Adaptive brightness equalization and specular reflection suppression of narrow-band endoscopic images : NBI image enhancement algorithm
- 11 August 2023
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 313-318
Abstract
In clinical endoscopic diagnosis, the endoscopic image quality is reduced due to interference. The interference factors mainly include local overexposure and shadow of images caused by light source irradiation angle and specular reflection caused by tissue fluid. As the primary task in developing a computer-aided diagnosis (CAD) model, high-performance image enhancement algorithms can help them obtain a more accurate recognition rate. This study takes the endoscopic images of Barrett’s esophagus (BE) collected under narrow-band imaging as the research object. An adaptive brightness equalization algorithm and a specular reflection suppression algorithm are proposed to preprocess datasets, and the existing segmentation model confirms the algorithm. The experimental results show that the image enhanced by the algorithm has a certain degree of improvement in the overall and local dark areas and texture details, eliminates the specular reflection, and effectively improves the recognition accuracy of the segmentation model. The proposed algorithm improves the Mean Intersection over Union (MIoU) by 10.8% and the F 1 -Score by 8.76% on the self-built BE datasets. It shows that it can play an excellent auxiliary diagnostic role and obtain higher diagnostic efficiency in clinical diagnosis.Keywords
Funding Information
- Southwest University
This publication has 21 references indexed in Scilit:
- Adaptive Specular Reflection Detection and Inpainting in Colonoscopy Video Frames2018
- Adaptive segmentation and mask-specific Sobolev inpainting of specular highlights for endoscopic images2016
- 3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation2016
- A Specular Reflection Suppression Method for Endoscopic Images2016
- U-Net: Convolutional Networks for Biomedical Image Segmentation2015
- Adaptive local tone mapping based on retinex for high dynamic range images2013
- Contrast image correction methodJournal of Electronic Imaging, 2010
- Automatic Segmentation and Inpainting of Specular Highlights for Endoscopic ImagingEURASIP Journal on Image and Video Processing, 2010
- Appearance of enhanced tissue features in narrow-band endoscopic imagingJournal of Biomedical Optics, 2004
- Adaptive Logarithmic Mapping For Displaying High Contrast ScenesComputer Graphics Forum, 2003