Sparse graph regularization for robust crop mapping using hyperspectral remotely sensed imagery with very few in situ data

详细信息    查看全文

• 作者：Zhaohui Xue^a ; Peijun Du^b ; ^c ; ^d ; ^{dupjrs@gmail.com} ; Jun Li^e ; Hongjun Su^a
• 关键词：Hyperspectral remote sensing ; Crop mapping ; Sparse graph regularization (SGR) ; Heihe watershed ; CASI/SASI
• 刊名：ISPRS Journal of Photogrammetry and Remote Sensing
• 出版年：2017
• 出版时间：February 2017
• 年：2017
• 卷：124
• 期：Complete
• 页码：1-15
• 全文大小：5531 K
• 卷排序：124

文摘

The generally limited availability of training data relative to the usually high data dimension pose a great challenge to accurate classification of hyperspectral imagery, especially for identifying crops characterized with highly correlated spectra. However, traditional parametric classification models are problematic due to the need of non-singular class-specific covariance matrices. In this research, a novel sparse graph regularization (SGR) method is presented, aiming at robust crop mapping using hyperspectral imagery with very few in situ data. The core of SGR lies in propagating labels from known data to unknown, which is triggered by: (1) the fraction matrix generated for the large unknown data by using an effective sparse representation algorithm with respect to the few training data serving as the dictionary; (2) the prediction function estimated for the few training data by formulating a regularization model based on sparse graph. Then, the labels of large unknown data can be obtained by maximizing the posterior probability distribution based on the two ingredients. SGR is more discriminative, data-adaptive, robust to noise, and efficient, which is unique with regard to previously proposed approaches and has high potentials in discriminating crops, especially when facing insufficient training data and high-dimensional spectral space. The study area is located at Zhangye basin in the middle reaches of Heihe watershed, Gansu, China, where eight crop types were mapped with Compact Airborne Spectrographic Imager (CASI) and Shortwave Infrared Airborne Spectrogrpahic Imager (SASI) hyperspectral data. Experimental results demonstrate that the proposed method significantly outperforms other traditional and state-of-the-art methods.