# US 2014/0153844 A1 · Interpolation method and prediction method using same > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2014/0153844 A1 - Document ID: US20140153844A1 - Status: Published application - Kind: A1 - Application number: 14/130,180 - Filed: 2011-07-28 - Published: 2014-06-05 - Earliest claimed: 2011-06-30 - Assignees: LG Electronics Inc. - Inventors: Yong Joon Jeon; Jie JIA; Byeong Moon Jeon; Seung Wook Park - Web page: https://patentgrep.com/patent/US20140153844A1 ## CPC - G06T 3/4007: based on interpolation, e.g. bilinear interpolation (image demosaicing G06T3/4015; edge-driven or edge-based scaling G06T3/403) - H04N 19/523: with sub-pixel accuracy - H04N 19/82: involving filtering within a prediction loop - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL - H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION ## Abstract The present invention relates to a method of performing interpolation for prediction between screens by using a pixel sample of a current picture and a pixel sample of a reference picture, and a prediction method using same. The interpolation method according to the present invention includes: calculating movement information for generating a prediction block; and generating a partial pixel sample on the basis of an integer sample set including integer pixel samples to which an interpolation filter is applied. Here, the integer sample set includes at least one current integer sample used to generate a partial pixel sample among restored integer pixel samples of a current picture, and the partial pixel sample corresponds to the offset positions of a partial pixel unit from a predetermined reference pixel sample among integer pixel samples. ## Claims ### Claim 1 An interpolation method comprising the steps of: calculating motion information for generating a predicted block; and generating fractional pixel samples on the basis of an integer sample set including integer pixel samples to which an interpolation filter is applied, wherein the integer sample set includes at least one current integer sample used to generate the fractional pixel samples out of reconstructed integer pixel samples of a current picture, and wherein the fractional pixel samples correspond to offset positions in the unit of fractional pixels from a predetermined reference pixel sample out of the integer pixel samples. ### Claim 2 (depends on 1) The interpolation method according to claim 1, wherein the interpolation method is performed on luma pixels, and wherein the step of generating the fractional pixel samples includes setting a predetermined number of current integer samples to be included in the integer sample set depending on the position of the reference pixel sample. ### Claim 3 (depends on 2) The interpolation method according to claim 2, wherein three current integer samples are included in the integer sample set when the reference pixel sample is located in a first integer sample column or row from a boundary between a reconstructed pixel sample area of a current picture and a reference sample area of a reference picture used for the interpolation, wherein two current integer samples are included in the integer sample set when the reference pixel sample is located in a second integer pixel sample column or row from the boundary, and wherein one current integer sample is included in the integer sample set when the reference pixel sample is located in a third integer pixel sample column or row form the boundary. ### Claim 4 (depends on 3) The interpolation method according to claim 3, further comprising a step of applying a smoothing filter, which uses at least one reference pixel sample and at least one reconstructed pixel sample of the current picture, to at least one current integer sample out of the current integer samples before generating the fractional pixel samples. ### Claim 5 (depends on 3) The interpolation method according to claim 3, further comprising a step of applying a smoothing filter, which uses at least one reference pixel sample and at least one reconstructed pixel sample of the current picture, to the current integer sample closest to the boundary between the reconstructed pixel sample area of the current picture and the reference sample area of the reference picture, which are used for the interpolation, before generating the fractional pixel samples. ### Claim 6 (depends on 1) The interpolation method according to claim 1, wherein the interpolation method is performed on chroma pixels, and wherein the step of generating the fractional pixel samples includes setting a predetermined number of current integer samples to be included in the integer sample set depending on the position of the reference pixel sample. ### Claim 7 (depends on 6) The interpolation method according to claim 6, wherein one current integer sample is included in the integer sample set when the reference pixel sample is located in a first integer sample column or row from a boundary between a reconstructed pixel sample area of a current picture and a reference sample area of a reference picture which are used for the interpolation. ### Claim 8 (depends on 7) The interpolation method according to claim 7, further comprising a step of applying a smoothing filter, which uses at least one reference pixel sample and at least one reconstructed pixel sample of the current picture, to at least one current integer sample out of the current integer samples before generating the fractional pixel samples. ### Claim 9 (depends on 1) The interpolation method according to claim 1, further comprising a step of determining whether an implicit or explicit instruction to use the reconstructed integer pixel samples of the current picture to generate the fractional pixel samples is transmitted from an encoder, wherein when it is determined that the instruction is transmitted, the reconstructed integer pixel samples of the current picture are used to generate the fractional pixel samples. ### Claim 10 (depends on 9) The interpolation method according to claim 9, wherein the explicit instruction is 1-bit flag information which is transmitted from the encoder to a decoder along with pixel position information and information on an array of reference pixel samples. ### Claim 11 (depends on 9) The interpolation method according to claim 9, wherein when a difference between a motion vector of a current prediction unit and a motion vector of a neighboring block of the current prediction unit is equal to or less than a predetermined threshold value, it is determined that an implicit instruction to use the reconstructed integer pixel samples of the current picture to generate the fractional pixel samples is transmitted. ### Claim 12 (depends on 1) The interpolation method according to claim 1, wherein the interpolation method is performed on luma pixels, and wherein the reconstructed integer pixel samples of the current picture are used for the interpolation when a reference integer sample is located within a third integer sample row/column from an upper or left boundary of a reference sample area of a reference picture used for the interpolation. ### Claim 13 (depends on 1) The interpolation method according to claim 1, wherein the interpolation method is performed on chroma pixels, and wherein the reconstructed integer pixel samples of the current picture are used for the interpolation when a reference integer sample is located within a first integer sample row/column from an upper or left boundary of a reference sample area of a reference picture used for the interpolation. ### Claim 14 An inter prediction method comprising the steps of: calculating a motion vector for generating a predicted block; performing an interpolation process of generating pixel samples in the unit of sub pixels when the motion vector indicates a sub pixel position; and generating the predicted block on the basis of the motion vector and the pixel samples in the unit of sub pixels, wherein reconstructed pixel samples of a current picture and pixel samples of a reference picture are together used for the interpolation process. ### Claim 15 (depends on 14) The inter prediction method according to claim 14, wherein the reconstructed pixel samples of the current picture are located in an upper or left area of a current prediction unit. ### Claim 16 (depends on 14) The inter prediction method according to claim 14, wherein a smoothing process using at least one of the reconstructed pixel samples of the current picture and at least one pixel sample of the reference picture is performed on the reconstructed pixel samples of the current picture before the step of performing the interpolation process. ### Claim 17 An interpolation method comprising the steps of: calculating a motion vector for generating a predicted block; and performing an interpolation process of generating pixel samples in the unit of sub pixels when the motion vector indicates a sub pixel position, wherein reconstructed pixel samples of a current picture and pixel samples of a reference picture are together used for the interpolation process. ### Claim 18 (depends on 17) The interpolation method according to claim 17, wherein the reconstructed pixel samples of the current picture are located in an upper or left area of a current prediction unit. ### Claim 19 (depends on 17) The interpolation method according to claim 17, wherein a smoothing process using at least one of the reconstructed pixel samples of the current picture and at least one pixel sample of the reference picture is performed on the reconstructed pixel samples of the current picture before the step of performing the interpolation process.