Noise-Robust Image Registration via Radon and FFT Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image registration methods are inadequate for accurately realigning distorted images with reference images, especially in the presence of noise, and fail to provide robust visual quality assessment under affine transforms.
Innovation Solution
The method involves determining a scale factor and rotation angle between the reference and distorted images using Radon transforms and Fast Fourier Transforms, followed by image resizing, rotation, and translation to align the images, while also assessing visual quality using metrics like PSNR, SSIM, and VIF.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image registration methods are used, then the registration process is simple, but the accuracy deteriorates in the presence of noise and affine transforms
Solution Approach 1:
The patent applies Radon transform to convert spatial domain image data into frequency domain representation, enabling the detection of affine transformation parameters (rotation, scaling, shearing) by analyzing spectral characteristics. This dimensional transformation from spatial to frequency domain allows robust estimation of transformation parameters even in noisy conditions, resolving the contradiction between registration accuracy and noise sensitivity.
Solution Approach 2:
The method estimates affine transformation parameters by analyzing the spectral signature of the image and comparing it with the reference image. By changing the domain of analysis from spatial to frequency domain and using spectral characteristics, the method achieves accurate parameter estimation that is robust to noise, thereby improving registration accuracy without being adversely affected by noise.
2Measurement precision
If affine transform correction is applied, then visual quality assessment accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent performs preliminary estimation of affine transformation parameters using Radon transform and spectral analysis before conducting visual quality assessment. By pre-correcting the distorted image for rotation, scaling, and shearing transformations, the method enables subsequent quality metrics (PSNR, SSIM, VIF) to operate on properly aligned images, significantly improving assessment accuracy. The computational overhead is justified by the substantial improvement in quality measurement reliability.
Data Source
AI summary
An image registration method is disclosed for processing a distorted image into a registered image that is aligned with reference to an original image. Distortions from the original image may include scaling, rotation, and noise. The method is based on correlating Radon transforms of both images to determine the rotation angle, and the scaling factor is determined by dividing averages of the overall luminance of each image on the assumption that any added noise will cancel. The Fast Fourier Transform (FFT) is used to estimate global spatial shifts. In one embodiment, the distorted image is first scaled to the size of the original image before being rotated. In another embodiment, the original image is first scaled to the size of the distorted image before rotating the distorted image, and finally scaling it to match the original image. A corresponding system for image registration is also provided.