Video Decoder Reference Line Selection for TMRL Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital video compression standards face challenges in reducing bandwidth and traffic pressure during video transmission, necessitating improved decoding performance to enhance efficiency.
Innovation Solution
A decoding method that determines the maximum line quantity of reference lines for a current block, allowing the use of a template-based multiple reference line (TMRL) mode, and performs prediction based on the determined reference line and mode to improve decoding performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If existing digital video compression standards are used, then video transmission and storage can be achieved, but bandwidth and traffic pressure cannot be effectively reduced
Solution Approach 1:
The patent changes the parameter of reference line quantity from fixed to variable, allowing the decoder to adaptively determine the maximum number of reference lines based on block characteristics. This parameter optimization enables better compression efficiency without sacrificing decoding performance, directly addressing the contradiction between reducing bandwidth pressure and maintaining productivity.
2Productivity
If the number of reference lines is increased to improve prediction accuracy, then decoding performance improves, but system complexity increases
Solution Approach 1:
The patent introduces dynamic adaptation by allowing the decoder to determine the maximum line quantity of reference lines based on the characteristics of the current block. This dynamic approach enables the system to use more reference lines when beneficial for prediction accuracy while using fewer reference lines when not necessary, thus improving decoding performance without consistently increasing system complexity.
Solution Approach 2:
The patent applies different reference line quantities to different blocks based on their specific characteristics. Instead of using a uniform number of reference lines for all blocks, the system locally optimizes the reference line quantity for each block, allowing high-complexity processing only where it provides the most benefit to decoding performance.
3Measurement precision
If template-based multiple reference line mode is always used, then prediction accuracy improves, but processing overhead increases
Solution Approach 1:
The patent implements partial use of the template-based multiple reference line mode by first determining whether to use the mode based on block characteristics. Rather than always applying the computationally intensive template matching approach, the system selectively applies it only when the determined maximum line quantity and block characteristics indicate it will provide meaningful prediction accuracy improvement, thereby reducing processing overhead while maintaining prediction accuracy where needed.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
Embodiments of the present application provide a decoding method, an encoding method, a decoder, and an encoder. The decoding method comprises: determining a maximum line number of reference lines allowed to be used by a current block; on the basis of the maximum line number, determining whether the current block uses a template-based multiple reference line (TMRL) mode; if the current block uses the TMRL mode, determining a reference line used by the current block and a prediction mode used by the current block; on the basis of the reference line used by the current block, predicting the current block using the prediction mode used by the current block, to obtain a prediction block. In the present embodiment, when determining whether to use the TMRL mode to predict the current block, the maximum line quantity must be used as a condition to determine whether the current block uses the TMRL mode, so that when the current block uses the TMRL mode, and the maximum line quantity can meet a use condition, decoding performance can be improved.