基于遥感影像增强的林地植被树种分类研究

Classification of forest vegetation species based on remote sensing image enhancement

在基于遥感影像的植被树种提取研究中,受气候、光照等外界因素的影响,导致遥感影像模糊,难以区分树种。为此,本文开展了基于遥感影像增强的林地植被树种分类研究。结合Retinex增强和引导滤波两种方法,增强林地植被树木遥感影像视觉效果、影像颜色及影像边缘细节后,提取增强后影像中林地的线性地形、植被指数和纹理特征,组成特征集输入旋转森林算法中,获取待分类样本的最大可信度值,由此得出林地植被树种分类结果。结果表明,该方法的遥感影像增强性能良好,影像结构相似性均在0.90以上,可显著提升影像视觉效果和质量,清晰呈现山脊和山谷的地形线性特征,实现杉木、油茶树、松树、映山红和草地的分类,Kappa系数高达0.805。

It is often difficult to distinguish tree species in the research of tree species extraction based on remote sensing images due to the influence of such external factors as climate and lighting,resulting in blurred remote sensing images.Therefore,a study on forest vegetation tree species classification based on remote sensing image enhancement is conducted in this paper.The visual effect,image color and image edge details of remote sensing images of forest vegetation and trees are enhanced by combining Retinex enhancement and guided filtering.The linear terrain,vegetation index,and texture features of the enhanced images are extracted,the feature set is input into the rotating forest algorithm to obtain the maximum confidence value of the samples to be classified,and the classification results of forest vegetation species are obtained.The results show that the remote sensing image enhancement performance from this method is good,and the similarity of image structure is above 0.90.It can significantly improve the visual effect and quality of images,clearly present the linear features of mountain ridges and valleys,and get the classification of fir,camellia oleifera,pine,azalea,and grassland,and the Kappa coefficient is 0.805.