Automated 3D volumetric reconstruction of multiple-room building interiors for as-built BIM

Abstract

Currently, fully automated as-built modeling of building interiors using point-cloud data still remains an open challenge, due to several problems that repeatedly arise: (1) complex indoor environments containing multiple rooms; (2) time-consuming and labor-intensive noise filtering; (3) difficulties of representation of volumetric and detail-rich objects such as windows and doors. This study aimed to overcome such limitations while improving the amount of details reproduced within the model for further utilization in BIM. First, we input just the registered three-dimensional (3D) point-cloud data and segmented the point cloud into separate rooms for more effective performance of the later modeling phases for each room. For noise filtering, an offset space from the ceiling height was used to determine whether the scan points belonged to clutter or architectural components. The filtered points were projected onto a binary map in order to trace the floor-wall boundary, which was further refined through subsequent segmentation and regularization procedures. Then, the wall volumes were estimated in two ways: inside- and outside-wall-component modeling. Finally, the wall points were segmented and projected onto an inverse binary map, thereby enabling detection and modeling of the hollow areas as windows or doors. The experimental results on two real-world data sets demonstrated, through comparison with manually-generated models, the effectiveness of our approach: the calculated RMSEs of the two resulting models were 0.089 m and 0.074 m, respectively.

Introduction

A Building Information Model (BIM) is a digital representation of facilities that records all information relevant to a building’s life cycle from construction to demolition, and includes three-dimensional (3D) design drawings, schedules, material properties, costs, and safety specifications [1], [2]. The use of BIM has been thoroughly examined by many researchers and practitioners, who have proved that it is particularly beneficial for reducing construction costs and errors while increasing productivity [1], [3], [4], [5], [6], [7], [8]. Nowadays, building owners, facilities management groups and governments are leading efforts to use BIM for new construction, renovation or refurbishment of a variety of facilities such as buildings, large shopping malls, subway systems, and airports.

It is important to note that the BIM created in the design stage of a facility is called as-designed BIM, and the BIM that reflects a facility in its as-built condition is called as-built BIM [2], [9], [10]. Typically, as-built BIMs are created for facilities that are not equipped with as-designed BIM or where the as-built conditions differ from the as-designed BIM [2]. As-built BIM has a number of potential benefits, for example detection of defects, timely quality control, support for decision making, and operation and maintenance of existing buildings [11], [12]. Meanwhile, as-designed BIM is not available for most existing buildings, as those had been built prior to the “BIM era” [13]. Even though some existing buildings have BIM, current building conditions still cannot be accurately reflected, because they periodically undergo renovation [14]. Therefore, the demand for efficient and accurate as-built BIM is significantly increasing in the academic, industrial, and governmental fields [15], [16], [17], [18], [19], [20].

The commonly used process for generation of as-built BIM is Scan-to-BIM by 3D laser scanning [9], [13], [21]. Typically in a BIM tool, the point cloud is used as a guide enabling modelers to effectively identify and trace the object’s shape [22]. For this conversion, unfortunately, the available commercial and academic tools require extensive human intervention, making them time consumptive, costly, and labor intensive [2], [23], [24], [25], [26]. Automatic as-built BIM creation from point-cloud data is a key objective, as it would reduce production costs while improving utilization of as-built BIM.

Typical as-built BIM creation involves three main aspects: geometric modeling of the components, attribution of categories and material properties to them, and establishment of their interrelations [27]. The geometric model serves as the main medium for the structuring and navigating of all of the other data contained in a BIM model [9]. In particular, geometric modeling of indoor scene is more challenging than modeling of outdoor scene due to high amount of clutter and occlusion [28]. The focus of this paper is to automate geometric as-built model creation for indoor environments containing multiple rooms. More specifically, we developed a framework to accomplish the following objectives:

• Propose an automated framework to model the rectilinear form of building interiors containing multiple rooms that uses only a point-cloud data with no additional observations

• Develop a methodology to filter raw point-cloud data contaminated by noise and clutter

• Develop a methodology to create the detail-rich volumetric wall components

• Adjust the generated models and evaluate them with thorough quantitative analysis

The remainder of this paper is organized as follows: Section 2 explores the challenges preventing automated as-built model creations for indoor environments using point-cloud data. Section 3 describes our main procedure for realization of detail-rich 3D volumetric as-built modeling for multi-room environments; Section 4 reports evaluations of the proposed approach with two real-world scanned data sets and discusses the problems that need to be improved; and finally, Section 5 draws conclusions and anticipates future work.