IP Library › Granted Patent US 12,732,640
Granted Patent B2
US 12,732,640 · App. 19/117,247 · Granted Sep 8, 2026

Method and apparatus of inheriting cross-component model parameters in video coding system

Inventors: Chia-Ming Tsai (Hsinchu City, TW); Chih-Wei Hsu (Hsinchu City, TW); Yi-Wen Chen (San Jose, CA); Ching-Yeh Chen (Hsinchu City, TW); Tzu-Der Chuang (Hsinchu City, TW); Cheng-Yen Chuang (Hsinchu City, TW)
Assignee: MEDIATEK INC.
H04N19/70H04N19/105H04N19/176H04N19/186H04N19/593
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 12,732,640
App. No.
19/117,247
Granted
Sep 8, 2026
Kind
B2
Abstract

Methods and apparatus for inheriting model parameters for cross-component prediction modes. According to one method, a target model parameter set associated with a target inherited prediction model is derived based on an inherited model parameter set associated with the target inherited prediction model selected from the prediction candidate list, where the target model parameter set comprises at least one re-derived or refined parameter being re-derived or refined from the inherited model parameter set, and wherein said at least one re-derived or refined parameter comprises a CCLM offset parameter when the target inherited prediction model corresponds to CCLM (Cross Component Linear Model), or two separate inherited model parameter sets are used for a second-colour block. At least one second-colour block is encoded or decoded using prediction data comprising cross-colour prediction generated by applying the target inherited prediction model with the target model parameter set to reconstructed first-colour block.

Claims (30)

1 . A method of coding colour pictures using coding tools including one or more cross component model related modes, the method comprising:

receiving input data associated with a current block comprising a first-colour block and one or more second-colour blocks, wherein the input data comprises pixel data to be encoded at an encoder side or data associated with the current block to be decoded at a decoder side;

determining a prediction candidate list comprising one or more inherited cross-component prediction candidates;

deriving one or more target model parameter sets associated with a target inherited prediction model based on one or more inherited model parameter sets associated with the target inherited prediction model selected from the prediction candidate list, wherein when the target inherited prediction model corresponds to GLM (Gradient Linear Model), one or more gradient patterns associated with the GLM are also inherited, or when the target inherited prediction model corresponds to MMLM (Multiple Model CCLM Mode) or CCCM (Convolutional Cross-Component Model) with multi-model, a classification threshold associated with the MMLM or the CCCM with multi-mode is also inherited; and

encoding or decoding at least one of said one or more second-colour blocks using prediction data comprising cross-colour prediction generated by applying the target inherited prediction model with said one or more target model parameter sets to reconstructed first-colour block.

2 . The method of claim 1 , wherein when the target inherited prediction model corresponds to the MMLM or the CCCM with multi-model, the classification threshold is used to classify neighbouring samples of the current block into multiple groups.

3 . The method of claim 2 , wherein one inherited model parameter set is assigned to each of the multiple groups.

4 . The method of claim 3 , wherein an offset parameter of each group is re-derived based on an inherited scaling parameter and an average value of neighbouring first-colour and second-colour samples of each group of the current block.

5 . A method of coding colour pictures using coding tools including one or more cross component models related modes, the method comprising:

receiving input data associated with a current block comprising a first-colour block and one or more second-colour blocks, wherein the input data comprise pixel data to be encoded at an encoder side or data associated with the current block to be decoded at a decoder side;

deriving a prediction candidate list comprising one or more inherited cross-component prediction candidates;

checking similarity among said one or more inherited cross-component prediction candidates and one or more existing candidates in the prediction candidate list before said one or more inherited cross-component prediction candidates are inserted into the prediction candidate list;

inserting a target inherited cross-component prediction candidate into the prediction candidate list only if a target similarity among the target inherited cross-component prediction candidate and one existing candidate in the prediction candidate list satisfies a condition;

deriving a target predictor from the prediction candidate list; and

encoding or decoding at least one of said one or more second-colour blocks using the target predictor.

6 . The method of claim 5 , wherein the condition corresponds to the similarity being larger than a threshold.

7 . The method of claim 6 , wherein the similarity is measured by comparing model parameters of two inherited cross-component prediction candidates.

8 . The method of claim 7 , wherein the target inherited cross-component prediction candidate is not inserted into the prediction candidate list if the model parameters of the target inherited cross-component prediction candidate is the same as the model parameters of another inherited cross-component prediction candidate.

