IP Library › Granted Patent US 12,407,818
Granted Patent B2
US 12,407,818 · App. 18/241,358 · Granted Sep 2, 2025

Method and apparatus for video coding using derivation of intra prediction mode

Inventors: Joo Hyung Byeon (Seoul, KR); Sea Nae Park (Seoul, KR); Dong Gyu Sim (Seoul, KR); Seung Wook Park (Yongin-si, KR)
Assignees: HYUNDAI MOTOR COMPANY; KIA CORPORATION; KWANGWOON UNIVERSITY INDUSTRY-ACADEMIC COLLABORATION FOUNDATION
H04N19/11H04N19/105H04N19/119H04N19/132H04N19/136H04N19/176H04N19/182H04N19/593H04N19/70
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,407,818
App. No.
18/241,358
Granted
Sep 2, 2025
Kind
B2
Abstract

A method and an apparatus for video coding using derivation of intra prediction modes are disclosed. The video coding method and the video coding apparatus derive the intra prediction mode of a current block using pre-restored, neighboring reference sample values and generate a prediction block of the current block based on the derived prediction mode.

Claims (66)

1. A method performed by a video decoding apparatus for intra prediction of a current block, the method comprising:

parsing a prediction mode derivation flag indicating whether to derive at least one prediction mode of the current block from a bitstream; and

checking the prediction mode derivation flag,

wherein, when the prediction mode derivation flag is true, the method includes:

determining a calculation area used for calculating gradient values from restored neighboring samples of the current block;

calculating a gradient histogram of directional modes in the calculation area for the current block;

deriving the at least one prediction mode of the current block based on the gradient histogram; and

generating a prediction block of the current block by performing intra prediction using the at least one derived prediction mode, and

wherein calculating the gradient histogram comprises:

calculating vertical and horizontal gradient values for restored reference samples included in the calculation area;

calculating gradient directions and gradient magnitudes at the restored reference samples using the vertical and horizontal gradient values;

calculating directional modes of intra prediction closest to the gradient directions; and

accumulating the gradient magnitudes in the gradient histogram corresponding to the directional modes.

2. The method of claim 1 , further comprising:

parsing, from the bitstream, one or a plurality of flags that indicate whether the at least one prediction mode of the current block is a non-directional mode,

wherein, when all the flags indicating one of the non-directional modes are false, the prediction mode derivation flag is parsed.

3. The method of claim 1 , wherein, when the prediction mode derivation flag is true, a subblock partitioning application flag is parsed from a bitstream to determine whether to perform subblock partitioning of the current block.

4. The method of claim 1 , wherein determining the calculation area comprises:

determining three rows of restored reference pixels on left and top of the current block as the calculation area, and

wherein a length of the calculation area located on left and top of the current block is set based on a width and a height of the current block.

5. The method of claim 1 , further comprising:

parsing a flag indicating the calculation area from the bitstream to set the calculation area between left and upper parts of the current block according to the flag indicating the calculation area.

6. The method of claim 1 , wherein calculating the gradient histogram comprises:

calculating the vertical and horizontal gradient values for restored reference samples located on a top second line and a left second line based on the current block by using a preset boundary detection filter.

7. The method of claim 1 , wherein deriving the at least one prediction mode comprises:

determining a first direction mode having a largest value from the gradient histogram as the at least one prediction mode of the current block.

8. The method of claim 1 , wherein deriving the at least one prediction mode comprises:

determining a first directional mode having a first largest value and a second directional mode having a second largest value from the gradient histogram as a first prediction mode of the current block and a second prediction mode of the current block.

9. The method of claim 7 , wherein deriving the at least one prediction mode comprises:

when a histogram value of the first directional mode is smaller than a preset first threshold or a sum of values from the gradient histogram is smaller than a preset second threshold, determining a non-directional prediction mode as the at least one prediction mode of the current block.

10. The method of claim 7 , wherein deriving the at least one prediction mode comprises:

when a ratio between a histogram value of the first directional mode and a sum of gradient histogram values is smaller than a preset ratio, determining a non-directional mode as the at least one prediction mode of the current block.

11. The method of claim 3 , wherein, when the subblock partitioning application flag is true, the current block is partitioned into subblocks, and a derived prediction mode of the current block is shared as the same intra prediction mode of the subblocks.

12. The method of claim 3 , wherein, when the subblock partitioning application flag is true, the current block is partitioned into subblocks, and a prediction mode of a current subblock is derived based on restored samples of a restored previous subblock, and

wherein whether to derive a prediction mode for each subblock is implicitly determined based on sizes of the subblocks.

13. The method of claim 8 , wherein generating the prediction block comprises:

generating the prediction block of the current block by generating a first prediction block of the current block using the first directional mode, generating a second prediction block of the current block using the second directional mode, and performing weighted averaging on the first prediction block and the second prediction block.

14. The method of claim 13 , wherein generating the prediction block comprises:

when the first prediction mode of the current block and the second prediction mode of the current block are directional modes, generating a third prediction block of the current block by additionally using a non-directional mode, and

wherein the prediction block of the current block is generated by performing weighted averaging on the first prediction block, the second prediction block, and the third prediction block.

