本研究以三明尤溪某地作为研究区,以DJI Phantom 4 Pro V2.0无人机航测影像为数据源,采用面向对象分类方法进行植被分类。经测试,取尺度参数100、形状参数0.3和紧致度参数0.5分割效果最好。利用光谱、纹理、形状等信息创建规则进行分类...本研究以三明尤溪某地作为研究区,以DJI Phantom 4 Pro V2.0无人机航测影像为数据源,采用面向对象分类方法进行植被分类。经测试,取尺度参数100、形状参数0.3和紧致度参数0.5分割效果最好。利用光谱、纹理、形状等信息创建规则进行分类,将地物细分为芦苇、芒萁、杉木、马尾松、竹林、水田、百香果、柑橘、茶树、草地和3种农作物等13种植被类型,房屋、道路、水体、裸地等4种非植被类型,总体精度达到85.2%,Kappa系数为0.84。展开更多
With the wide use of color in many areas, the interest on the color perception and processing has been growing rapidly. An important topic in color image processing is the development of efficient tools capable of fil...With the wide use of color in many areas, the interest on the color perception and processing has been growing rapidly. An important topic in color image processing is the development of efficient tools capable of filtering images without blurring them and without changing their original chromatic contents. In this paper, a new technique reducing noise of color image is developed. A class of color-scale morphological operations is introduced, which extend mathematical morphology to color image processing, representing a color image as a vector function. The correlation between color components is utilized to perform noise removal. Color-scale morphological niters with multiple structuring elements (CSMF-MSEs) are proposed. Their properties are discussed and proved. Experimental results show that CSMF-MSEs are suitable and powerful to eliminate noise and preserve edges in color image because of efficient utilization of inherent correlation between color components, and they perform better than vector展开更多
文摘本研究以三明尤溪某地作为研究区,以DJI Phantom 4 Pro V2.0无人机航测影像为数据源,采用面向对象分类方法进行植被分类。经测试,取尺度参数100、形状参数0.3和紧致度参数0.5分割效果最好。利用光谱、纹理、形状等信息创建规则进行分类,将地物细分为芦苇、芒萁、杉木、马尾松、竹林、水田、百香果、柑橘、茶树、草地和3种农作物等13种植被类型,房屋、道路、水体、裸地等4种非植被类型,总体精度达到85.2%,Kappa系数为0.84。
基金Supported by the Natural Science Foundation of China,No.69775004
文摘With the wide use of color in many areas, the interest on the color perception and processing has been growing rapidly. An important topic in color image processing is the development of efficient tools capable of filtering images without blurring them and without changing their original chromatic contents. In this paper, a new technique reducing noise of color image is developed. A class of color-scale morphological operations is introduced, which extend mathematical morphology to color image processing, representing a color image as a vector function. The correlation between color components is utilized to perform noise removal. Color-scale morphological niters with multiple structuring elements (CSMF-MSEs) are proposed. Their properties are discussed and proved. Experimental results show that CSMF-MSEs are suitable and powerful to eliminate noise and preserve edges in color image because of efficient utilization of inherent correlation between color components, and they perform better than vector
