Remote sensing data fusion using support vector machine

The paper introduces a support vector machine (SVM) into research of remote sensing data fusion, and a new approach of remote sensing data fusion based on SVM is presented. In the selected test area in Shaoxing City, Zhejiang Province, China, a data fusion experiment was conducted using Landsat TM multispectral data (30 m) and SPOT-4 panchromatic (PAN) data (10 m). The results show that the overall classification accuracy of the fusion data reached 76.8%. The new fusion method could detect effectively the ground objects, which have close spectrum.