Spatial Filter Convolution for Image Resolution Anisotropy Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies fail to effectively improve resolution anisotropy in digital camera images, leading to image blurring and directional moiré patterns due to mechanical vignetting and limited finite spatial filters.
Innovation Solution
An image processing apparatus that acquires a first finite spatial filter with image resolution anisotropy and computes a second spatial filter by convolving a finite high-pass filter with the first spatial filter, ensuring the sum of elements is 0 and at least two elements are non-zero, to correct image blurring and improve resolution anisotropy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a finite spatial filter with anisotropy is used to correct image blurring, then resolution anisotropy is improved, but directional moiré patterns are induced due to limited filter elements
Solution Approach 1:
The patent changes the parameters of the spatial filter by convolving it with a high-pass filter that has specific properties (sum of elements equals 0, at least two non-zero elements). This parameter modification allows the filter to maintain resolution improvement while suppressing the directional moiré pattern generation.
Solution Approach 2:
The patent creates a composite filtering approach by combining the original finite spatial filter with a high-pass filter through convolution. This composite filter structure integrates the edge-enhancement capabilities of the high-pass filter with the deblurring functionality of the original spatial filter, achieving both resolution improvement and moiré suppression.
2Reliability
If different filtering is performed according to image positions to correct blurring, then image quality is improved, but resolution anisotropy cannot be effectively improved
Solution Approach 1:
The patent applies local quality by using different spatial filters for different positions in the image. Each filter is tailored to the specific characteristics of that region, taking into account the radial distance from the image center and the directional characteristics of blurring at that location. This allows effective correction of position-dependent blurring while addressing resolution anisotropy.
Solution Approach 2:
The patent employs asymmetric filtering by using filters with different characteristics for different directions and positions. The spatial filters are designed with anisotropic properties that match the asymmetric nature of optical blurring, which varies with radial distance and angle from the image center. This asymmetric approach enables effective correction of resolution anisotropy.
3Reliability
If a large number of operations are performed in the frequency domain for filtering, then filtering function is achieved, but hardware implementation becomes complex
Solution Approach 1:
The patent replaces frequency domain operations with spatial domain convolution operations. Instead of performing computationally intensive Fourier transforms and frequency domain multiplications, the system uses direct spatial convolution with pre-computed spatial filters. This substitution significantly reduces computational complexity and enables efficient hardware implementation while maintaining the same filtering functionality.
Data Source
AI summary
An image processing apparatus includes an acquisition unit configured to acquire a first finite spatial filter having image resolution anisotropy, and a calculation unit configured to compute a second spatial filter by convolving a finite filter with respect to the first spatial filter, the finite filter having a sum of elements being 0 and at least two of the elements being non-0.