Hyperspectral image classification has been well acknowledged as one of the challenging tasks of hyperspectral data processing. In this paper, we propose a novel hyperspectral image classification framework based on local binary pattern (LBP) features and PCANet. In the proposed method, linear prediction error (LPE) is first employed to select a subset of informative bands, and LBP is utilized to extract texture features. Then, spectral and texture features are stacked into a high dimensional vectors. Next, the extracted features of a specified position are transformed to a 2-D image. The obtained images of all pixels are fed into PCANet for classification. Experimental results on real hyperspectral dataset demonstrate the effectiveness of the proposed method.
Urban land accounts for a small fraction of the Earth’s surface area but rapid increases in urban land have a disproportionate influence on the environment. China is a living laboratory in urbanization and has witnessed fast urban growth in recent decades. The timely and accurate mapping of urban land in China is an urgent and basic issue toward clarifying the urbanization process and revealing its environmental impacts. Nighttime stable light (NSL) data obtained by the Defense Meteorological Satellite Program’s Operational Linescan System (DMSP/OLS) can provide an economical way to map urban land nationwide. However, it is difficult to apply existing methods to accurately extract urban land from DMSP/OLS NSL data covering the entirety of China due to China’s large area and substantial regional variation. A stratified support vector machine (SSVM)-based method used to map the urban land in China in 2008 at a national scale using DMSP/OLS NSL and SPOT normalized difference vegetation index data is presented. The results show that measurement of urban land in China in 2008 using SSVM achieves an average overall accuracy of 90% and an average Kappa of 0.69. The success of this research demonstrates the great potential of SSVM for clarifying the urbanization process in continental and global research.