3D Video Encoding Method Sharing Search Window Pixel Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The encoding of 3D video requires reading pixel data from memory for reference frames of different view angles, leading to increased memory bandwidth demand.
Innovation Solution
The encoding method and device execute motion estimations by sharing pixel data of a search window across multiple frames with different view angles, allowing for sequential or simultaneous processing of macro blocks and search windows to reduce memory bandwidth requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If pixel data of reference frames belonging to different view angles are read from memory for motion estimation, then the encoding accuracy is improved, but the memory bandwidth demand increases
Solution Approach 1:
The patent merges the processing of multiple view angle frames by executing motion estimations for different view angles simultaneously or sequentially while sharing the same search window pixel data from a reference frame, thereby consolidating memory access operations and reducing overall bandwidth demand
Solution Approach 2:
The reference frame pixel data serves multiple functions by being reused across different motion estimation operations for multiple view angles, allowing a single data loading operation to support multiple encoding tasks simultaneously
2Adaptability or versatility
If two reference frames are required for encoding one current frame from different view angles, then the encoding completeness is improved, but the memory bandwidth demand increases
Solution Approach 1:
The patent loads the search window pixel data from the reference frame into a buffer memory in advance, before it is needed for multiple motion estimation operations, thereby preparing the data for reuse and avoiding repeated memory access operations
Solution Approach 2:
The patent introduces a buffer memory as an intermediary component between the main memory and the motion estimation units, allowing pixel data to be cached and shared across multiple view angle encodings without requiring continuous access to the main memory
Data Source
AI summary
An encoding method and an encoding device for 3D video are provided. The method includes the following. Pixel data of a first macro block in a first frame to be encoded is loaded, wherein the first frame belongs to a first view angle. Pixel data of a search window in a reference frame is loaded. A first motion estimation is executed according to the pixel data of the search window and the first macro block. Pixel data of a second macro block in a second frame to be encoded is loaded, wherein the second frame belongs to a second view angle different from the first view angle. A second motion estimation is executed according to the pixel data of the search window and the second macro block, thereby sharing the pixel data of the search window with the first motion estimation. Accordingly, the demand for a memory bandwidth is decreased.