Terrestrial laser scanners (TLS) have demonstrated great potential in estimating structural attributes of forest canopy, such as leaf area index (LAI). However, the inversion accuracy of LAI is highly dependent on the measurement configuration of TLS and spatial characteristics of the scanned tree. Therefore, a modified gap fraction model integrating the path length distribution is developed to improve the accuracy of retrieved single-tree leaf area (LA) by considering the shape of a single-tree crown. The sensitivity of TLS measurement configurations on the accuracy of the retrieved LA is also discussed by using the modified gap fraction model based on several groups of simulated and field-measured point clouds. We conclude that (1) the modified gap fraction model has the potential to retrieve LA of an individual tree and (2) scanning distance has the enhanced impact on the accuracy of the retrieved LA than scanning step. A small scanning step for broadleaf trees reduces the scanning time, the storage volume, and postprocessing work in the condition of ensuring the accuracy of the retrieved LA. This work can benefit the design of an optimal survey configuration for the field campaign.
Building boundary is important for the urban mapping and real estate industry applications. The reconstruction of building boundary is also a significant but difficult step in generating city building models. As Light detection and
ranging system (Lidar) can acquire large and dense point cloud data fast and easily, it has great advantages for building
reconstruction. In this paper, we combine Lidar data and images to develop a novel building boundary reconstruction
method. We use only one scan of Lidar data and one image to do the reconstruction. The process consists of a sequence of three steps: project boundary Lidar points to image; extract accurate boundary from image; and reconstruct boundary
in Lidar points. We define a relationship between 3D points and the pixel coordinates. Then we extract the boundary in the image and use the relationship to get boundary in the point cloud. The method presented here reduces the difficulty of data acquisition effectively. The theory is not complex so it has low computational complexity. It can also be widely used
in the data acquired by other 3D scanning devices to improve the accuracy. Results of the experiment demonstrate that
this method has a clear advantage and high efficiency over others, particularly in the data with large point spacing.