Extraction Of Land Cover Information From SPOT5Based On Characteristic Bands In Taibai Mountain, China

With high spatial resolution, SPOT5satellite imagery contains rich shape and texture information. However, traditional statistical classification technology based on the spectrum of pixels couldn’t make full use of the shape and texture information of SPOT5satellite imagery because SPOT5-XS imagery only includes four multi-spectral bands, i.e. green(G) with wavelengths ranging from0.50to0.59microns, red(R)0.61to0.68, near infrared (NIR)0.78to0.89and shortwave infrared (SWIR)0.15to1.75. If only these four bands are applied to identify land cover by using remote sensing data, it will often bring some problems, such as the divisibility among various land cover types reduced and the automated information extraction more difficult, which mainly results in the lack of band information and complexity of spectrum. Therefore spectrum characteristic, relations among bands were analyzed and many characteristic bands were built through Band Operation method and merged into the original SPOT5-XS imagery, which includes Normalized Difference Water Index (NDWI), Normalized Difference Vegetation Index (NDVI), Normalized Difference soil Index (NDSI), Adjusted Normalized Difference Vegetation Index (ANDVI), Sum of Vegetation Index (SVI). Object-oriented classification approach based on patches and decision tree classification method based on pixels was then applied to the new generated imagery. Then by using the same evaluation samples, Error Matrix for accuracy assessment of these two classification ways was formed. The results indicate that the classification rates from both of these two classification methods are high. However, compared to the result from Decision tree classification, the precision from object-oriented method is high and it also avoids the "salt and pepper" noise. Based on above analysis, methods based on the new constructed spectrum features, and the use of texture and shape feature could take advantage of spatial information of high resolution imagery, improve the divisibility among various land cover types and make the information extraction more easily and the high classification accuracy obtained. What’s more, ideal segmentation scales are useful to improve the classification precision.