Scalable Video Motion Estimation via Optical Flow
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motion estimation techniques, such as block matching, face difficulties in scalable video coding, especially when applied to wavelet transforms, leading to inefficiencies and visual artifacts due to discontinuities, and fail to accurately represent different space resolutions effectively.
Innovation Solution
The optical flow technique is adapted for scalable coding by computing motion fields at the lowest resolution and then iteratively refining them for higher resolutions with a regularization term that measures similarity to the lower resolution solutions, using wavelet filtering and sub-sampling to constrain the motion field and distribute errors across resolutions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If block matching is used for motion estimation in scalable video coding with wavelet transforms, then motion estimation can be performed, but visual artifacts and discontinuities occur at block edges
Solution Approach 1:
The patent replaces the mechanical block matching approach with an optical flow-based motion estimation method. Instead of dividing the image into blocks and matching them, the optical flow technique computes motion fields continuously across the entire image, eliminating block edges and associated visual artifacts while maintaining compatibility with wavelet transforms.
Solution Approach 2:
The patent changes the fundamental parameters of motion estimation by transitioning from discrete block-based matching to continuous optical flow computation. This involves solving a system of linear equations with regularization terms, transforming the motion representation from piecewise constant block vectors to smooth continuous motion fields that are compatible with wavelet decomposition.
2Object-affected harmful factors
If optical flow is used for motion estimation, then smooth motion fields are produced suitable for wavelet transforms, but computational complexity increases
Solution Approach 1:
The patent applies multi-resolution analysis by segmenting the motion estimation process across different resolution levels. Motion fields are computed iteratively from coarse to fine resolutions, with each level providing a refined approximation. This segmentation reduces computational complexity at each stage while maintaining the smoothness required for wavelet transform compatibility.
Solution Approach 2:
The patent performs preliminary motion estimation at lower resolutions before refining at higher resolutions. By computing motion fields at coarser levels first and using them as initial estimates for finer levels, the method reduces the computational burden of solving the full-resolution optical flow equations while still achieving smooth, accurate motion fields suitable for wavelet transforms.
3Measurement precision
If motion fields are computed for each resolution level independently, then each level optimizes its own motion estimation, but scalability and error distribution across resolutions are compromised
Solution Approach 1:
The patent merges the motion estimation across multiple resolution levels by computing motion fields iteratively from coarse to fine resolutions. Each resolution level's motion field is computed considering the previously computed lower-resolution field, creating a unified multi-resolution motion representation that maintains scalability and allows error distribution across all levels while preserving accuracy at each individual level.
Data Source
AI summary
A method for estimating motion for the scalable video coding, includes the step of estimating the motion field of a sequence of photograms which can be represented with a plurality of space resolution levels including computing the motion field for the minimum resolution level and, until the maximum resolution level is reached, repeating the steps of: rising by one resolution level; extracting the photograms for such resolution level; and computing the motion field for such resolution level. The motion field is computed through an optical flow equation which contains, for every higher level than the minimum resolution level, a regularization factor between levels which points out the difference between the solution for the considered level and the solution for the immediately lower resolution level. A more or less high value of the regularization factor implies more or less relevant changes of the component at the considered resolution during the following process iterations.