Method and apparatus for reconstructing 360-degree image according to projection format

1. A method for decoding a 360-degree image, the method comprising:

receiving a bitstream in which the 360-degree image is encoded, the bitstream including data of an extended 2-dimensional image, the extended 2-dimensional image including a 2-dimensional image and a predetermined extension region, and the 2-dimensional image being from an image with a 3-dimensional projection structure and including one or more faces;

generating a prediction image by referring to syntax information obtained from the received bitstream;

obtaining a decoded image by adding the generated prediction image to a residual image, the residual image being obtained by inverse-quantizing and inverse transforming quantized transform coefficients, the quantized transform coefficients being obtained by performing entropy-decoding or variable-length-decoding for residual information included in the bitstream; and

reconstructing the decoded image into the 360-degree image according to a projection format,

wherein the projection format is selectively determined based on identification information, among a plurality of pre-defined projection formats including an ERP format in which the 360-degree image is projected in a two-dimensional plane or a CMP format in which the 360-degree image is projected in a cube,

wherein a size of the extension region is variably determined based on at least one of first information indicating a width of the extension region or second information indicating a height of the extension region, independently of a size of the 2-dimensional image,

wherein the extension region is not included in the image with the 3-dimensional projection structure,

wherein at least one of the identification information, the first information or the second information is obtained from the bitstream,

wherein the prediction image is generated by selecting one prediction mode among a plurality of prediction modes including intra prediction and inter prediction, and performing prediction based on the selected prediction mode, and

information on the selected prediction mode is obtained from the bitstream.

2. The method of claim 1 , wherein reconstructing the decoded image comprises:

obtaining, from the bitstream, arrangement information according to a regional packing, the arrangement information specifying one of a plurality of conversion types pre-defined in a decoding apparatus; and

rearranging at least one of a plurality of faces in the decoded image according to the arrangement information.

3. The method of claim 2 , wherein the pre-defined conversion types include at least one of 90-degree rotation, 180-degree rotation, 270-degree rotation, horizontal flipping, or vertical flipping.

4. The method of claim 1 , wherein the extension region is adjacent to a boundary of at least one of the faces.

5. The method of claim 4 , wherein the boundary of the face is discontinuous in the 2-dimensional image and continuous in the image with the 3-dimensional projection structure.

6. The method of claim 1 , wherein the first information includes at least one of width information of the extension region on a left side of the face or width information of the extension region on a right side of the face, and

wherein the second information includes at least one of height information of the extension region on a top side of the face or height information of the extension region on a bottom side of the face.

7. The method of claim 1 , wherein the one or more faces are specified based on position information and size information, and

wherein the position information includes at least one of horizontal start position information or vertical start position information, and the size information includes at least one of horizontal length information or vertical length information.

8. The method of claim 1 , wherein the identification information equal to a first value indicates the ERP format, and the identification information equal to a second value indicates the CMP format.

9. A method for encoding a 360-degree image, the method comprising:

generating a bitstream in which the 360-degree image is encoded, the bitstream including data of an extended 2-dimensional image, the extended 2-dimensional image including a 2-dimensional image and a predetermined extension region, and the 2-dimensional image being from an image with a 3-dimensional projection structure and including one or more faces;

wherein the bitstream comprises syntax information, the syntax information being referred to for generating a prediction image,

the bitstream comprises residual information, the residual information being entropy-decoded or variable-length-decoded for obtaining quantized transform coefficients, the quantized transform coefficients being inverse-quantized and inverse transformed for obtaining a residual image,

the prediction image is added to the residual image for obtaining a decoded image, and

the decoded image is reconstructed into the 360-degree image according to a projection format,

wherein the projection format is selectively determined based on identification information, among a plurality of pre-defined projection formats including an ERP format in which the 360-degree image is projected in a two-dimensional plane or a CMP format in which the 360-degree image is projected in a cube,

wherein a size of the extension region is variably determined based on at least one of first information indicating a width of the extension region or second information indicating a height of the extension region, independently of a size of the 2-dimensional image,

wherein the extension region is not included in the image with the 3-dimensional projection structure,

wherein at least one of the identification information, the first information or the second information is encoded into the bitstream, and

wherein the prediction image is generated by selecting one prediction mode among a plurality of prediction modes including intra prediction and inter prediction, and performing prediction based on the selected prediction mode, and

information on the selected prediction mode is encoded into the bitstream.

10. A method for transmitting a bitstream, the method comprising:

generating the bitstream in which the 360-degree image is encoded, the bitstream including data of an extended 2-dimensional image, the extended 2-dimensional image including a 2-dimensional image and a predetermined extension region, and the 2-dimensional image being from an image with a 3-dimensional projection structure and including one or more faces; and

transmitting the generated bitstream to an image decoding apparatus,

wherein the bitstream comprises syntax information, the syntax information being referred to for generating a prediction image,

the bitstream comprises residual information, the residual information being entropy-decoded or variable-length-decoded for obtaining quantized transform coefficients, the quantized transform coefficients being inverse-quantized and inverse transformed for obtaining a residual image,

the prediction image is added to the residual image for obtaining a decoded image, and

the decoded image is reconstructed into the 360-degree image according to a projection format,

wherein the projection format is selectively determined based on identification information, among a plurality of pre-defined projection formats including an ERP format in which the 360-degree image is projected in a two-dimensional plane or a CMP format in which the 360-degree image is projected in a cube,

wherein a size of the extension region is variably determined based on at least one of first information indicating a width of the extension region or second information indicating a height of the extension region, independently of a size of the 2-dimensional image,

wherein the extension region is not included in the image with the 3-dimensional projection structure,

wherein at least one of the identification information, the first information or the second information is encoded into the bitstream, and

wherein the prediction image is generated by selecting one prediction mode among a plurality of prediction modes including intra prediction and inter prediction, and performing prediction based on the selected prediction mode, and

information on the selected prediction mode is encoded into the bitstream.