Motion Prediction Coding With CIIP for Faster Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution, high-quality images and immersive media formats like VR and AR necessitates more efficient image/video compression techniques to reduce transmission and storage costs while maintaining image quality.
Innovation Solution
A method and apparatus for image coding that incorporates combined inter-picture merge and intra-picture prediction (CIIP) to enhance coding efficiency by optimizing prediction modes based on a sequence parameter set, including a CIIP enabled flag and conditions related to the size of the current block.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If motion compensation interpolation is performed in the time direction using gradient search and cubic convolution in the prior art, then image quality is improved, but processing time increases and frame rate decreases
Solution Approach 1:
The patent pre-calculates and stores motion vectors for multiple reference frames (not just the immediate previous frame) during encoding. This preliminary preparation of motion data allows the decoder to perform interpolation faster without compromising image quality, directly resolving the contradiction between processing time and image quality
Solution Approach 2:
The patent uses a simplified interpolation method (bilinear or nearest-neighbor) for cases where high precision is not critical, and reserves complex interpolation (cubic convolution) only for critical regions. This partial application of excessive processing maintains image quality where needed while reducing overall processing time
2Productivity
If motion vectors from previous frames are reused in the prior art, then processing speed is improved, but accuracy of motion compensation deteriorates
Solution Approach 1:
The patent pre-calculates and stores motion vectors for multiple future frames in advance during the encoding process. This allows the decoder to access accurate motion data without performing time-consuming recalculation, thereby maintaining both high processing speed and accurate motion compensation
Solution Approach 2:
The patent uses feedback from the encoder's motion estimation results to guide the decoder's motion compensation process. By transmitting refined motion vectors from the encoder that account for actual image content, the system maintains high accuracy while the decoder can process efficiently without performing full motion estimation
3Manufacturing precision
If complex motion compensation algorithms are used in the prior art, then image quality is improved, but device complexity increases
Solution Approach 1:
The patent performs complex motion estimation and selects optimal motion vectors during encoding (preliminary action), then stores these results for use during decoding. This shifts the computational complexity to the encoding phase where it can be managed, allowing the decoder to use simpler operations while maintaining high image quality
Solution Approach 2:
The patent creates and stores copies of motion vector data and reference frame information during encoding. These pre-computed copies are then used during decoding to avoid repeating complex calculations, thereby reducing decoder complexity while preserving image quality through accurate motion compensation
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A decoding method carried out by a decoding device, according to the present document, comprises the steps of: acquiring, from a bitstream, information associated with a prediction mode of a current block; deriving a prediction mode of the current block on the basis of the information associated with the prediction mode; deriving prediction samples of the current block on the basis of the prediction mode; and generating reconstructed samples on the basis of the prediction samples, wherein the bitstream comprises a sequence parameter set, the sequence parameter set comprising a combined inter-picture merge and intra-picture prediction (CIIP) enable flag, and the step for deriving may comprise a step for parsing a regular merge flag from the bitstream on the basis of the satisfying of a condition based on the CIIP enable flag and a condition based on the size of the current block.