Method for decoding immersive video and method for encoding immersive video
A method of processing an immersive video according to the present disclosure includes performing pruning for an input image, generating an atlas based on patches generated by the pruning and generating a cropped atlas by removing a background region of the atlas.
1 . A method of encoding an immersive video, the method comprising:
performing pruning for input images;
generating an atlas by packing a plurality of patches generated by the pruning; and
generating a cropped atlas from the atlas,
wherein the cropped atlas is generated by removing from the atlas, a background region in which none of the patches are packed,
wherein the cropped atlas is encoded instead of the atlas,
wherein metadata for the cropped atlas is further encoded as well as the cropped atlas, and
wherein the metadata comprises information representing whether the cropped atlas is encoded instead of the atlas.
2 . The method according to claim 1 , wherein in response that the information is encoded to indicate that the cropped atlas is encoded instead of the atlas, the metadata further comprises position information of the cropped atlas in the atlas and size information of the cropped atlas.
3 . The method according to claim 2 , wherein the position information comprises horizontal position information representing a position of a left boundary of the cropped atlas in the atlas and vertical position information representing a position of a top boundary of the cropped atlas in the atlas.
4 . The method according to claim 2 , wherein the size information includes width information of the cropped atlas and height information of the cropped atlas.
5 . The method according to claim 1 , wherein whether to generate the cropped atlas is determined based on whether object-based coding is applied, and
wherein in response to the object-based coding being applied, only the patches derived from a first object are packed into the atlas while patches derived from objects other than the first object are excluded.
6 . The method according to claim 1 , wherein based on at least one of a size of the background region in the atlas or an attribute of an object from which the patches are derived, whether to generate the cropped atlas is determined.
7 . A method of decoding an immersive video, the method comprising:
decoding an object map and either an atlas or a cropped atlas from a bitstream; and
synthesizing a view image for a predetermined object based on the object map and either the atlas or the cropped atlas,
wherein the atlas is divided into a data region and a background region, the data region being composed of a plurality of patches, the background region being a region in which none of patches are packed,
wherein the cropped atlas comprises only the data region excluding the background region,
wherein metadata for the cropped atlas is further decoded from the bitstream, and
wherein the metadata comprises information representing whether the cropped atlas is decoded instead of the atlas.
8 . The method according to claim 7 , wherein in response that the information indicates that the cropped atlas is decoded instead of the atlas, the metadata further comprises position information of the cropped atlas in the atlas and size information of the cropped atlas.
9 . The method according to claim 8 , wherein the position information comprises horizontal position information representing a position of a left boundary of the cropped atlas in the atlas in an original size and vertical position information representing a position of a top boundary of the cropped atlas in the atlas in the original size.
10 . The method according to claim 8 , wherein the size information includes width information of the cropped atlas and height information of the cropped atlas.
11 . The method according to claim 7 , wherein in response to object-based coding being applied, encoding the cropped atlas is decoded instead of the atlas, and
wherein in response to the object-based coding being applied, only patches derived from a first object are present in the atlas while patches derived from objects other than the first object are excluded.