15. A method performed by a video encoding apparatus for intra prediction of a current block, the method comprising:

obtaining a prediction mode derivation flag indicating whether to derive at least one prediction mode of the current block; and

checking the prediction mode derivation flag,

wherein, when the prediction mode derivation flag is true, the method includes:

determining a calculation area used for calculating gradient values from restored neighboring samples of the current block;

calculating a gradient histogram of directional modes in the calculation area for the current block;

deriving the at least one prediction mode of the current block based on the gradient histogram; and

generating a prediction block of the current block by performing intra prediction using the derived at least one prediction mode, and

wherein calculating the gradient histogram comprises:

calculating vertical and horizontal gradient values for restored reference samples included in the calculation area;

calculating gradient directions and gradient magnitudes at the restored reference samples using the vertical and horizontal gradient values;

calculating directional modes of intra prediction closest to the gradient directions; and

accumulating the gradient magnitudes in the gradient histogram corresponding to the directional modes.

16. A method of storing a bitstream of a video into a non-transitory computer-readable recording medium, wherein the bitstream is generated by an encoding method for intra predicting a current block performed by a video encoding apparatus, wherein the method comprises:

obtaining a prediction mode derivation flag indicating whether to derive at least one prediction mode of the current block; and

checking the prediction mode derivation flag,

wherein, when the prediction mode derivation flag is true, the method includes:

determining a calculation area used for calculating gradient values from restored neighboring samples of the current block;

calculating a gradient histogram of directional modes in the calculation area for the current block;

deriving the at least one prediction mode of the current block based on the gradient histogram; and

generating a prediction block of the current block by performing intra prediction using the at least one derived prediction mode, and

wherein calculating the gradient histogram comprises:

calculating vertical and horizontal gradient values for restored reference samples included in the calculation area;

calculating gradient directions and gradient magnitudes at the restored reference samples using the vertical and horizontal gradient values;

calculating directional modes of intra prediction closest to the gradient directions; and

accumulating the gradient magnitudes in the gradient histogram corresponding to the directional modes.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 1, 2023
From: BYEON, JOO HYUNG; PARK, SEA NAE; SIM, DONG GYU; PARK, SEUNG WOOK
To: HYUNDAI MOTOR COMPANY; KIA CORPORATION; KWANGWOON UNIVERSITY INDUSTRY-ACADEMIC COLLABORATION FOUNDATION
Reel/Frame 064775/0188 →
Priority Claims (2)
KR 10-2021-0028795 · Mar 4, 2021 · national
KR 10-2022-0026549 · Mar 2, 2022 · national
Continuity (2)
Continuation PCTKR2022002985 · Mar 3, 2022
Related Publication 20230412799A1 · Dec 21, 2023
References Cited (21)
US 10219003B2 · Huang et al. · 2019 [cited by applicant]
US 11252403B2 · Lee · 2022 [cited by applicant]
US 20110002385A1 · Kobayashi · 2011 [cited by examiner]
US 20160014421A1 · Cote · 2016 [cited by examiner]
US 20170078696A1 · Huang et al. · 2017 [cited by applicant]
US 20190110076A1 · Lim · 2019 [cited by examiner]
US 20190215521A1 · Chuang et al. · 2019 [cited by applicant]
US 20200296356A1 · Mora · 2020 [cited by examiner]
US 20210227208A1 · Lee · 2021 [cited by applicant]
US 20210377568A1 · Koo · 2021 [cited by examiner]
US 20220094912A1 · Lee · 2022 [cited by applicant]
US 20220224922A1 · Wang · 2022 [cited by examiner]
US 20230109828A1 · Lee · 2023 [cited by examiner]
US 20240137560A1 · Kim · 2024 [cited by examiner]
KR 101909170B1 · 2018 [cited by applicant]
KR 20200041810A · 2020 [cited by applicant]
“Decoder-Side Intra Mode Derivation With Texture Analysis in VVC Test Model”—Nasrallah et al., 2019 IEEE International Conference on Image Processing (ICIP); Date Added to IEEE Xplore: Aug. 26, 2019 (Year: 2019). [cited by examiner]
“Gradient-Based Intraprediction Fusion for Video Coding”—Abdoli et al., IEEE MultiMedia (vol. 28, Issue: 3, Jul. 1-Sep. 2021); Date of Publication: Dec. 3, 2020 (Year: 2020). [cited by examiner]
Abdoli, Mohsen, et al., Decoder-Side Intra Mode Derivation For Next Generation Video Coding, 2020 IEEE International Conference on Multimedia and Expo (ICME), Jul. 6-10, 2020, 7 pp, DOI: 10.1109/ICME46284.2020.9102799. [cited by applicant]
Abdoli, Mohsen, et al., Non-CE3: Decoder-side Intra Mode Derivation with Prediction Fusion Using Planar, Joint Video Experts Team (JVET) of ITU-T SG 16 WP 3 and ISO/IEC JTC 1/SC 29/WG 11 15th Meeting: Gothenburg, SE, Ju… [cited by applicant]
International Search Report and Written Opinion cited in corresponding international app No. PCT/KR2022/002985 ; Jun. 14, 2022; 11 pp. [cited by applicant]