Related Experiment Video
Updated: Jul 16, 2026
09:10
Digital Hybrid Model Preparation for Virtual Planning of Reconstructive Dentoalveolar Surgical Procedures
Published on: August 5, 2021
Automated Adaptive Approach for Specular Highlight Removal in Digital Dentistry: A Proof-of-Concept Study for
Ji Su Han1, Sung-Ae Son2, Il-Ho Park3
1Department of AI Convergence, Sahmyook University, Seoul 01795, Republic of Korea.
Journal of Clinical Medicine
|July 15, 2026
Summary
This study developed an automated method to remove specular highlights from dental images, preserving tooth texture for better AI analysis and clinical documentation.
Area of Science:
- Dentistry
- Medical Imaging
- Computer Vision
Background:
- Digital intraoral photography is crucial for dental documentation and AI analysis.
- Specular highlights from saliva and lighting obscure tooth texture, reducing image quality.
- Existing methods struggle to balance highlight removal with texture preservation.
Purpose of the Study:
- To develop an automated pipeline for removing specular highlights from intraoral images.
- To preserve tooth surface texture during highlight removal.
- To improve the fidelity of dental images for clinical and AI applications.
Main Methods:
- A three-stage pipeline: adaptive threshold prediction, mask generation, and image inpainting.
- Utilized HSV statistical features for threshold prediction and CIE LAB color space for refinement.
- Employed an Aggregated Contextual Transformation (AOT)-based generator for image restoration.
Main Results:
- The dynamic adjustment refined masking, reducing over-masking and preserving tooth texture.
- Significantly reduced distributional discrepancies (Wasserstein distance from 2.96 to 1.35, KL divergence from 0.35 to 0.06).
- Clinical experts preferred the proposed method in 69.5% of evaluations (p < 0.001).
Conclusions:
- The pipeline effectively reduces specular highlights while preserving tooth surface texture.
- This method shows potential for standardized preprocessing of dental image datasets.
- Further validation on real clinical photographs is needed for broader applications.
