Reference Frame Storage for Lower-Memory Video Codecs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video codecs require large buffered-frame memory and excessive data access for motion estimation, leading to high memory usage and power consumption, particularly in standards like H.264, which deteriorate video quality and fail to adaptively manage storage types of blocks.
Innovation Solution
A method for a video codec that determines storage types and compression data storage methods based on the content of compressed video signals and estimated unused memory, using five schemes for storage type determination and four compression data storage methods, allowing for efficient storage of reference frames using either compressed or reconstructed data to balance memory and computational requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If reference frames are stored using reconstructed data in the buffered-frame storage device, then motion estimation can be performed with high accuracy, but the memory size and data access amount significantly increase
Solution Approach 1:
The reference frame storage is segmented into two types: compressed data storage and reconstructed data storage. Different blocks of reference frames are stored in different formats depending on their usage patterns. This segmentation allows the system to maintain high motion estimation accuracy for frequently accessed blocks while using compressed storage for less critical blocks, thereby reducing overall memory requirements.
Solution Approach 2:
Different storage qualities are applied to different regions of the reference frame based on local requirements. Blocks that are frequently referenced for motion estimation are stored in full reconstructed form, while blocks that are less frequently accessed are stored in compressed form. This local quality differentiation optimizes the balance between motion estimation accuracy and memory usage.
2Reliability
If reference frames are stored using reconstructed data, then video quality is maintained, but power consumption increases due to excessive data access
Solution Approach 1:
The reference frame data is segmented into compressed and reconstructed portions. By segmenting the storage format rather than using a uniform approach, the system reduces the total amount of data that needs to be accessed and processed, thereby lowering power consumption while maintaining video quality in the reconstructed portions.
Solution Approach 2:
The storage parameter (compressed vs. reconstructed) is changed dynamically based on the reference frame block's characteristics and usage patterns. This parameter change allows the system to optimize between power consumption and video quality by selecting the appropriate storage format for each block.
3Productivity
If H.264 codec references up to five previous frames for motion estimation, then compression ratio is improved, but the amount of data accessed and memory required increases
Solution Approach 1:
The five reference frames are segmented into different storage formats based on their importance and usage patterns. Not all five frames need to be stored in full reconstructed form - some can be stored in compressed form or selectively reconstructed, reducing the total data access amount while maintaining the compression benefits of referencing multiple frames.
Solution Approach 2:
Instead of fully reconstructing and storing all five reference frames, the system applies partial reconstruction only to the extent necessary for maintaining compression performance. Some reference frames or blocks within frames are stored in compressed form, applying partial action rather than complete reconstruction.
4Quantity of substance
If adaptive storage type determination is implemented for different blocks, then memory efficiency is improved, but device complexity increases
Solution Approach 1:
The storage type determination is made dynamically based on block characteristics and usage patterns rather than using a static uniform approach. This dynamic adaptation allows the system to optimize memory efficiency by selecting appropriate storage formats for different blocks while managing complexity through rule-based decision making.
Data Source
AI summary
A method for reducing memory size and amount of data accessed of a buffered-frame storage device in a video codec is provided. It is applicable to a codec compliant with a video encoding and decoding standard. When a block is decompressed, the block is stored in the buffered-frame storage device using either compression data or reconstructed data based on its encoded mode and the positions of its reference frames, or based on a topological pattern and a pre-determined memory capacity constraint. With this method, not only the memory size of the buffered-frame storage device, but also the computational complexity requirement for decoding compression data of reference frames is reduced. The present invention compromises among the buffered-frame memory size, amount of data accessed, and computational complexity requirement.