Panorama Motion Estimation via Border Padding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional motion estimation techniques are ineffective for panorama images with 360° omni-directional views due to the high spatial relation between the right and left borders, leading to inaccuracies in motion vector estimation and compensation.
Innovation Solution
The method involves padding regions connected to the left and right sides of a basic reference frame using border pixel values, expanding the frame, and adjusting sub-pixel coordinates to determine similarity using evaluation functions like SAD, SATD, or SSD, ensuring accurate motion vector estimation and compensation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If conventional motion estimation technique is applied to panorama image, then the processing method is simple, but the motion vector estimation accuracy deteriorates due to high spatial relation between borders
Solution Approach 1:
The patent applies preliminary action by padding the reference frame with border pixel values before performing motion estimation. This preprocessing step ensures that when motion vectors point to border regions, the referenced pixel values are already available, preventing estimation errors caused by unavailable data and improving motion vector accuracy without complicating the overall process
Solution Approach 2:
The patent changes the parameter of pixel value availability by introducing padded pixel values at border regions. This parameter change allows the motion estimation algorithm to access valid pixel values even when motion vectors point beyond the original frame boundaries, thereby improving estimation accuracy while maintaining algorithmic simplicity
2Productivity
If motion estimation is performed without padding border regions, then the processing speed is fast, but the compensation accuracy at image borders deteriorates
Solution Approach 1:
The patent performs preliminary padding of border regions with replicated pixel values before motion compensation. This ensures that when motion vectors reference border or extrapolated regions, valid pixel values are already in place, enabling accurate compensation at image borders without adding complex processing steps during the main computation phase
Solution Approach 2:
The padded border pixel values act as an intermediary that bridges the gap between the original frame data and the motion-compensated reference regions. This intermediary layer allows the motion compensation algorithm to operate continuously without special case handling, maintaining processing speed while improving border region accuracy
3Adaptability or versatility
If padding region is created using opposite border region, then the spatial relation characteristic is utilized, but the reference frame size increases
Solution Approach 1:
The patent applies local quality by creating padding regions only at the border areas where spatial wraparound characteristics are relevant, rather than modifying the entire reference frame. This localized approach utilizes the panoramic spatial relation where left and right borders are connected, improving adaptability to panorama-specific motion patterns while minimizing the increase in overall reference frame size
Solution Approach 2:
The patent effectively utilizes the temporal dimension by repurposing historical pixel values from opposite borders to create padding regions. This dimensional transformation allows the system to handle panoramic spatial relations without physically expanding the spatial dimensions of the reference frame, as the padding values are derived from temporal relationships in the video sequence
Data Source
AI summary
Provided are a method and device for motion estimation and compensation to be performed on a panorama image. The motion estimation and compensation are performed on a panorama image with a 360° omni-directional view based on the spatial relation between left and right borders of the panorama image being very high. Accordingly, it is possible to improve image quality through effective and precise estimation and compensation for the motion of a panorama image. In particular, it is possible to improve the image quality at the right and left edges of the panorama image.