Abstract: A single wood feature parameter extraction and detection method based on 3D point cloud modeling was proposed to improve the convenience and accuracy of non-contact measurement of single wood breast diameter using 3D point cloud. Firstly, the IMAGER 5010 terrestrial 3D laser scanner was used to obtain single tree multi-directional point cloud data, and the nearest point search method was used to achieve multi-directional point cloud registration. Secondly, an improved bilateral filtering algorithm was used to eliminate small-scale point cloud noise, and then point cloud data was simplified using the envelope grid method by combining statistical filtering and geometric filtering to remove large-scale point cloud noise. Then, the scattered point clouds at the height of a single tree chest using ant colony algorithm were optimized and sorted. Finally, cubic B-spline contour fitting and single tree diameter calculation were performed on the sorted point cloud at the chest height. The experimental results showed that the method proposed in this paper can effectively achieve multi-directional point cloud registration and filter out point cloud noise of different scales. The ant colony algorithm can approach the global optimal solution well and has high convergence efficiency. After optimizing the sorting of scattered point clouds at the chest height using the ant colony algorithm, the measurement accuracy and stability of single wood breast diameter are improved significantly. The relative measurement error is within 1.5% of the technical specifications, and the long-distance non-contact rapid and accurate measurement of single wood breast diameter can be achieved.

Key words: measurement of single-tree breast diameter, multi directional point cloud registration, large and small scale point cloud denoising, ant colony, cubic B-spline curve fitting