Picture Order Count Derivation for VVC Decoder Interoperability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards, particularly in the Versatile Video Coding (VVC) text, face issues with the handling of discardable pictures and access units (AUs), leading to confusion and interoperability problems due to incorrect semantics and format rules.
Innovation Solution
The proposed solution involves modifying the format rules and semantics in the VVC specification to address specific issues, such as making certain variables and messages AU-specific, updating the derivation of picture order counts, and correcting the handling of random access skipped leading (RASL) and random access decodable leading (RADL) pictures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing video coding standards are used to handle discardable pictures and access units, then video coding can be performed, but confusion and interoperability problems occur due to incorrect semantics and format rules
Solution Approach 1:
The patent modifies specific parameters and flags in the video coding standard, such as changing the semantics of ph_non_ref_pic_flag and its interaction with other flags like delta_poc_msb_cycle_present_flag. These parameter changes correct the incorrect semantics and format rules that were causing interoperability problems and decoder crashes.
2Reliability
If the derivation of picture order count does not consider ph_non_ref_pic_flag, then decoding process is simpler, but incorrect POC values are generated causing decoder crashes
Solution Approach 1:
The patent applies preliminary action by checking the ph_non_ref_pic_flag before performing the derivation of picture order count. This preliminary check ensures that incorrect POC values are not generated in the first place, preventing decoder crashes while maintaining a relatively simple derivation process.
3Productivity
If RASL pictures are incorrectly handled as reference pictures, then more pictures can be used for prediction, but incorrect decoding behavior occurs
Solution Approach 1:
The patent inverts the conventional approach by explicitly preventing RASL pictures from being used as reference pictures, rather than allowing them and then correcting errors. This is achieved through modified semantics of flags that control reference picture selection, ensuring decoding correctness while maintaining prediction efficiency through proper use of valid reference pictures.
Data Source
AI summary
Methods, systems and devices for coding the picture order count in video bitstream processing. An example method of video processing includes performing a conversion between a video comprising one or more pictures and a bitstream of the video, wherein the bitstream conforms to a format rule, wherein the format rule specifies that a derivation of a picture associated with a first flag and in a decoding process for a picture order count is based on a second flag, wherein the picture associated with the first flag is a previous picture in a decoding order that has (i) a first identifier that is the same as a slice or picture header referring to a reference picture list syntax structure, (ii) a second identifier and the second flag being equal to zero, and (iii) a picture type different from a random access skipped leading picture and a random access decodable leading picture.