Abstract: Abstract: For the assessment and improvement of the safety performance as well as historical and cultural inheritance of ancient wooden architecture, the dimension information of various key components of such architecture acts as the important foundation. However, an extraction method for such dimension information with high efficiency and accuracy is rarely reported. It is well acknowledged that the three dimensional (3D) laser scanning technology has the potential to provide a scientific solution for this problem. However, the point cloud data obtained by 3D laser scanning technology is usually enormous, and the dimension information herein cannot be directly obtained from this data. According to the important characteristics of the key component (i.e. multi-level of details (multi-LoD)) in ancient wooden architecture, a preliminary framework of multi-LoD models is proposed for various types of key components in the ancient wooden architecture, and the correspondingly salient geometric feature parameters, which aim to represent the dimension information of key components, are also recommended according to different LoD. Based on these multi-LoD models and massive high-fidelity point cloud data, an automatic extraction method of multi-LoD dimension information for the key components in ancient wooden architecture is proposed. This method is considered to be capable of accurately and efficiently extracting the multi-LoD dimension information of key components. To validate the reliability and high efficiency of this method, multi-LoD dimension information of two typical key components are extracted using the proposed method. The results indicate that this method is capable of extracting dimension information from millions of point cloud data within 7 minutes. Furthermore, the relative and absolute errors of such information are less than 2% and 0.5 mm respectively, thus validating the high efficiency and reliability of the proposed method.

Key words: Keywords: key components of ancient wooden architecture, LoD model, salient geometric features, point cloud data, automatic extraction method