Multilayer Video Motion Vector Scaling for Data Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video compression techniques face challenges in efficiently deriving and utilizing motion information across layers in a multi-layer video structure, particularly for high-resolution and stereoscopic video content, leading to increased transmission and storage costs.
Innovation Solution
A method and device for video encoding and decoding that specify a current layer criterion location and a corresponding reference layer criterion location, derive motion information from the reference layer based on a motion information storage unit size, and scale it to create a motion vector for reconstructing the current layer's picture.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion information is derived independently for each layer in a multi-layer video structure, then motion accuracy for each layer is maintained, but transmission and storage costs increase due to redundant data
Solution Approach 1:
The patent copies motion information from a reference layer to a current layer instead of deriving it independently. The motion vector of the current layer is obtained by scaling the motion vector from the corresponding block in the reference layer, significantly reducing data redundancy while maintaining acceptable motion compensation accuracy
Solution Approach 2:
The patent performs preliminary derivation of motion information in the reference layer, which is then reused and scaled for the current layer. This preliminary action avoids redundant computation and data transmission, as the reference layer's motion information serves as a foundation for deriving current layer motion vectors
2Quantity of substance
If motion information from reference layer is used for current layer, then transmission and storage costs reduce, but motion compensation precision may deteriorate
Solution Approach 1:
The patent applies parameter changes by scaling the motion vector from the reference layer according to the resolution ratio between reference layer and current layer. This scaling operation adjusts the motion vector magnitude to match the current layer's resolution, maintaining motion compensation accuracy while utilizing reference layer data
Solution Approach 2:
The patent introduces a scaling factor as an intermediary between the reference layer motion vector and the current layer motion vector. This intermediary element (scaling factor based on resolution ratio) bridges the gap between layers with different resolutions, enabling accurate motion compensation without requiring independent motion estimation
3Manufacturing precision
If high-resolution video is transmitted without compression, then video quality is maintained, but transmission and storage costs increase significantly
Solution Approach 1:
The patent segments video coding into multiple layers (reference layer and current layer), where the reference layer is coded at lower resolution and the current layer contains enhancement information. This segmentation allows efficient compression while maintaining high-resolution video quality through layered reconstruction
Solution Approach 2:
The patent adds a layer dimension to video coding by using multi-layer structure. Instead of compressing a single high-resolution layer, it creates a hierarchy where a base reference layer provides structural information and additional current layers provide enhancement, effectively utilizing the layer dimension for compression
Data Source
AI summary
The present invention relates to a method for encoding a video and a method for decoding a video in a multilayer structure, and an apparatus using the same. The method for decoding the video according to the present invention comprises the steps of: specifying a current layer reference location for specifying a current block from a current layer; specifying a reference layer reference location corresponding to the reference location of the current layer, from the reference layer; inducing motion information from the reference layer reference location based on the size of motion information saving units; and scaling the induced motion information and inducing the same into a motion vector used for recovering a picture in the current layer.