LTRP Reference Picture Lists for Efficient Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution and high-quality video has led to increased information transfer and storage costs due to the inefficiencies in existing video encoding and decoding processes, particularly in managing reference picture lists.
Innovation Solution
The method involves constructing a reference picture set using entropy-decoding and specifying long-term reference pictures (LTRPs) through sequence parameter sets (SPS), with flag information transmitted to improve inter prediction efficiency, and managing LTRPs using picture order counts (POCs) to optimize video encoding and decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If high-resolution and high-quality video is used, then video quality is improved, but information transfer cost and storage cost increase
Solution Approach 1:
The patent segments the reference picture list into different types (first reference picture list and second reference picture list) with different roles. The first list contains pictures for temporal prediction while the second list contains pictures for spatial prediction. This segmentation allows the system to manage large amounts of video information more efficiently by organizing it into structured categories, enabling better compression while maintaining high video quality.
Solution Approach 2:
The patent introduces flag information (e.g., `use_first_ref_pic_list_flag`, `use第二_ref_pic List flag`) to dynamically control which reference picture lists are used for different prediction operations. This parameter-based control allows the system to adaptively select from multiple reference pictures, optimizing the balance between compression efficiency and video quality without requiring all possible reference pictures to be stored and transmitted.
2Productivity
If inter prediction is performed using reference picture lists, then video compression efficiency is improved, but complexity of reference picture list management increases
Solution Approach 1:
The patent divides the reference picture list into multiple segmented lists (first reference picture list for temporal prediction, second reference picture list for spatial prediction). Each list has a specific function and can be managed independently, reducing the overall complexity of managing a single large reference picture list while maintaining the benefits of efficient inter prediction.
Solution Approach 2:
The patent performs preliminary actions by pre-defining the structure and usage rules for different reference picture lists through flag information in the bitstream. This allows the decoder to automatically understand and manage the reference picture lists without complex runtime decisions, simplifying the management process while maintaining high compression efficiency.
3Measurement precision
If more reference pictures are used for inter prediction, then prediction accuracy is improved, but information transfer cost increases
Solution Approach 1:
The patent applies local quality by differentiating between different types of reference pictures and their specific uses. Not all reference pictures are treated equally - the first reference picture list is used for temporal prediction while the second is used for spatial prediction. This localized approach allows the system to select the most appropriate reference pictures for each prediction task, improving accuracy without unnecessarily transmitting all possible reference pictures.
Solution Approach 2:
The patent uses flag information as a parameter to control which reference picture lists are activated and how they are used during prediction. By dynamically adjusting these parameters based on the specific prediction needs, the system can achieve high prediction accuracy using only the necessary reference pictures, thereby reducing information transfer costs.
Data Source
AI summary
An image decoding method according to the present invention comprises the steps of: acquiring information to form a reference picture set of a current picture by entropy decoding the received bitstream information; and performing prediction on a prediction block inside the current picture by using a reference picture list which is formed based on the reference picture set.