Fusion classification of multispectral and panchromatic image using improved decision tree algorithm
Fusion classification of multispectral and panchromatic image using improved decision tree algorithm
复制标题
DOI：
10.1109/icspct.2014.6884944
复制
发表时间：
2014-07
期刊：
影响因子：
--
通讯作者：
P. Shingare;Priya M. Hemane;Duhita S. Dandekar
中科院分区：
文献类型：
--
作者：
P. Shingare;Priya M. Hemane;Duhita S. Dandekar
本文从卫星图像中提取植被、水体、土壤、建筑物等区域。Landsat 7 ETM+卫星用于图像数据集。它给出了低分辨率的多光谱图像和高分辨率的全色图像。为了检测城市区域的特征，我们需要在图像中的空间和光谱信息。因此，这两个图像首先使用不同的方法融合。然后将得到的融合图像用于各个区域的分类。采用决策树算法对图像进行分类。决策树规则采用NDVI、NDWI、SAVI和NDBI等多种指数。在NDBI公式中做了一些修改。最后，将改进后的NDBI算法与原NDBI算法的结果进行了比较，发现改进后的NDBI决策树算法具有更高的精度。
In this paper, efforts are made to detect the areas such as vegetation, water, soil, built-up area etc. from the satellite images. Landsat 7 ETM+ satellite is used for data set of images. It gives multispectral image with low resolution and panchromatic image with high resolution. For detecting the features of the urban area we require both spatial and spectral information in image. Hence these both images are first fused using different methods. Resultant fused image is then used for classification in various areas. Decision tree algorithm is used to divide the image in various classes. Various indexes such as NDVI, NDWI, SAVI and NDBI are used as decision tree rules. Some modification is done in the NDBI formula. The final results of decision tree algorithm using original and modified NDBI are compared and it was found that the decision tree algorithm using modified NDBI gives more accurate results.