An airport and oil depot recognition method based on salient semantics model

Different from the conventional ground object recognition based on feature matching technique, an airport and oil depot recognition method is presented in this paper, which is based on salient semantics model. Specifically, the proposed method utilizes the visual attention model to decompose the aerial image into several visual salient subgraphs in low-level feature space, which are the candidate regions object may exist. Meanwhile the training images are applied to construct the bag-of-features (BoF) semantics model via SIFT local features, and the salient semantic features of the subgraphs can be extracted with feature dictionary of BoF model. Consequently the recognition for airport and oil depot can be quickly implemented. Multiple typical ground object database is used to test, which is acquired under different imaging conditions from Google Earth. Experiments on the database demonstrate the proposed method has better recognition performance and higher efficiency compared with traditional feature matching methods, and also more robust to the influence of illumination, viewpoints and scale.