Raw Image Restoration Using Channel-Guided Demosaicing Denoising

Solution Overview

Problem

Existing joint demosaicing and denoising methods in computer vision fail to fully utilize color filter array data, leading to false artifacts and image blurring, which degrades the quality of processed images.

Innovation Solution

An image processing method that includes preliminary image restoration on channel images, using density maps and neural networks to distinguish texture regions, followed by adaptive convolution processing to enhance denoising and retain image details.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If joint demosaicing and denoising is performed using conventional methods, then processing speed is maintained, but false artifacts appear and image quality deteriorates

Engineering Contradiction:

Improveimage qualityVSAvoidfalse artifacts

Core Design Contradiction:

Manufacturing precisionVSObject-generated harmful factors

Solution Approach 1:

The patent performs preliminary image restoration processing on channel images before using them to guide raw image restoration. This preliminary action separates the restoration of channel images from raw images, allowing the channel images to be optimized first and then used as guidance to reduce false artifacts in the final demosaiced and denoised image.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent segments the image processing into distinct channel images (e.g., R, G, B channels) and processes each channel separately through preliminary restoration before combining them. This segmentation allows targeted processing of each color channel, improving overall image quality while reducing artifacts through the guided restoration process.

Inventive Principle:

Principle #1Segmentation

2Manufacturing precision

If joint demosaicing and denoising is performed using conventional methods, then computational complexity is reduced, but image details are lost and blurring occurs

Engineering Contradiction:

Improveimage detailsVSAvoidprocessing complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent performs preliminary image restoration processing on channel images before using them to guide raw image restoration. This preliminary action separates the restoration of channel images from raw images, allowing the channel images to be optimized first and then used as guidance to reduce false artifacts in the final demosaiced and denoised image.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent applies different processing parameters and strategies to different channel images during preliminary restoration, and adjusts the guidance strength based on channel-specific characteristics. This parameter differentiation allows detailed preservation in each channel while managing overall processing complexity through structured guidance mechanisms.

Inventive Principle:

Principle #35Parameter changes

3Manufacturing precision

If color filter array data is not fully utilized in joint demosaicing and denoising, then processing simplicity is maintained, but image quality deteriorates with false artifacts

Engineering Contradiction:

Improveimage qualityVSAvoidcolor filter array data utilization

Core Design Contradiction:

Manufacturing precisionVSLoss of information

Solution Approach 1:

The patent performs preliminary image restoration processing on channel images before using them to guide raw image restoration. This preliminary action separates the restoration of channel images from raw images, allowing the channel images to be optimized first and then used as guidance to reduce false artifacts in the final demosaiced and denoised image.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent uses restored channel images as guidance feedback for the raw image restoration process. The preliminarily restored channel images provide feedback information that guides the demosaicing and denoising of the raw image, ensuring that color filter array data is fully utilized while maintaining processing efficiency and reducing false artifacts.

Inventive Principle:

Principle #23Feedback

Data Source

PatentEP4109392B1Image processing method and image processing device

Publication Date: 2026.02.25 HUAWEI TECH CO LTD

Figure 1~2
Figure 3~4
Figure 5

AI summary

This application discloses an image processing method and apparatus in the computer vision field of the artificial intelligence field. The image processing method includes: obtaining a channel image corresponding to a to-be-processed raw (raw) image, where the channel image includes at least two first channel images; performing image restoration processing on the first channel image to obtain the restored first channel image; and performing image restoration processing on the to-be-processed raw image based on the restored first channel image to obtain a raw image obtained through joint demosaicing and denoising processing. The technical solution of this application can improve image quality of the raw image obtained through joint demosaicing and denoising processing.