In-loop filtering method according to adaptive pixel classification standard

1. A method of performing in-loop filtering according to a sample classification standard, wherein the method is performed in an image decoding apparatus, the method comprising:

dividing, based on division information signaled through a bitstream, a picture including a current block into tiles, one of the tiles being divided into tile segments by at least one division boundary line in only one of a vertical direction or a horizontal direction independently of other tiles, and each of the tile segments being composed of an integer number of consecutive complete coding tree unit rows and having a rectangular shape;

reconstructing the current block, the current block resulting from partitioning a coding block based on at least one of a quad-tree partition or a binary-tree partition;

classifying a reconstructed sample in the current block based on the sample classification standard, the sample classification standard being an absolute classification standard or a relative classification standard;

obtaining offset information based on a result of classifying the reconstructed sample;

adding an offset value to the reconstructed sample based on the obtained offset information; and

outputting the reconstructed sample to which the offset value is added,

wherein, when a size of the coding block is greater than or equal to a pre-defined threshold size, only the quad-tree partition is performed for the partitioning of the coding block,

wherein the pre-defined threshold size is 128×128,

wherein the binary-tree partition is performed based on a binary-tree partition flag indicating whether to perform the binary-tree partition and a binary-tree partition direction flag indicating horizontal partition or vertical partition, and

wherein the binary-tree partition is performed only when the quad-tree partition is no longer performed.

2. The method of claim 1 ,

wherein the classifying of the reconstructed sample includes:

when the sample classification standard is the absolute classification standard, classifying the reconstructed sample according to a band to which a brightness value of the reconstructed sample belongs.

3. The method of claim 1 ,

wherein the classifying of the reconstructed sample includes:

when the sample classification standard is the relative classification standard, classifying the reconstructed sample on the basis of at least one of gradient information or edge information derived by comparing a pixel value of the reconstructed sample with pixel values of neighboring samples adjacent to the reconstructed sample.

4. A method of performing in-loop filtering according to a sample classification standard, wherein the method is performed in an image encoding apparatus, the method comprising:

dividing a picture including a current block into tiles, one of the tiles being divided into tile segments by at least one division boundary line in only one of a vertical direction or a horizontal direction independently of other tiles, and each of the tile segments being composed of an integer number of consecutive complete coding tree unit rows and having a rectangular shape;

reconstructing the current block, the current block resulting from partitioning a coding block based on at least one of a quad-tree partition or a binary-tree partition;

classifying a reconstructed sample in the current block based on the sample classification standard, the sample classification standard being an absolute classification standard or a relative classification standard;

obtaining offset information based on a result of classifying the reconstructed sample;

adding an offset value to the reconstructed sample based on the obtained offset information; and

outputting the reconstructed sample to which the offset value is added,

wherein division information on the dividing of the picture is encoded in a bitstream,

wherein, when a size of the coding block is greater than or equal to a pre-defined threshold size, only the quad-tree partition is performed for the partitioning of the coding block,

wherein the pre-defined threshold size is 128×128,

wherein the binary-tree partition is performed only when the quad-tree partition is no longer performed, and

wherein, when the binary-tree partition is performed, a binary-tree partition flag indicating whether to perform the binary-tree partition and a binary-tree partition direction flag indicating horizontal partition or vertical partition are encoded in the bitstream.

5. A non-transitory computer readable medium having stored thereon instructions that when executed by a processor perform a decoding method, including:

dividing, based on division information signaled through a bitstream, a picture including a current block into tiles, one of the tiles being divided into tile segments by at least one division boundary line in only one of a vertical direction or a horizontal direction independently of other tiles, and each of the tile segments being composed of an integer number of consecutive complete coding tree unit rows and having a rectangular shape;

reconstructing the current block, the current block resulting from partitioning a coding block based on at least one of a quad-tree partition or a binary-tree partition;

classifying a reconstructed sample in the current block based on a sample classification standard, the sample classification standard being an absolute classification standard or a relative classification standard,

obtaining offset information based on a result of classifying the reconstructed sample;

adding an offset value to the reconstructed sample based on the obtained offset information; and

outputting the reconstructed sample to which the offset value is added,

wherein, when a size of the coding block is greater than or equal to a pre-defined threshold size, only the quad-tree partition is performed for the partitioning the coding block,

wherein the pre-defined threshold size is 128×128,

wherein the binary-tree partition is performed based on a binary-tree partition flag indicating whether to perform the binary-tree partition and a binary-tree partition direction flag indicating horizontal partition or vertical partition, and

wherein the binary-tree partition is performed only when the quad-tree partition is no longer performed.

6. The method of claim 1 ,

wherein a first division boundary line for a first tile of the tiles and a second division boundary line for a second tile adjacent to the first tile are discontinuous at a boundary between the first tile and the second tile.