Individual tooth image segmentation with correcting of specular reflections

Seong Taek Lee, Kyeong Seop Kim, Tae Ho Yoon, Jeong Whan Lee, Kee Deog Kim, Wonse Park

Research output: Contribution to journalArticlepeer-review

Abstract

In this study, an emcient removal algonthm tor specular reflections in a tooth color image is proposed to minimize the artefact interrupting color image segmentation. The pixel values of RGB color channels are initially reversed to emphasize the features in reflective regions, and then those regions are automatically detected by utilizing perception artificial neural network model and those prominent intensities are corrected by applying a smoothing spatial filter. After correcting specular reflection regions, multiple seeds in the tooth candidates are selected to find the regional minima and MCWA(Marker-Controlled Watershed Algorithm) is applied to delineate the individual tooth region in a CCD tooth color image. Therefore, the accuracy in segmentation for separating tooth regions can be drastically improved with removing specular reflections due to the illumination effect.

Original languageEnglish
Pages (from-to)1136-1142
Number of pages7
JournalTransactions of the Korean Institute of Electrical Engineers
Volume59
Issue number6
Publication statusPublished - 2010 Jun

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Individual tooth image segmentation with correcting of specular reflections'. Together they form a unique fingerprint.

Cite this