Fast Motion Search Logic for Multiview Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multiview video encoding is computationally expensive and power-intensive due to exhaustive macroblock searches, limiting real-time encoding capabilities, especially in battery-powered devices.
Innovation Solution
Implementing a fast motion search logic that downscales views, determines shifts between downscaled views, and encodes using these shifts, reducing processing power and time while maintaining visual quality by selecting between fast and traditional motion search logic based on user commands.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If exhaustive macroblock searches are performed to determine motion shifts between views, then encoding precision is improved, but processing time and power consumption increase significantly
Solution Approach 1:
The patent segments the encoding process into two stages: first performing exhaustive macroblock searches on downscaled (reduced-resolution) views to determine preliminary motion shifts, then applying these motion shifts to the full-resolution views. This segmentation allows the computationally intensive search to be performed on smaller data, reducing processing time while maintaining encoding precision through the subsequent refinement stage.
Solution Approach 2:
The patent introduces downscaled views as an intermediary representation between the original full-resolution views. By performing motion estimation on these intermediate downscaled versions first, the system reduces the computational burden of exhaustive searches while still achieving accurate motion compensation when applied to the original high-resolution images.
2Manufacturing precision
If exhaustive macroblock searches are performed to determine motion shifts between views, then encoding precision is improved, but power consumption increases significantly
Solution Approach 1:
The encoding process is segmented into preliminary motion estimation on downscaled views followed by refinement on full-resolution views. This segmentation reduces the number of operations requiring full processing power, thereby lowering overall energy consumption while preserving encoding quality.
Solution Approach 2:
Downscaled views serve as an energy-efficient intermediary that enables motion estimation to be performed with reduced computational resources. The intermediate results from processing these lower-resolution representations require significantly less power while still providing accurate enough motion parameters for the final encoding stage.
3Speed
If fast motion search logic with downsampling is used, then processing speed is improved, but measurement precision deteriorates
Solution Approach 1:
The patent segments the motion estimation process into two phases: a fast preliminary phase using downscaled views to obtain approximate motion shifts, followed by a refinement phase using full-resolution views to improve accuracy. This segmentation allows the system to benefit from both the speed of downsampling and the precision of full-resolution analysis.
Solution Approach 2:
The patent performs preliminary motion estimation on downscaled views before conducting the final encoding. This preliminary action provides initial motion parameters that guide subsequent more precise processing, reducing the search space and improving overall efficiency while maintaining accuracy through the two-stage approach.
Data Source
AI summary
Generally, an apparatus and a method for encoding multiview video downscale two views of the multiview video, determine a shift between the downscaled views, and encode one the two views using the shift.