In-situ indoor 3-D land mapping and radioactive source localization

Recent years, RGB-D cameras have been widely used in many three-dimensional (3-D) applications, such as mapping, segmentation, recognition and other important tasks. These cameras enable machines perceiving the world in 3-D space and translate these perceptions into a synchronized depth image in the same way that humans do. Likewise, radiation detectors enable detecting harmful radiation in a dangerous environment. In the research, the information from RGB-D cameras and radiation detectors can be reconstructed and integrated to establish a radiation-localized 3-D map for effectively exploring hazardous environment and damage control. In this approach, color images are first employed to extract sparse visual features, hence pairs of matched features between two consecutive images are established. To guarantee the convergence of the data registration processing, a novel filtering algorithm is developed to reject invalid correspondences. Meanwhile, an algorithm for localizing the radioactive parameters is also developed to pinpoint radiation sources in the space in the reconstructed 3-D land maps. Some preliminary experiments have been effectively conducted to verify the feasibility of the development.