Plot-level trunk detection and reconstruction using one-scan-mode terrestrial laser scanning data

The applicability of terrestrial laser scanning (TLS) data for quantitative forest inventory has received increasing attention in the last decade. So far, research has been carried out for individual trunk modeling and plot-level forest parameter determination mainly from multi-scan-mode (MSM) TLS data. While MSM data, on the average, provide whole coverage of trunk and potentially lead to high reconstruction accuracy, it is of rising practical interest to study how well one-scan-mode (OSM) laser data could provide plot-level forest information, e.g. location, number of trees and breast height diameter of individual trees, in mainly one-storey stands, with lower expense, faster data collection and enhanced processing to collect e.g. reference and calibration data for airborne laser scanning based forest inventory. In general, to achieve plot-level trunk modeling, three main problems need to be solved. First, meaningful laser points need to be identified from data set, originally consisting of several millions points, for computational reasons; second, trunk points need to be recognized as precisely as possible, to facilitate localized modeling process; third, trunk reconstruction needs to be automatic and computationally acceptable, to give certain level details, but still enable fast processing. In this paper, a new tree detecting and trunk modeling mechanism is proposed, based on point distribution analysis, trunk finding and slice-by-slice circle fitting. The emphasis of this paper is on exploring the applicability of OSM laser data for plot-level inventory and automatic solution. The test area is a pine-dominated forest. Reference measurements from intensity image are used for validation. Experimental result shows that OSM-based TLS data is feasible for plot-level automation to deliver basic plot level information: detection of the most of the trunks, and reconstructing the DBH.