Indoor Scene Recognition Based on the Weighting Spatial Information Fusion
Zhiliang Wang, Rong Wang, Xirong Ma · 2012
Scene recognition is high level image understanding in machine vision field. It is usually difficult to recognize variety indoor scene of which the characteristics between many classes are changing significantly. This paper proposes an indoor home scene recognition model based on Weighting Spatial Information Fusion of PLSA (WSIF_PLSA), build many PLSA models and SVM classifiers using spatial information of indoor scene and fuse the recognition result with weight. So it is considering not only global visual characteristic but also local visual characteristic. The experiment constructs a database of IHSD which is for indoor home scene recognition and the results shows the higher recognition efficiency with this method.