Abstract
Geo-referenced 3D models are currently in demand as an initial knowledge base for cultural heritage projects and forest inventories. The mobile laser scanning (MLS) used for geo-referenced 3D models offers ever greater efficiency in the acquisition of 3D data and their subsequent application in the fields of forestry. In this study, we have analysed the performance of an MLS with simultaneous localisation and mapping technology (SLAM) for compiling a tree inventory in a historic garden, and we assessed the accuracy of the estimates of diameter at breast height (DBH, a height of 1.30 m) calculated from three fitting algorithms: RANSAC, Monte Carlo, and Optimal Circle. The reference sample used was 378 trees from the Island Garden, a historic garden and UNESCO World Heritage site in Aranjuez, Spain. The time taken to acquire the data by MLS was 27 min 37 s, in an area of 2.38 ha. The best results were obtained with the Monte Carlo fitting algorithm, which was able to estimate the DBH of 77% of the 378 trees in the study, with a root mean squared error (RMSE) of 5.31 cm and a bias of 1.23 cm. The proposed methodology enabled a supervised detection of the trees and automatically estimated the DBH of most trees in the study, making this a useful tool for the management and conservation of a historic garden.
Highlights
Compiling inventories of elements and cataloguing heritage spaces are all necessary operations for preserving any cultural landscape [1].Some cultural spaces that have been designated as historic gardens have singular characteristics that hinder the use of new sensors to compile tree inventories
Studies comparing terrestrial laser scanning (TLS) and mobile laser scanning (MLS) based on simultaneous localisation and mapping technology (SLAM) for use in forest inventories on various terrains revealed that MLS detected 96% of the trees, whereas TLS detected 78.5% [20], and took less time to collect the data
In our study there were 120 (32%) cases in which could not use geometric shape detection algorithms. These 120 specimens correspond eiwe could not use geometric shape detection algorithms. These 120 specimens correspond ther to trees with a DBH below 10 cm, or to trees located at the edges of the surveyed area, either to trees with a DBH below 10 cm, or to trees located at the edges of the surveyed and have been exclude from the DBH analysis
Summary
Some cultural spaces that have been designated as historic gardens have singular characteristics that hinder the use of new sensors to compile tree inventories. Landscape elements, such as the morphology of paths and hedges, signage, and ornamental fountains and statues, cause occlusion when collecting data by laser scanner [2], so an essential phase of this study was to develop an adequate methodology, including the optimum planning of the data acquisition process. Some studies on the quality of the point cloud obtained by MLS report a problem in the model due to noise [14] or errors in fitting the geometric shapes [22]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
