基于小波变换和随机森林的森林类型分类研究

以黑龙江省凉水国家级自然保护区为研究区，采用“高分一号”卫星提供的多光谱影像作为遥感数据源，通过对遥感影像进行小波变换处理，之后选择植被指数、纹理特征、地形因子作为分类特征，利用随机森林算法对该地区森林类型进行分类。结果表明，遥感图像在进行小波变换后，基于随机森林算法的森林类型分类精度为91.68%，Kappa系数为0.90，较未进行小波变换时的分类精度提高10.67%。总体来看，结合小波变换的随机森林分类方法可以获得比较高的分类精度。为森林类型分类提供一种新的思路，且为提高森林类型分类精度提供一种参考方法。

Classification of Forest Types Based on Discrete Wavelet Transform and Random Forests from GF-1 Images

Taking Liangshui Nature Reserve of Heilongjiang Province as the research object,the multi spectral images of the region from “Gaofen No.1” satellite were processed by wavelet transform.Vegetation index,texture and terrain factors were extracted from the images.Forest types of the region were classified by using the random forest algorithm.The results showed that random forests model achieved an overall accuracy of 91.68% and Kappa coefficient of 0.90 after the images were preprocessed with discrete wavelet transform,compared with the overall accuracy (81.01%) of random forests model established with original images.All the above illustrated that wavelet threshold denoising models could denoise the original images effectively,and the proposed method was able to achieve highly satisfactory forest type classification results.The method could provide a new thought for the classification of forest types and a new way to increase the classification accuracy.