An image-driven integrated system to construct as-is IFC BIM of existing buildings

Abstract

The existing buildings have been operated and maintained using traditional Operations and Maintenance (O&M) strategies and methods. However, building O&M has been notoriously known for its inefficient and ineffective practice, which often causes tragic accidents, and is consequently unsatisfactory to building occupants and owners. Building Information Modeling (BIM) could support different activities throughout the life cycle of a building and has been widely applied in design and construction phases nowadays. It has shown promising opportunities and advantages in BIM applications for the improvement in O&M. However, BIM has not been widely implemented in the O&M phase. As-is BIM for existing buildings and updating the models is still considered to be a time-consuming process that requires great effort, high cost, and skilled workers. Moreover, lack of accurate and complete as-is information is one of the key reasons leading to the low-level efficiency in O&M. Hence, an effective and convenient approach to record as-is conditions of the existing buildings and create as-is BIM would be the essential step for improving efficiency and effectiveness of O&M, and furthermore possibly refurbishment of the building. Firstly, this study conducts a systematic review and analysis for image-based as-is BIM construction techniques and methods to highlight the importance of the as-is BIM generation methods. Then, this research provides a comprehensive and comparative analysis of current practices of building O&M without and with BIM assists in a systematic and dynamic way based on the activity theory to figure out the practices and major problems of as-is BIM implementation in building O&M. In order to solve the research gaps, this study presents an image-driven integrated system to construct IFC-based as-is BIM of existing buildings. This integrated system includes three modules: (1) Module 1; to extract text information from CAD drawings and further support constructing building framework. (2) Module 2; to propose an image-based recognition system to recognize building objects and their materials from images and provide complementary building information for further create IFC BIM in Module 3. (3) Module 3; to develop the IFC BIM generation system to integrate building information from Module 1 and 2 and further create the IFC BIM for target building. In details, Module 1 extracts text information and locations from the CAD drawings using the Optical Character Recognition (OCR) algorithm. The Module 2 supports recognizing structure object types and their corresponding materials according to one single image collected by a handheld digital camera using Neuro Fuzzy System (NFS) with image classification techniques (supported by a developed texture library). With the goal of creating an effective, convenient and low-cost approach to construct as-is BIM, the developed system is further validated and tested in real projects. Knowledge gained from this research and the proposed computational system will form a solid foundation and a comprehensive platform for researchers and industries.