Evaluation of local features for scene classification using VHR satellite images
Authors:
- 1 April 2011
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 23340932,p. 385-388
Abstract
We compare the scene classification performance of 13 features, including structure, texture and color features. First, image classification are performed using a single feature and the performance of different features are compared. Both the k-nearest-neighbor (KNN) classifier and the support vector machine classifier (SVM) are employed. And for the KNN classifier, we use four different distance measures. Then, according to the classification results, three of these features with good performance are combined by simple concatenation. The combined feature is subsequently used for classification. This yields an overall comparison of the 13 features. Experiments on the very high resolution satellite images reveal that the combined feature consistently outperforms the other features and improves the results obtained.Keywords
This publication has 15 references indexed in Scilit:
- Evaluation of GIST descriptors for web-scale image search2009
- Matching Local Self-Similarities across Images and Videos2007
- Fast anisotropic gauss filteringIEEE Transactions on Image Processing, 2003
- Video Google: a text retrieval approach to object matching in videos2003
- Constructing models for content-based image retrieval2001
- Representing and Recognizing the Visual Appearance of Materials using Three-dimensional TextonsInternational Journal of Computer Vision, 2001
- The Earth Mover's Distance as a Metric for Image RetrievalInternational Journal of Computer Vision, 2000
- Color-based object recognitionPattern Recognition, 1999
- Color indexingInternational Journal of Computer Vision, 1991
- Two-dimensional spectral analysis of cortical receptive field profilesVision Research, 1980