机构地区: 石河子大学农学院农业资源与环境系
出 处: 《西北农业学报》 2006年第3期45-49,共5页
摘 要: 利用图像分析方法,通过准确识别冠层和背景像素进行棉花冠层生物学产量和叶面积系数估测。采用Olympus C740 Ultra Zoom数码相机拍摄棉花不同生育期冠层图像,在棉花冠层数码照片特征分析的基础上提出了棉花冠层图片计算机自动判读的方法,即混合采用图像色度(H)、绿光(G)、红光(R)灰度值构造提取条件,通过多重判断识别棉花冠层和背景,并编写了相应的计算机程序。利用该程序分析棉花不同施氮量下、不同生育期提取地面覆盖度参数与棉花生物学产量、叶面积系数间的关系,发现棉花冠层地面覆盖度指标可以有效预测棉花生物学产量和叶面积系数,二者间指数相关系数达到r=0.97以上,为极显著相关。 This research is attempted to buildup an innovational method based on object features of cotton canopy digital images to identify plant and background pixels and hence to estimate percent ground cover of vegetation (PGCV) and leaf area index (LAI). Cotton canopy was captured by using an Olympus C740 Ultra Zoom digital camera. A protocol mixed hue of HLS color space with R, G value was setup and multiple judgment process was designed to extract cotton canopy pixels from background noise. A computer program based on the protocol above was designed simultaneity. Relation betwcen PGCV that derived from the designed program and cotton canopy biomass or LAI with differ cnt N application and in different growth stages suggested that PGCV could reliably evaluate both PGCV and LAI, empirical statistical showed that a significantly high coefficient between PGCV and cotton canopy biomass or LAI reached (r=0.97, respectively). The results indicated that image anal ysis is a promising method for quick diagnosis of crop growth and development.