The Method of Matching Single Tree Information Extracted by Point Cloud to the Reference Data from Field Work through Bidirectional Selection

Objective: Based on the principle of bidirectional selection and judgment, a method was proposed to reasonably match the individual tree information extracted from point cloud data(LiDAR) with the reference information measured by the field work. Method: Using airborne LiDAR point cloud data, individual tree information such as tree position, number, height, and crown diameter was extracted. Firstly, the candidate reference trees were selected according to the information of the LiDAR tree. Then whether such candidate trees were the most reasonable LiDAR trees from the reference tree or not were evaluated again. Result: The matching accuracy, the heights and crown diameters accuracy after matching were used as the accuracy indicators. Compared with the other three commonly used matching methods, the height accuracy of individual tree using the proposed matching method was increased from 75.21% to 91.01%, and the crown diameter accuracy was also increased from 60.50% to 68.64% under the conditions with the same matching accuracy. When the height and crown diameter accuracy were controlled with the same value, the proposed method improved the matching accuracy from 33.52% to 61.11% comparing to the traditional method. Conclusion: The proposed method in this paper could match the single tree information quickly and efficiently between the ones extracted by remote sensing and the reference information measured on the field work. Compared with the traditional method, it could show some superiority when used in high-density and multi-layer stands.