Robust Multi-Frame Image Registration for Low-Light Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-frame image processing techniques struggle with frame registration failures due to noise, especially in low-light conditions and scenarios with limited scene content, leading to blurry or smeared images.
Innovation Solution
A method involving tile-based and feature-based registration of non-reference frames to a reference frame, followed by aggregation of motion vectors and warping to generate aligned frames, with segmentation to optimize registration in different scene portions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional single-method registration is used, then processing speed is maintained, but registration accuracy deteriorates in low-light conditions and scenes with limited content
Solution Approach 1:
The patent segments the registration process into two distinct methods: tile-based registration for general areas and feature-based registration for key content areas. This segmentation allows each method to be optimized for its specific use case, improving overall registration accuracy across different scene conditions.
Solution Approach 2:
The patent dynamically selects between tile-based and feature-based registration methods based on scene content analysis. The system adapts its registration approach in real-time, switching methods depending on whether the scene contains distinguishable features or is better suited for tile-based alignment, thereby improving adaptability to different conditions.
2Reliability
If dual registration methods (tile-based and feature-based) are combined, then registration robustness is improved, but processing complexity increases
Solution Approach 1:
The patent performs preliminary scene analysis to identify suitable feature points before executing the dual registration process. This preliminary action allows the system to prepare and pre-process potential feature matches, reducing the complexity of the subsequent registration operations by having candidate features ready in advance.
Solution Approach 2:
The patent introduces motion vectors as an intermediary element that bridges tile-based and feature-based registration results. Motion vectors serve as a common representation that allows the system to integrate results from both methods without directly combining their complex processing pipelines, thereby managing overall system complexity.
3Measurement precision
If motion vectors are aggregated from both registration methods, then alignment accuracy is improved, but computational load increases
Solution Approach 1:
The patent applies partial action by selectively processing only the most relevant tiles and features for aggregation. Instead of combining all possible motion vectors from both registration methods, the system identifies and processes only those that contribute most significantly to alignment accuracy, reducing unnecessary computational overhead while maintaining precision.
4Manufacturing precision
If warping is applied based on aggregated motion vectors, then frame alignment quality is improved, but processing time increases
Solution Approach 1:
The patent segments the warping operation into multiple passes or stages, applying transformations to different regions of the image sequentially. This segmentation allows the system to optimize warping computations for different areas, improving overall alignment quality while managing processing time through efficient region-based processing.
Data Source
AI summary
A method includes obtaining, using at least one processing device of an electronic device, multiple image frames capturing a scene. The method also includes selecting, using the at least one processing device, a reference frame among the image frames. The method further includes aligning, using the at least one processing device, each of one or more non-reference frames among the image frames with the reference frame by (i) performing tile-based registration of the non-reference frame to the reference frame, (ii) performing feature-based registration of the non-reference frame to the reference frame, (iii) aggregating first motion vectors generated during the tile-based registration and second motion vectors generated during the feature-based registration, and (iv) warping the non-reference frame based on the aggregated motion vectors to generate an aligned non-reference frame. The reference frame and the one or more aligned non-reference frames may be blended to generate a final image of the scene.