License Plate Image Reconstruction Using Optical Flow Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing license plate recognition models perform poorly with low-quality video inputs due to degradation and vehicle movement, leading to reduced visibility and recognition performance.
Innovation Solution
An image generation method involving generating crop images, determining a reference crop image, calculating optical flow values, aligning images based on these values, and enhancing the license plate image using algorithms like CLAHE to improve resolution and recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If low-resolution vehicle license plate images are converted into high-resolution images, then image resolution is improved, but the license plate cannot be readily restored due to degradation and continuous movement
Solution Approach 1:
The patent divides the video into multiple frames and extracts multiple crop images containing the license plate area from different time points. Each crop image is processed independently through alignment and super-resolution, allowing the system to handle the degradation and movement issues by segmenting the reconstruction task across multiple temporal snapshots rather than attempting to restore a single degraded frame.
Solution Approach 2:
The patent performs preliminary actions by extracting crop images containing the license plate area from multiple video frames before the final super-resolution reconstruction. This preliminary extraction and alignment of multiple frames enables the system to prepare and align multiple high-quality source images that can then be fused to produce the final high-resolution license plate image, overcoming the limitations of single-frame degradation.
2Measurement precision
If multiple crop images are generated and aligned, then recognition performance is improved, but processing time and computational complexity increase
Solution Approach 1:
The patent segments the video stream into discrete frames and extracts only the relevant license plate area from each frame using detection algorithms. This segmentation allows the system to process only the necessary portions of the video data rather than the entire video stream, reducing computational load while maintaining recognition accuracy through the alignment and super-resolution of these segmented crop images.
Solution Approach 2:
The patent creates multiple copies of the license plate area from different time points in the video and then aligns and fuses these copies through super-resolution. By working with multiple copies extracted from different frames, the system can achieve higher recognition performance without needing to process the entire original video in detail, thus managing computational complexity effectively.
Data Source
AI summary
A method of generating a license plate image of a vehicle performed by an electronic device includes generating a plurality of crop images including a license plate area of a target vehicle based on a plurality of basic images of a video capturing the target vehicle, determining a reference crop image among the plurality of crop images, calculating a plurality of optical flow values between the reference crop image and each of the plurality of crop images, determining a plurality of alignment images among the plurality of crop images based on the plurality of optical flow values, and generating the license plate image based on the plurality of alignment images.