遥感作为人类获取地表信息最重要的手段之一,对于湿地研究具有重要的价值。文章运用机器学习规则推理的方法克服常规遥感研究方法在湿地研究中的弊端,使分类精度得到很大的提高。与传统的最大似然法相比,分类精度提高了13.51%, 达到了83.81%。基于机器学习规则推理分类,不仅利用了光谱信息,而且利用了纹理信息和坐标信息,并使用了掩膜技术。研究发现,这种方法能使湿地类型以及湿地与高地之间的混淆现象得到有效的克服。
Wetland, which lies between land and aquatic systems, is one of the most important systems on the earth, and one of crucial links of matter and energy exchange in the earth systems. Remote sensing, one of the significant methods for the acquisition of the information on the earth surface, is a valuable tool to the researches of wetland. In the paper, the machine learning methodology is used to identify the wetland in order to overcome the shortcomings of traditional methodology in the research of wetland. Compared with traditional classification method, rule-based inferring method has improved the accuracy of classification by 13.81%. In the classification of wetland by using machine learning methodology, not only the spectral features of geo-objects, but also other features such as textural features, location features are used. Moreover, the mask technique is used too. It is proven that the technology can really improve the accuracy of classification of wetland greatly.
[1] Mitchell T M(著).曾华军,张银奎(译). 机器学习[M]. 北京:机械工业出版社, 2003.
[2] Huang X Q, Jensen J R. A machine-learning approach to automated knowledge-based building for remote sensing image analysis with GIS data [J]. Photogrammetric Engineering & Remote Sensing, 1997, 63 (10): 1185-1194.
[3] Sader S A, Ahl D, Liou W S. Accuracy of Landsat TM and GIS rule-based methods for forest wetland classification in Maine [J]. Remote Sensing of Environment, 1995, 53 (3): 133-144.
[4] Lunetta R S, Barlogh M E. Application of multi-temporal Landsat 5 TM imagery for wetland identification [J]. Photogrammetric Engineering & Remote Sensing, 1999,65 (11): 1303-1310.
[5] 陆健健. 中国湿地 [M]. 上海:华东师范大学出版社,1990.
[6] Haralick R M. Statistical and structural approaches to texture [J]. IEEEProceedings, 1979, 67 (5): 786-804.
[7] 夏德深,傅德胜. 现代图像处理技术与应用 [M]. 南京:东南大学出版社,1997.1~200.
[8] Narasimha Rao P V, Sesha Sai M V R, Sreenivas K, et al. Textural analysis of IRS-1D panchromatic data for land cover classification [J]. International Journal of Remote Sensing, 2002, 23 (17): 3327-3345.
[9] Lawrence R L, Wright A. Rule-based classification systems using classification and regression tree (CART) analysis[J]. Photogrammetric Engineering & Remote Sensing, 2001, 67 (10): 1137-1142.
[10] Congalton R G. A review of assessing the accuracy of classifications of remotely sensed data [J]. Remote Sensing of Environment, 1991, 37 (1):35-46.
