Unsupervised Optimal Fuzzy Clustering
IEEE Transactions on Pattern Analysis and Machine Intelligence
Normalized Cuts and Image Segmentation
IEEE Transactions on Pattern Analysis and Machine Intelligence
Computer Vision
An Efficient k-Means Clustering Algorithm: Analysis and Implementation
IEEE Transactions on Pattern Analysis and Machine Intelligence
Efficient Graph-Based Image Segmentation
International Journal of Computer Vision
WSEAS Transactions on Signal Processing
A study on multiple objects detection, loading and control in video for augmented reality
WSEAS Transactions on Computers
Hi-index | 0.00 |
With more applications of multispectral remote sensing images, how to effectively and correctly make automated classification of multispectral images is still a great challenge. Utilizing both spatial contextual information and spectral information can achieve better classification performance. In order to make better utilization of the spatial contextual information, we apply graph model to the multispectral image, and use graph-based segmentation to produce units of pixels for further classification. In this paper, we present an unsupervised approach for multispectral remote sensing image classification with graph-based segment and fuzzy c-means clustering. Our method mainly involves following steps: First, represent image as graph H = (V,E) based on the feature vector of per pixel and the relationships among neighboring pixels, and segment the graph into groups of sub-regions as basic object units using the effective graph segmentation algorithm. Then according to those global feature vectors of sub-regions, the fuzzy c-means clustering is used to obtain the classification map based on these sub-regions. Experiments shows the results by different segmentation scales, and then turn out that the approach proposed in this paper can achieve better accuracy and efficiency.