4种人工林下可燃物的计算机图像分类  

Classification of 4 Fuels under different Plantations by Their Computer Images

作　　者：金森[1] 王亮[1] 

机构地区：[1]东北林业大学,哈尔滨150040

出　　处：《东北林业大学学报》2003年第5期1-3,共3页Journal of Northeast Forestry University

基　　金：国家自然科学基金青年基金项目 (30 0 0 0 1 35);哈尔滨市学科后备带头人基金项目资助

摘　　要：对水曲柳 (Fraxinusmandshurica)、胡桃楸 (Juglansmandshurica)、蒙古栎 (Quercusmongolica)和落叶松(Larixgemlini)人工林下的可燃物的图像进行特征提取 ,建立了以距离为 5个像素的自相关系数、距离为 10个像素的自相关系数和行程长度为 8的频数这三个特征为变量的线性分类器 ,将落叶松、蒙古栎、胡桃楸和水曲柳人工林下的可燃物较好地区分开来 ,分类准确率达 91% ,而且分类能力比较稳定。Characteristics of images for four kinds of fuels under different plantations, namely, Manchurian ash, Manchurian walnut , Mongolian oak and larch, were extracted, including Euler number, autocorrelation coefficients at different distance and travel length frequency. A linear classification machine was established based on the autocorrelation coefficients at the distance of 5 and 10 pixels and the number of travel length of 8. The results indicate that the classification machine can group the four fuels well with 91% correctness and steady performance.

关 键 词：人工林 可燃物 计算机图像 图像分类 水曲柳 胡桃楸 蒙古栎 落叶松 

分 类 号：S762[农业科学—森林保护学]