Image Denoising via Depth-Guided Edge Preservation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image denoising methods often blur object boundaries during noise reduction, especially in low light environments, and require multiple frames or long processing times, which affects continuous shooting and image clarity.
Innovation Solution
An image denoising method that performs edge detection on a color image and uses a depth image to guide a first noise reduction process, preserving edge information by identifying and reducing high-frequency noises in connected regions, thereby avoiding blurring and improving processing efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If multi-frame noise reduction technique is used to guarantee clarity of details or outlines, then image clarity is improved, but processing time is extended and continuous shooting is affected
Solution Approach 1:
The patent segments the image processing into distinct components: edge detection identifies boundary regions, depth information provides spatial context, and noise reduction is selectively applied to non-edge regions. This segmentation allows parallel processing and avoids the need for multiple frame captures, reducing processing time while maintaining clarity.
Solution Approach 2:
The patent introduces depth information as an intermediary element that mediates between the color image and the noise reduction process. The depth map provides spatial context that helps distinguish edge regions from non-edge regions, enabling selective noise reduction without requiring multiple frames, thus reducing processing time while preserving image clarity.
2Object-affected harmful factors
If conventional filters are used for partition denoising, then noise is reduced, but boundary blurring occurs
Solution Approach 1:
The patent applies local quality by using depth information to identify different regions (edge vs. non-edge) and applying appropriate processing to each. Edge regions are preserved with minimal filtering to maintain boundary clarity, while non-edge regions receive stronger noise reduction. This localized approach eliminates boundary blurring while effectively reducing noise.
Solution Approach 2:
The patent changes the parameter of noise reduction strength based on spatial location determined by depth information. By dynamically adjusting the filtering intensity according to whether a pixel lies on an edge or in a flat region, the method reduces noise in appropriate areas while preserving boundary sharpness, avoiding the boundary blurring that occurs with uniform filtering.
3Loss of information
If edge detection is performed on color image, then boundary information is extracted, but processing complexity increases
Solution Approach 1:
The patent merges edge detection with depth-based region classification into a unified processing framework. By combining the edge map with depth information, the system efficiently identifies regions requiring different processing without requiring separate complex analysis steps, reducing overall processing complexity while preserving edge information.
Data Source
AI summary
The present disclosure provides an image denoising method, an image denoising device, an image denoising apparatus and a storage medium. The method includes: performing edge detection on a color image to obtain a preprocessed image; acquiring a depth image having the same scene as the color image; and performing a first noise reduction process on the preprocessed image according to the depth image to obtain a first image.