Early Warning Method for Sea Typhoons using Remote-Sensing Imagery Based on Improved Support Vector Machines (SVMs)

Because traditional risk warning methods using early mass analysis of remote-sensing images have low efficiency, a typhoon warning method based on analysis of remote sensing imagery using improved support vector machines (SVMs) is proposed. The remote-sensing image data of the typhoon area are reconstructed by a wavelet transform and multiscale analysis algorithm. A redundant collective algorithm is then used to eliminate the redundant attributes, and the degree of support theory is used to reduce the algorithm and establish an efficient typhoon warning model. The experimental results show that this algorithm combined with remote-sensing imagery can effectively predict a typhoon disaster. The speed of this warning method is 22% faster than the traditional method.