Dimensionality reduction-based building recognition

Object recognition is being paid more and more attention in computer vision research and a variety of algorithms have been put forward to enhance the recognition performance. However, building recognition, a relatively specific recognition task, is still at a preliminary stage of development, because the challenging task includes rotation, scaling, illumination changes, occlusion, etc. A building recognition scheme is proposed in this paper, which integrates biologically-inspired feature extraction and dimensionality reduction. Experiments undertaken on our own constructed building database demonstrate that our proposed scheme can achieve satisfactory results.