Global Motion Estimation Using Least Median of Squares
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression techniques, such as MPEG, face challenges in accurately estimating global motion between frames due to noise in motion vector information, which complicates the generation of panoramic images and increases computational burden.
Innovation Solution
A robust least median of squares approach is employed to estimate global motion between frames, allowing for accurate motion estimation even with up to 50% noisy motion vectors, by selecting and processing sets of motion vectors to exclude noisy data and calculate median squared error values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If conventional motion-compensated coding is used to compress video data, then storage and network bandwidth requirements are reduced, but motion vector information becomes noisy and unreliable for global motion estimation
Solution Approach 1:
The patent uses an affine transformation model as an intermediary framework to reinterpret noisy motion vectors from compressed video. By fitting motion vectors to an affine model that represents global camera motion, the system extracts reliable global motion parameters despite the noise introduced by compression. This intermediary model bridges the gap between compressed representation and accurate motion estimation.
Solution Approach 2:
The patent replaces direct use of raw motion vectors with a statistical approach using least median of squares estimation. Instead of relying on individual motion vector measurements, the system substitutes a robust statistical method that processes the distribution of motion vectors to extract accurate global motion parameters, effectively replacing mechanical measurement with statistical inference.
2Productivity
If all motion vectors from compressed video frames are used for global motion estimation, then computational efficiency is improved, but estimation accuracy deteriorates due to noisy motion vector data
Solution Approach 1:
The patent extracts only the essential global motion information from motion vectors by fitting them to an affine transformation model. Rather than processing all motion vector details, the system extracts the six parameters representing global camera motion, effectively separating useful global motion information from noisy local motion details introduced by compression.
Solution Approach 2:
The patent changes the parameter representation from individual motion vectors to six affine transformation parameters. By transforming the problem from processing numerous noisy motion vector parameters to estimating a small set of global motion parameters, the system achieves both computational efficiency and accuracy through parameter dimensionality reduction and reparameterization.
3Device complexity
If traditional least mean squares method is used for motion estimation, then computational complexity is reduced, but reliability of motion estimation deteriorates in the presence of noisy motion vectors
Solution Approach 1:
The patent converts the harmful effect of noisy motion vectors into a benefit by using the distribution of noisy vectors as input to a robust estimation algorithm. The least median of squares method transforms the presence of outliers and noise into useful statistical information, where the median property naturally filters out corrupted motion vectors while preserving accurate global motion signals.
Data Source
AI summary
Global motion between frames of a motion-compensated inter-frame encoded video sequence is estimated directly from the motion vectors encoded within the frames. For any particular frame, the motion vectors are first decoded, and a finite number of sets of vectors are selected. An affine or other geometrical transform is then used to generate a motion estimation for each set, and then the least median squared error present in each motion estimation is calculated for each estimation. The motion estimation with the smallest least median squared error is then selected as being representative of the global motion in the image of the frame. A panoramic image generating method and system makes uses of the global motion estimations thus obtained.