint of a user is estimated using LSTM with respect to a 360-degree raw image and a viewpoint image that is a region of interest of the user and a tile including the estimated viewpoint is transmitted over a network, so that the amount of data to transmit compared to the network bandwidth may be reduced and a viewpoint of each individual may be estimated.

One variation of a method for autonomously generating an optimized audio-visual (AV) file from an original AV file comprises: a) generating a vector cube comprising a plurality of vector matrices for an original AV file; b) for each vector matrix within the vector cube, determining an optimal subframe having a subframe size larger than or equal to a predetermined minimum subframe size; and c) generating an optimized AV file based on the optimal subframes determined for each of the vector matrices within the vector cube.