Region-Adaptive Distance Filtering for Detailed 3D Image Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies struggle to balance noise reduction in distance data while accurately reproducing fine details such as organs, accessories, and hair asperities, leading to issues like noise retention on low-texture skin when using smoothing filters.
Innovation Solution
An image processing apparatus that divides images into regions, applies filter processing with varying characteristics based on classification, synthesizes the processed distance information, and generates 3D data, optimizing smoothing for different image areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the tap number of the smoothing filter is increased to remove noise from distance data, then noise reduction is improved, but fine details such as organs, accessories, and hair asperities cannot be reproduced
Solution Approach 1:
The patent applies different filter tap numbers to different regions of the image based on object classification. Specifically, regions containing fine details (organs, accessories, hair) use smaller tap numbers to preserve detail, while regions with less critical features use larger tap numbers for stronger noise reduction. This regional differentiation resolves the contradiction by optimizing noise reduction and detail preservation locally rather than applying a uniform filter across the entire image.
2Manufacturing precision
If the tap number of the filter is decreased to preserve fine details, then fine detail reproduction is improved, but noise tends to remain on skin having low texture
Solution Approach 1:
The patent performs object classification to identify different regions (skin, hair, organs, accessories) and applies appropriately sized filters to each. Low-texture skin regions receive filters with larger tap numbers for noise reduction, while fine-detail regions receive filters with smaller tap numbers. This selective application resolves the contradiction by matching filter strength to regional requirements.
3Ease of operation
If a single filter tap number is used for the entire image, then processing simplicity is maintained, but both noise reduction and fine detail reproduction cannot be optimized simultaneously
Solution Approach 1:
The patent segments the image into multiple regions based on object classification (skin, hair, organs, accessories) and applies different filter parameters to each segment. This segmentation enables optimized noise reduction and detail preservation for each region type while maintaining a systematic processing approach that balances complexity with performance.
Data Source
AI summary
An image processing apparatus comprises a region division unit configured to divide an image into a plurality of regions by classifying objects; a filter processing unit configured to perform, with respect to respective distance information corresponding to the plurality of regions of the image, filter processing having different characteristics according to the classification; a synthesis unit configured to synthesize the distance information on which the filter processing has been performed; and a 3D data generation processing unit configured to generate 3D data of a subject based on the distance information synthesized by the synthesis unit and the image.