It is known that laser scanners can provide accurate, dense and raw 3D data of large scenarios but, in order to obtain a representative 3D model of the scene, such information needs to be hardly processed. This research field, named automatic construction of building models, is a challenging topic which is has been progressing over the last five years. This paper presents an overview that covers all of our research in this field. Our method follows a sequential 3D data processing strategy composed by multiple algorithms which aim to obtain a 3D basic model on furnished interiors. The approach computes a Boundary Representation Model (B-Rep) of the scene and, with the help of Radio Frequency Identification (RFID) technologies, is able to recognize and precisely place pieces of furniture into the B-Rep model, thus generating a simplified accurate semantic 3D model of the inhabited interior. Extended experimentation of the method has been carried out in real scenarios with laser scanner data yielding encouraging results.
|Publication status||Published - Dec 2014|
|Event||Second International Conference on 3D Vision - Tokyo, Japan|
Duration: 8 Dec 2014 → 11 Dec 2014
|Conference||Second International Conference on 3D Vision|
|Period||8/12/14 → 11/12/14|