Super-Resolution Image Reconstruction via Temporal Frame Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Surveillance camera systems, such as those detecting illegally parked vehicles, face challenges in recognizing faces, road signs, and license plate numbers due to low-resolution images caused by camera lens blur, noise, or image compression, leading to loss of important information.
Innovation Solution
A method and system for generating super-resolution images by fusing pixel data of high-resolution images estimated at previous times with low-resolution images input at current times, using techniques like Kalman filter upsampling and image registration error calculation to improve resolution and reduce misalignment errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If image compression is applied to reduce storage requirements, then storage capacity is improved, but image resolution deteriorates
Solution Approach 1:
The system performs preliminary high-resolution image capture and processing before compression is applied. By generating super-resolution images through fusion of multiple low-resolution frames, the system recovers lost high-frequency information before final compression, thereby maintaining image quality while achieving storage efficiency.
Solution Approach 2:
The patent introduces an intermediate super-resolution image generation step between low-resolution capture and final storage. This intermediary process uses frame fusion and filtering to reconstruct high-frequency details that would otherwise be lost in compression, serving as a mediator that preserves image quality while enabling compression.
2Manufacturing precision
If multiple low-resolution images are captured to improve resolution, then image resolution is improved, but processing time increases
Solution Approach 1:
The processing is segmented into distinct stages: motion compensation, frame alignment, fusion, and filtering. By dividing the complex super-resolution task into manageable segments, the system can process multiple frames efficiently without excessive computational overhead, reducing overall processing time while maintaining high resolution output.
Solution Approach 2:
The system uses periodic frame sampling and temporal filtering to process multiple images. By selectively combining frames at regular intervals and applying temporal constraints, the system achieves high-resolution reconstruction from multiple inputs without requiring continuous processing of every captured frame, thereby reducing processing time.
3Manufacturing precision
If image fusion is performed to generate super-resolution images, then image resolution is improved, but memory consumption increases
Solution Approach 1:
The system extracts only the essential high-frequency information from multiple low-resolution frames rather than storing or processing complete high-resolution versions of all frames. By separating and processing only the critical detail information through fusion algorithms, memory consumption is reduced while still achieving super-resolution output.
Solution Approach 2:
The patent applies local adaptive filtering and fusion techniques that process different regions of the image with appropriate methods. By focusing computational resources on regions containing important details and using efficient local operations, the system achieves high resolution where needed while minimizing overall memory requirements compared to global processing approaches.
Data Source
AI summary
A method and system of generating a super-resolution image is provided. fusing The method includes inputting an image; and generating an estimated high-resolution image of a current time by fusing an input image and an estimated high-resolution image of a previous time corresponding to the input image.