By Topic

Parallel Sparse Spectral Clustering for SAR Image Segmentation

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$33 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

4 Author(s)
Shuiping Gou ; Key Laboratory of Intelligent Perception and Image Understanding for the Ministry of Education, Xidian university, Xi'an, China ; Xiong Zhuang ; Huming Zhu ; Tiantian Yu

A novel parallel spectral clustering approach is proposed by exploiting the distributed computing in MATLAB for SAR image segmentation quickly and accurately. For large-scale data applications, most existing spectral clustering algorithms suffer from the bottleneck problems of high computational complexity and large memory use. And in the absence of advanced hardware and software equipments with only the loosely coupled computer resources accessible, the framework of MATLAB Parallel Computing-based sparse spectral clustering is constructed in this paper. In the proposed frame, we use a distributed parallel computing model to accelerate computation, where each partition of data instances is assigned to different processor nodes for the similarity matrix calculation in spectral clustering. Further, by the construction of exact t-nearest neighbor sparse symmetric similarity matrix, the sparseness technique is employed to alleviate the storage stress. Besides, the problems of how to choose the number of nearest neighbors and the scaling parameter are also discussed. The segmentation results on artificial synthesis texture images and SAR images show that the proposed parallel algorithm can effectively handle large-size segmentation cases. Meanwhile, it can obtain better segmentation results compared with Nyström approximation spectral clustering and k-means clustering algorithm.

Published in:

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing  (Volume:6 ,  Issue: 4 )