基于高分影像的面向对象土地利用变化检测方法研究

以2013年与2014年2期高分一号卫星影像为数据源进行浙江省海盐县沿海地带土地利用类型变化检测。面向对象的变化检测方法包括单一波段差值比值法、多层次多波段差值比值法、变化矢量分析法，为对比检测效果还进行了基于像元的多波段差值法与比值法对多光谱影像与融合影像的检测。结果表明，面向对象的变化检测总体精度为86.29%，Kappa系数为0.72，优于基于像元的变化检测方法。在面向对象的变化检测中，运用融合影像进行分层次多波段差值比值法得到的检测效果最好，优于变化矢量分析法。而基于像元的变化检测中，运用融合影像进行多波段差值法得到的检测效果较好。

Object-oriented Detection of Land Use Changes Based on High Spatial Resolution Remote Sensing Image

The remote sensing images from GF-1 satellite in 2013 and 2014 were used to detect the land use changes of the coastal zone in Haiyan County, Zhejiang Province. Two detective methods were compared through the pixel-based and object-oriented changes. In the pixel-based land use change detection, the multi-band difference and ratio methods were used to detect the land use changes based on multi-spectral and fusion images. In the object-oriented land use change detection, the effect of multi-spectral and fusion images were researched using multi-band difference and ratio methods on the single-level and multi-level. On this basis, the detection results of human activities region combined with shape features were analyzed. In addition, the change vector analysis (CVA) was adopted to conduct land use change detection basing on the multi-spectral and fusion images. The results showed that the overall accuracy of object-oriented land use change detection was 86.29%, and Kappa coefficient was 0.72, which were better than those of the pixel-based land use change detection. In the object-oriented land use change detection, the detection results of multi-level multi-band difference and ratio methods which using fusion image were the best, and they were better than those of CVA. And in the pixel-based land use change detection, the results of multi-band difference method which used fusion image were better than those of the other methods.