Hyperspectral Image Classification Based on Compound Kernels of Support Vector Machine
Yuyong Cui +, Zhiyuan Zeng and Bitao Fu
Digital Engineering and Simulation Research Center, Huazhong University of Science and Technology, Wuhan 430074, China
Abstract. Support vector machine is a kind of pattern classification algorithm based on the statistics learning theory. This paper proposes to estimate abundances from hyperspectral image using probability outputs of support vector machines (SVM), training a SVM with a gauss kernel function,and we discussed the relationship between kernel functions and nonlinear mappings and mapped spaces. Then, a new kernel - compound kernel function is given. We applied the kernel in SVM and compared the kernel with other kernels in Hyperspectral image classification, and give a comparison with some result to present the validation in remote sensing, the result shows the evidence to the validity of the method and shows that this method is more accurate than the other method, which has a more accurate result.
Keywords: support vector machine, compound kernels, hyperspectral image, classification, accuracy
In: Wan, Y. et al. (eds) Proceeding of the 8th international symposium on spatial accuracy assessment in natural resources and environmental sciences, World Academic Union (Press).