Hill-Down strategy based DENsity CLUstEring and its application to medical image data

Conghua XIE; Jinyi CHANG; Yuqing SONG

2nd International ICST Conference on Scalable Information Systems

Research Article

Hill-Down strategy based DENsity CLUstEring and its application to medical image data

Download1029 downloads

Cite: BibTeX Plain Text

@INPROCEEDINGS{10.4108/infoscale.2007.968,
    author={Conghua XIE and Jinyi CHANG and Yuqing SONG},
    title={Hill-Down strategy based DENsity CLUstEring and its application to medical image data},
    proceedings={2nd International ICST Conference on Scalable Information Systems},
    proceedings_a={INFOSCALE},
    year={2010},
    month={5},
    keywords={Hill-down strategy; density clustering hill-climbing},
    doi={10.4108/infoscale.2007.968}
}

Conghua XIE
Jinyi CHANG
Yuqing SONG
Year: 2010
Hill-Down strategy based DENsity CLUstEring and its application to medical image data
INFOSCALE
ICST
DOI: 10.4108/infoscale.2007.968

Conghua XIE¹^,*, Jinyi CHANG¹^,*, Yuqing SONG²^,*

1: Depart. of Computer Science and Engineering Changshu Institute of Technology, China 215500
2: School of Computer Science and Engineering, Jiangsu University, China 212013

*Contact email: x7c8h5@yahoo.com, cjy@cslg.edu.cn, yqsong@ujs.edu.cn

Abstract

In order to overcome problems of DENCLUE and improve it for medical image segmentation, we propose HD-DENCLUE algorithm. It takes the optimization step length to find density attractor, designs a hill-down strategy to give different density thresholds for different clusters and stops at hill-down data i.e. the edge data of this cluster. Experiments show that HD- DENCLUE can decrease time overhead, have better clustering effects than DENCLUE and get the edges of each cluster.

Keywords: Hill-down strategy; density clustering hill-climbing

Published: 2010-05-16
Modified: 2011-09-11

: http://dx.doi.org/10.4108/infoscale.2007.968

Hill-Down strategy based DENsity CLUstEring and its application to medical image data

Abstract

About EAI

Community

Publish with EAI