Inter-image component prediction method, and image encoding and decoding method and device using same

1. An image decoding method, comprising:

determining an inter-component prediction mode of a chroma component block based on an inter-component prediction mode list and predetermined index information, wherein the inter-component prediction mode list is configured with k inter-component prediction modes;

downsampling reference samples for inter-component prediction of the chroma component block based on the determined inter-component prediction mode;

classifying the downsampled reference samples into a first group and a second group, wherein the first group represents a minimum value group of first downsampled reference samples having relatively small values among the downsampled reference samples and the second group represents a maximum value group of second downsampled reference samples having relatively large values among the downsampled reference samples;

obtaining a first average value between the first downsampled reference samples in the first group and a second average value between the second downsampled reference samples in the second group;

calculating a difference between the first average value and the second average value;

deriving a parameter of a linear prediction model based on the difference between the first average value of the first group and the second average value of the second group; and

performing the inter-component prediction on the chroma component block by using the parameter of the linear prediction model,

wherein a filter for the downsampling is variably determined depending on whether a top boundary of a luma component block corresponding to the chroma component block is on a top boundary of a coding tree unit (CTU) to which the luma component block belongs.

2. The method of claim 1 , wherein the downsampling is selectively performed for a reference sample at a specific position.

3. The method of claim 2 , wherein the specific position is determined based on encoding information on at least one of the chroma component block or the luma component block.

4. The method of claim 3 , wherein the downsampling is performed by using a neighboring sample adjacent to the reference sample at the specific position in at least one of a vertical direction, a horizontal direction or a diagonal direction.

5. The method of claim 4 , wherein the downsampling is performed based on at least one of availability of a neighboring block of the luma component block, a CTU related parameter or a predetermined flag, and

wherein the flag represents whether a sample of the chroma component block has a position shifted by a predetermined distance from a corresponding sample position of the luma component block.

6. An image encoding method, comprising:

determining an inter-component prediction mode of a chroma component block from an inter-component prediction mode list including k inter-component prediction modes;

downsampling reference samples for inter-component prediction of the chroma component block based on the determined inter-component prediction mode;

classifying the downsampled reference samples into a first group and a second group, wherein the first group represents a minimum value group of first downsampled reference samples having relatively small values among the downsampled reference samples and the second group represents a maximum value group of second downsampled reference samples having relatively large values among the downsampled reference samples;

obtaining a first average value between the first downsampled reference samples in the first group and a second average value between the second downsampled reference samples in the second group;

calculating a difference between the first average value and the second average value;

deriving a parameter of a linear prediction model based on the difference between the first average value of the first group and the second average value of the second group; and

performing the inter-component prediction on the chroma component block by using the parameter of the linear prediction model,

wherein a filter for the downsampling is variably determined depending on whether a top boundary of a luma component block corresponding to the chroma component block is on a top boundary of a coding tree unit (CTU) to which the luma component block belongs.

7. The method of claim 6 , wherein the downsampling is selectively performed for a reference sample at a specific position.

8. The method of claim 7 , wherein the specific position is determined based on encoding information on at least one of the chroma component block or the luma component block.

9. The method of claim 8 , wherein the downsampling is performed by using a neighboring sample adjacent to the reference sample at the specific position in at least one of a vertical direction, a horizontal direction or a diagonal direction.

10. The method of claim 9 , wherein the downsampling is performed based on at least one of availability of a neighboring block of the luma component block, a CTU related parameter or a predetermined flag, and

wherein the flag represents whether a sample of the chroma component block has a position shifted by a predetermined distance from a corresponding sample position of the luma component block.

11. A non-transitory computer-readable storage medium for storing a bitstream generated by an image encoding method,

the image encoding method comprising:

determining an inter-component prediction mode of a chroma component block from an inter-component prediction mode list including k inter-component prediction modes;

downsampling reference samples for inter-component prediction of the chroma component block based on the determined inter-component prediction mode;

classifying the downsampled reference samples into a first group and a second group, wherein the first group represents a minimum value group of first downsampled reference samples having relatively small values among the downsampled reference samples and the second group represents a maximum value group of second downsampled reference samples having relatively large values among the downsampled reference samples;

obtaining a first average value between the first downsampled reference samples in the first group and a second average value between the second downsampled reference samples in the second group;

calculating a difference between the first average value and the second average value;

deriving a parameter of a linear prediction model based on the difference between the first average value of the first group and the second average value of the second group; and

performing the inter-component prediction on the chroma component block by using the parameter of the linear prediction model, wherein a filter for the downsampling is variably determined depending on whether a top boundary of a luma component block corresponding to the chroma component block is on a top boundary of a coding tree unit (CTU) to which the luma component block belongs.

12. The method of claim 1 , wherein, in response to a case where the top boundary of the luma component block is on the top boundary of the coding tree unit, a first filter having n-taps is used for the downsampling,

wherein, in response to a case where the top boundary of the luma component block is not on the top boundary of the coding tree unit, a second filter having m-taps is used for the downsampling, and

wherein n and m are different from each other.

13. The method of claim 12 , wherein n is equal to 3 and m is equal to 6.