Spectral edit distance method for image clustering

Authors:
Nian Wang;Jun Tang;Jiang Zhang;Yi-Zheng Fan;Dong Liang
Affiliations:
Key Laboratory of Intelligent Computing & Signal Processing, Anhui University, Education Ministry, Hefei, P.R. China;Key Laboratory of Intelligent Computing & Signal Processing, Anhui University, Education Ministry, Hefei, P.R. China;Key Laboratory of Intelligent Computing & Signal Processing, Anhui University, Education Ministry, Hefei, P.R. China;Key Laboratory of Intelligent Computing & Signal Processing, Anhui University, Education Ministry, Hefei, P.R. China and Department of Mathematics, Anhui University, Hefei , P.R. China;Key Laboratory of Intelligent Computing & Signal Processing, Anhui University, Education Ministry, Hefei, P.R. China
Venue:
APWeb/WAIM'07 Proceedings of the joint 9th Asia-Pacific web and 8th international conference on web-age information management conference on Advances in data and web management
Year:
2007

Citing 8
Cited 0

Combination of neural and statistical algorithms for supervised classification of remote-sensing images

Pattern Recognition Letters
Shape Matching and Object Recognition Using Shape Contexts

IEEE Transactions on Pattern Analysis and Machine Intelligence
Introducing a weighted non-negative matrix factorization for image classification

Pattern Recognition Letters
Content-based image classification using a neural network

Pattern Recognition Letters
A spectral approach to learning structural variations in graphs

Pattern Recognition
Graph matching and clustering using spectral partitions

Pattern Recognition
Graph clustering using the weighted minimum common supergraph

GbRPR'03 Proceedings of the 4th IAPR international conference on Graph based representations in pattern recognition
ACM attributed graph clustering for learning classes of images

GbRPR'03 Proceedings of the 4th IAPR international conference on Graph based representations in pattern recognition

Quantified Score

Hi-index	0.00

Visualization

Abstract

The spectral graph theories have been widely used in the domain of image clustering where editing distances between graphs are critical. This paper presents a method for spectral edit distance between the graphs constructed on the images. Using the feature points of each image, we define a weighted adjacency matrix of the relational graph and obtain a covariance matrix based on the spectra of all the graphs. Then we project the vectorized spectrum of each graph to the eigenspace of the covariance matrix, and derive the distances between pairwise graphs. We also conduct some theoretical analyses to support our method. Experiments on both synthetic data and real-world images demonstrate the effectiveness of our approach.