Abstract:A novel wavelet-based data clustering method is presented in this paper, which includes wavelet feature extraction and cluster growing algorithm. Wavelet transform can provide rich and diversified information for representing the global and local inherent structures of dataset, therefore, it is a very powerful tool for clustering feature extraction. As an unsupervised classification, the target of clustering analysis is dependent on the specific clustering criteria. Several criteria that should be considered for general-purpose clustering algorithm are proposed. And the cluster growing algorithm is also constructed to connect clustering criteria with wavelet features. Compared with other popular clustering methods, our clustering approach provides multi-resolution clustering results, needs few prior parameters, correctly deals with irregularly shaped clusters, and is insensitive to noises and outliers. As this wavelet-based clustering method is aimed at solving two-dimensional data clustering problem, for high-dimensional datasets, self-organizing map and U-matrix method are applied to transform them into two-dimensional Euclidean space, so that high-dimensional data clustering problem becomes two-dimensional problem. Therefore, our method is also suitable for high-dimensional data clustering analysis. Results on some simulated data and standard test data are reported to illustrate the power of our method.

A Cluster-Based Wavelet Feature Extraction Method and Its Application

Texture classification using wavelets with a cluster-Based feature extraction

Wavelets-Based Feature Extraction for Texture Classification

A New Cluster-Based Feature Extraction Method for Surface Defect Detection

Texture Clustering Combining Complex Wavelet Features and Local Binary Pattern

Spectral Clustering Algorithm of Image Texture Based on Rotated Complex Wavelet Transform

Data Clustering Analysis Based On Wavelet Feature Extraction

A Cluster-Based Wavelet Feature Extraction Method for Machine Fault Diagnosis

Image Texture Clustering Based on Locality Preserving Projection

Feature Extraction Through Contourlet Subband Clustering for Texture Classification.

A Novel Wavelet Transform Algorithm for Feature Extraction of Hyperspectral Remote Sensing Image

Featrure Extraction and Classification of an Underwater Target Based on CWT

Shape-based Shape Feature Extraction Using Wavelet Multi-Scale Analysis and Moment

Feature Extraction Method of Ultrasonic Signal Based on Wavelet Coefficients Clustering

Automation Textured and Non-Textured Images Classification and Retrieval

Texture Classification Using Wavelet Frame Representation Based Feature

Classification Based on Texture Feature of Wavelet Transform

Feature Extraction and Segmentation for Texture Based on Dual-tree Complex Wavelet Transform

Wavelet Transform-Based Texture Classification with Feature Weighting.

Image Segmentation Based on Fussing Multi-feature and Spatial Spectral Clustering

A Wavelet Model of Ganglion Cells Array and Its Application in Image Representation.