Abstract
Abstract. This paper presents a Maximum Sequential Similarity Reasoning (MSSR) algorithm based method for co-registration of 3D TLS data and 2D floor plans. The co-registration consists of two tasks: estimating a transformation between the two datasets and finding the vertical locations of windows and doors. The method first extracts TLS line sequences and floor plan line sequences from a series of horizontal cross-section bands of the TLS points and floor plans respectively. Then each line sequence is further decomposed into column vectors defined by using local transformation invariant information between two neighbouring line segments. Based on a normalized cross-correlation based similarity score function, the proposed MSSR algorithm is then used to iteratively estimate the vertical and horizontal locations of each floor plan by finding the longest matched consecutive column vectors between floor plan line sequences and TLS line sequences. A group matching algorithm is applied to simultaneously determine final matching results across floor plans and estimate the transformation parameters between floor plans and TLS points. With real datasets, the proposed method demonstrates its ability to deal with occlusions and multiple matching problems. It also shows the potential to detect conflict between floor plan and as-built, which makes it a promising method that can find many applications in many industrial fields.
Highlights
Terrestrial Laser Scanning (TLS) can provide dense, highly accurate 3D information, which makes it become a key data acquisition tool providing accurate, detailed spatial information
We present a Maximum Sequential Similarity Reasoning (MSSR) algorithm based method for co-registration of 3D TLS data and 2D floor plans
In order to overcome the difficulty and uncertainties caused by occlusions and similar building feature structures, we propose a Maximum Sequential Similarity Reasoning (MSSR) based method to re-represent the line sequences and estimate the locations of windows and doors
Summary
Terrestrial Laser Scanning (TLS) can provide dense, highly accurate 3D information, which makes it become a key data acquisition tool providing accurate, detailed spatial information. When the facade features (e.g., windows and doors) have similar structures or some parts of a facade are occluded, it may produce wrong results Another related research is (Khoshelham et al, 2009), in which a method for automated point cloud-to-map registration using a plane matching technique is presented to georeference the 3D point clouds using 2D maps. Because it uses 2D building footprint maps for the registration, which have less detailed building semantic and geometric information than floor plans, the registration cannot further facilitate TLS data processing towards 3D building modelling and other aforementioned industrial applications.
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
