基于树影与快鸟图像的单木树高提取

Extraction of Individual Tree Height Using Quickbird Images Based on Tree Shadow

利用黑龙江省塔河地区2008年的快鸟影像,研究了基于高空间分辨率的卫星影像的单木树高提取方法。在试验区内,实测孤立木的树高、胸径和冠幅,并在遥感图像对应位置上一一标记。在对快鸟影像进行裁剪、几何校正等处理的基础上,提取孤立木树冠顶点区域在遥感图像上的灰度值,然后建立灰度值与其树高之间的回归模型,其决定系数达到0.88,并依此计算树高。同时,根据树影长度、太阳方位角和坡度等因子,用几何光学的方法计算单木树高。最后,用地面实测的树高数据对回归模型和几何光学模型估测的树高进行检验。回归模型的平均精度达到85.56%,几何光学模型精度达到80.45%。回归模型优于几何光学模型。证明了采用回归模型方法提取单木树高的有效性和可行性。

A study was conducted to extract individual tree height using high spatial resolution remote sensing images and the Quickbird images of Tahe area in Heilongjiang Province in 2008.In the experimental area,the height,diameter at breast height（DBH） and crown width of the isolated trees were measured,and these trees were marked respectively on the remote sensing images.Based on the fusion and geometric correction of the Quickbird images,the gray value of the crown peak area of the isolated trees were obtained,and then a regression model of the relationship between gray value and tree height was established,with a coefficient of determination of 0.88.The model was then applied to calculate tree height.Moreover,the height of the individual trees was calculated according to the shadow length,sun＇s azimuth,slope gradient and some other factors by the geometrical optics method.The estimated height from the regression model and the geometric optics model was tested by the measured height from ground level.Result showed that the mean accuracy of the regression model reached 85.56%, while that of the geometric optics model was 80.45%.The regression model was better than the geometric optics model.The extraction of individual tree height by regression model was proved to be effective and feasible.