Feature-based disconnected edge segments linker

ASIAIR > College of Computer Science > Department of Applied Informatics and Multimedia > Journal Article > Item 310904400/18886

Please use this identifier to cite or link to this item: http://asiair.asia.edu.tw/ir/handle/310904400/18886

Title:	Feature-based disconnected edge segments linker
Authors:	詹啟祥;Chan, Chi-Shiang
Contributors:	資訊多媒體應用學系
Date:	2011
Issue Date:	2012-11-26 07:10:12 (UTC+0)
Abstract:	Object contour detection plays an important role in image processing. Many edge and object contour detectors have been proposed, such as, Canny edge detector, ACM, and level set methods. However, those detection methods often cannot provide the fully edge and object contour perfectly. In the ACM and level set methods, the initial contour of object has to be specified by a user in advance, and the successful segmentation results highly depend on the given initial object contour. In addition, the ACM and level set methods are significantly sensitive to noise. Giving an initial contour of object is not required for Canny edge detector, but it may bring about "broken edge problem." To solve the broken edge problem, a feature-based disconnected edge segments (FBDES) linker is provided in this paper. For extracting objects from an image, in this paper Canny edge detector is first used to detect the object contour, and then the FBDES linker is employed to link the disconnected edge segments based on the features of the gray-level difference at vicinity of, the gradient and gradient direction on, and the length of the line segment connecting two disconnected edge segments. The experimental results show that the FBDES linker can give more impressive object segmentation results than the level set method.
Relation:	2011 IEEE 3rd International Conference on Communication Software and Networks
Appears in Collections:	[Department of Applied Informatics and Multimedia] Journal Article

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	379	View/Open

Loading...