An Agent-Based Method for Automatic Building Recognition Based on Lidar Data

Lidar data has proved to be a promising data source for various mapping and 3D modeling of buildings. Nevertheless, traditional manually building extraction from Lidar data is highly labor-intensive, time-consuming and very expensive. During the past decade many researchers in photogrammetry, remote sensing and computer vision communities have been trying to study and develop the automatic or semi-automatic approaches for building extraction and reconstruction. Although, several studies related to the building recognition have been published during recent years, the performance of obtained results is still dependent to several assumptions and simplifications. In this research, an agent-based method is proposed for automatic building recognition based on Lidar data. The proposed methodology in this paper has two main steps; first one is pre-processing and second one is the agent-based building recognition. In the pre-processing step, ground, vegetation and near ground objects are removed from the first pulse range Lidar data. The second step of this algorithm is an agent-based method for using the advantages of all Lidar range and intensity data for improving initial candidate of building regions obtained from the previous step. Using an agent as a basic concept of the artificial intelligence can be useful also in the field of building recognition because in the agent-based method, it is possible to define relational knowledge between objects and complete contextual information in such a way to recognize buildings from other objects, precisely. By using agents, one can apply adaptive image processing algorithms on each dataset, based on local contextual information, and then fuse the results in decision level. The results of implementation the proposed methodology, show that using the best pre-defined characteristics of an agent, can solve most of the problems in the field of building detection. Keywords; Building Recognition, Agent-based Method, Contextual Information, Decision Level Fusion