Arna Fariza, - and Agus Zainal Arifin, - and Eha Renwi Astuti, - (2018) Interactive Segmentation of Conditional Spatial FCM with Gaussian Kernel-Based for Panoramic Radiography. IEEE. pp. 157-161. ISSN 00189219, 15582256,

	Text (VALIDASI) 44. Interactive Segmentation of Conditional Spatial FCM with Gaussian Kernel-Based for Panoramik.pdf Download (207kB)
	Text (KARIL) 38. Interactive Segmentation of Conditional Spatial FCM.pdf Download (112kB)
	Text (TURNITIN) 18 % Interactive Segmentation of Conditional Spatial FCM with Gaussian Kernel-Based for Panoramic Radiography.pdf Download (1MB)
	Text (ARTIKEL) Interactive Segmentation of Conditional Spatial FCM.pdf Download (303kB)

Abstract

Dental image segmentation is widely used for various real applications such as dental diagnosis, teeth numbering, dental age estimation, dental plaque analysis and etc. Dental image segmentation is a challenging task in panoramic radiography because the difficulty due to noise, low contrast, uneven illumination, complicate topology of objects and unclear lines of demarcation of the panoramic radiography. Unsupervised segmentation of Conditional Spatial FCM with Gaussian Kernel-Based incorporate spatial information and gaussian kernel function to overcome inhomogeneous regions. However, it encountered significant obstacles in obtaining effective segmentation to differentiate teeth with other dental features. To alleviate the problem, an interactive segmentation method involves the user to engage in the segmentation process by incorporating prior-knowledge, thus lead to accurate segmentation results. This paper proposes a novel strategy of conditional spatial FCM with Gaussian Kernel-Based in interactive segmentation for panoramic radiography image. The representative sample area chosen by the user causes the initialization value will affect the membership function in the segmentation process, thus it will overcome the lack of algorithm in distinguishing the tooth and background areas. This strategy gives a higher segmentation accuracy than automatic segmentation method with a few user samples.

Actions (login required)

View Item