9 . An apparatus for video coding, the apparatus comprising one or more electronics or processors arranged to:

receive input data associated with a current block comprising a first-colour block and one or more second-colour blocks, wherein the input data comprise pixel data to be encoded at an encoder side or data associated with the current block to be decoded at a decoder side;

determine a prediction candidate list comprising one or more inherited cross-component prediction candidates;

derive one or more target model parameter sets associated with a target inherited prediction model based on one or more inherited model parameter sets associated with the target inherited prediction model selected from the prediction candidate list, wherein when the target inherited prediction model corresponds to GLM (Gradient Linear Model), one or more gradient patterns associated with the GLM are also inherited, or when the target inherited prediction model corresponds to MMLM (Multiple Model CCLM Mode) or CCCM (Convolutional Cross-Component Model) with multi-model, a classification threshold associated with the MMLM or the CCCM with multi-mode is also inherited; and

encode or decode at least one of said one or more second-colour blocks using prediction data comprising cross-colour prediction generated by applying the target inherited prediction model with said one or more target model parameter sets to reconstructed first-colour block.

10 . An apparatus for video coding, the apparatus comprising one or more electronics or processors arranged to:

receive input data associated with a current block comprising a first-colour block and a second-colour block, wherein the input data comprise pixel data to be encoded at an encoder side or coded data associated with the current block to be decoded at a decoder side;

derive a prediction candidate list comprising one or more inherited cross-component prediction candidates;

check similarity among said one or more inherited cross-component prediction candidates and one or more existing candidates in the prediction candidate list before said one or more inherited cross-component prediction candidates are inserted into the prediction candidate list;

insert a target inherited cross-component prediction candidate into the prediction candidate list only if a target similarity among the target inherited cross-component prediction candidate and one existing candidate in the prediction candidate list satisfies a condition;

derive a target predictor from the prediction candidate list; and

encode or decode at least one of said second-colour blocks using the target predictor.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Mar 31, 2025
From: TSAI, CHIA-MING; HSU, CHIH-WEI; CHEN, YI-WEN; CHEN, CHING-YEH; CHUANG, TZU-DER; CHUANG, CHENG-YEN
To: MEDIATEK INC.
Reel/Frame 070683/0737 →
Continuity (3)
Provisional Application 63380762 · Oct 25, 2022
Provisional Application 63378707 · Oct 7, 2022
Related Publication 20260107020A1 · Apr 16, 2026
References Cited (34)
US 10708591B2 · Zhang · 2020 [cited by examiner]
US 11172216B1 · Zhang · 2021 [cited by examiner]
US 11546592B2 · Zhao · 2023 [cited by examiner]
US 11910020B2 · Zhang · 2024 [cited by examiner]
US 12101472B2 · Racape · 2024 [cited by examiner]
US 12108090B2 · Zhang · 2024 [cited by examiner]
US 12143616B2 · Zhang · 2024 [cited by examiner]
US 12531994B2 · Tsai · 2026 [cited by examiner]
US 20150172693A1 · Yang · 2015 [cited by examiner]
US 20150195571A1 · Boyce · 2015 [cited by examiner]
US 20180109814A1 · Chuang · 2018 [cited by applicant]
US 20190238877A1 · Sakurai · 2019 [cited by examiner]
US 20190364278A1 · Lee · 2019 [cited by examiner]
US 20190387226A1 · Lee · 2019 [cited by examiner]
US 20200186805A1 · Lee · 2020 [cited by examiner]
US 20200275129A1 · Deshpande · 2020 [cited by examiner]
US 20210014479A1 · Lee · 2021 [cited by examiner]
US 20210029356A1 · Zhang · 2021 [cited by examiner]
US 20210058637A1 · Zhang · 2021 [cited by examiner]
US 20210092431A1 · Zhang · 2021 [cited by examiner]
US 20210227229A1 · Ma · 2021 [cited by applicant]
US 20210235101A1 · Sim · 2021 [cited by examiner]
US 20210337197A1 · Lee · 2021 [cited by examiner]
US 20210400142A1 · Jorasch · 2021 [cited by examiner]
US 20210400310A1 · Zhang · 2021 [cited by applicant]
US 20220053194A1 · Zhao · 2022 [cited by examiner]
US 20220086439A1 · Tsai · 2022 [cited by examiner]
US 20220283433A1 · Christmas · 2022 [cited by examiner]
US 20230179805A1 · He · 2023 [cited by examiner]
US 20250126284A1 · Jhu · 2025 [cited by examiner]
US 20250168366A1 · Francois · 2025 [cited by examiner]
WO 2018183533A1 · 2018 [cited by applicant]
WO 2020096877A1 · 2020 [cited by applicant]
International Search Report and Written Opinion dated Dec. 19, 2023, issued in application No. PCT/CN2023/123199. [cited by applicant]