28 October 2014 Fuzzy texture model and support vector machine hybridization for land cover classification of remotely sensed images
Author Affiliations +
Abstract
Accuracy of land cover classification in remotely sensed images relies on the utilized classifier and extracted features. Texture features are significant in land cover classification. Traditional texture models capture only patterns with discrete boundaries, whereas fuzzy patterns should be classified by assigning due weightage to uncertainty. When a remotely sensed image contains noise, the image may have fuzzy patterns characterizing land covers and fuzzy boundaries separating them. Therefore, a fuzzy texture model is proposed for the effective classification of land covers in remotely sensed images. The model uses a Sugeno fuzzy inference system. A support vector machine (SVM) is used for the precise, fast classification of image pixels. The model is a hybrid of a fuzzy texture model and an SVM for the land cover classification of remotely sensed images. To support this proposal, experiments were conducted in three steps. In the first two steps, the proposed texture model was validated for supervised classifications and segmentation of a standard benchmark database. In the third step, the land cover classification of a remotely sensed image of LISS-IV (an Indian remote sensing satellite) is performed using a multivariate version of the proposed model. The classified image has 95.54% classification accuracy.
© 2014 Society of Photo-Optical Instrumentation Engineers (SPIE)
S. Jenicka, A. Suruliandi, "Fuzzy texture model and support vector machine hybridization for land cover classification of remotely sensed images," Journal of Applied Remote Sensing 8(1), 083540 (28 October 2014). https://doi.org/10.1117/1.JRS.8.083540 . Submission:
JOURNAL ARTICLE
17 PAGES


SHARE
Back to Top