Dual Demosaicing Image Processing for Checkered Pattern Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing demosaicing processes using neural networks, while effective in suppressing false color and moiré, introduce new image defects such as false patterns, particularly checkered patterns, due to insufficient training data or training biases.
Innovation Solution
A dual demosaicing process is employed, using a first demosaicing process with a neural network followed by a second demosaicing process that addresses the false patterns, combined to generate a composite image that suppresses these defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If a neural network-based demosaicing process is used, then false color and moiré are suppressed, but false patterns such as checkered patterns are introduced
Solution Approach 1:
The demosaicing process is segmented into multiple stages: first demosaicing using a neural network to suppress false color and moiré, detection of checkered pattern regions, and second demosaicing specifically for those regions to eliminate false patterns. This segmentation allows targeted handling of different defect types.
Solution Approach 2:
Different demosaicing methods are applied to different regions of the image based on local characteristics. Neural network-based demosaicing is applied to general regions, while conventional demosaicing is applied specifically to detected checkered pattern regions, optimizing quality for each local area.
2Productivity
If a single demosaicing process is used, then processing is simple and fast, but image quality suffers from various artifacts
Solution Approach 1:
The system dynamically selects and combines different demosaicing methods based on detected image characteristics. A detection process identifies regions with checkered patterns, and the system adaptively applies appropriate demosaicing algorithms to different regions, balancing speed and quality.
Solution Approach 2:
Multiple demosaicing results are merged into a final composite image. The system combines the output from neural network-based demosaicing for general regions with output from conventional demosaicing for checkered pattern regions, achieving both speed and quality.
3Ease of manufacture
If neural network training data is insufficient or has biases, then training is faster and simpler, but checkered patterns and other artifacts occur
Solution Approach 1:
The system performs preliminary detection of checkered pattern regions before final demosaicing completion. By detecting these problematic regions in advance, the system can apply corrective demosaicing specifically to those areas, preventing artifact propagation to the final image.
Data Source
AI summary
An image processing apparatus is disclosed. The image processing apparatus acquires a mosaic image. The image processing apparatus generates a first demosaic image by subjecting the mosaic image to a first demosaicing process in which a neural network is used, and generates a second demosaic image by subjecting the mosaic image to a second demosaicing process that is different from the first demosaicing process. The image processing apparatus generates a composite image in which the first demosaic image and the second demosaic image are combined.