基于SfM的城市树木参数提取研究

Usage of Structure-from-Motion for Urban Forest Inventory

  • 摘要: 通过在城市森林样地水平的树干三维模型重建与估算胸径试验(半径6 m圆形样地),研究运动恢复结构(SfM)算法在城市森林调查中应用的可行性,并将SfM算法得到的胸径与手持式激光雷达结果进行比较。结果表明:在胸径估算结果上,SfM算法的rRMSE为5.38%(rBias=2.74%),手持式激光雷达的rRMSE为3.35%(rBias=1.04%),2种方法的估测值均与野外实测值高度呈正相关( R 2大于0.97),满足树木参数提取精度要求。SfM算法成本低、易操作,具有创新意义,且估算树干胸径在森林资源调查允许误差范围内,在城市森林资源调查中具有一定应用价值。

     

    Abstract: In this paper, we evaluated the urban trees reconstruction and diameter at breast height(DBH) measurements by SfM based on a circular urban forest plot(radius = 6 m) to verify the feasibility for urban forest inventory. Furthermore, hand-held laser scanning was also used as a counterpart technique. The results indicated that SfM can obtain DBH of urban trees with satisfying accuracy for urban forest inventory measurements(rRMSE = 5.38%, rBias= 2.74%), which was similar to the hand-held laser scanner(rRMSE = 3.35%, rBias = 1.04%). Both of the evaluated DBH from SfM and hand-held laser scanner were highly and positively correlated with that from field measurements( R 2>0.97). SfM is a revolutionary, low-cost, and user-friendly technique, which can also meet the requirements of tree-parameter precision. Thus, the SfM technique is valuable in urban tree inventory.

     

/

返回文章
返回