Moiré Removal via Laplacian Pyramid Feature Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies face challenges in effectively removing moiré patterns from images captured from electronic display screens, leading to poor viewing experiences and increased data volume due to added texture and noise, with existing edge extraction algorithms being complex and inefficient.
Innovation Solution
An image processing method utilizing a network model with a multi-band module that processes images through an N-layer Laplacian pyramid to generate feature maps across different spatial frequency bands, enabling comprehensive moiré pattern removal by training on a moiré pattern training dataset and employing techniques like multi-scale learning and attention mechanisms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If edge extraction algorithm is used to remove moiré patterns, then moiré pattern removal is achieved, but the algorithm complexity increases and removal effect is poor
Solution Approach 1:
The patent segments the image processing task by decomposing the image into multiple frequency bands using Laplacian pyramid transformation. This allows different processing strategies to be applied to different frequency components, specifically targeting moiré patterns in certain bands while preserving other image content, thereby improving removal effectiveness without requiring complex global processing algorithms.
Solution Approach 2:
The patent replaces traditional mechanical edge extraction algorithms with a deep learning-based neural network model. This substitution enables the system to learn complex moiré pattern characteristics automatically from training data, achieving better removal effects while simplifying the algorithmic approach compared to manual edge detection and extraction methods.
2Reliability
If traditional moiré pattern removal methods are used, then processing is simpler, but the removal effect is poor and texture details are lost
Solution Approach 1:
The patent applies local quality enhancement by processing different frequency bands with different operations. The Laplacian pyramid decomposition allows the system to apply targeted filtering to specific frequency regions where moiré patterns appear, while preserving high-frequency details and texture information in other bands, thus removing moiré patterns without losing important image content.
Solution Approach 2:
The patent transforms the problem from a single-dimensional spatial domain approach to a multi-dimensional frequency domain approach using Laplacian pyramid decomposition. This dimensional transformation enables the system to analyze and process image content at different frequency scales, allowing selective removal of moiré patterns while preserving texture details that would be lost in traditional single-band processing.
3Quantity of substance
If moiré patterns are present in the image, then the original image is captured, but viewing experience deteriorates and data volume increases
Solution Approach 1:
The patent extracts and removes the harmful moiré pattern components from the image through frequency band analysis. By identifying and eliminating the specific frequency ranges where moiré patterns appear, the system reduces unnecessary data while improving viewing experience, as the processed image contains only the essential visual information without the distracting moiré artifacts.
Data Source
AI summary
Methods, apparatuses and computer-readable storage mediums for processing images to remove moiré patterns. The method entails running an image processing model to perform moiré pattern removal processing on an original image to obtain a target image. The image processing model may include a network model trained according to a moiré pattern training data set. The image processing model can also include a multi-band module, which may be configured to process the original image to obtain an N-layer Laplacian pyramid of the original image, and obtain a first processing result feature map based on feature maps corresponding to N spatial frequency bands of the N-layer Laplacian pyramid, with N being a positive integer greater than or equal to 2. The image processing model can include obtaining the target image according to the first processing result feature map.