A novel method of aerial image classification based on attention-based local descriptors

Sheng Xu, Tao Fang, Hong Huo, Deren Li
2009 Procedia Earth and Planetary Science  
This paper proposes a novel method for object-based classification in very high spatial resolution aerial image. It combines the saliency maps very closely to extract the conspicuous local regions for better description of object and classification. Unlike the previous work on detection of the local regions, a biologically motivated selective attention model is presented in this paper, since not all the local regions are important for describing the objects. In order to model the attention
more » ... n, we propose a new attention-based local descriptor using the saliency map and relative local features to reflect the region of interest (ROI). The experimental results on the VHR aerial image dataset show that the proposed approach can obtain the state-of-the-art classification performance.
doi:10.1016/j.proeps.2009.09.174 fatcat:kjzmssb54rgtnbqm6q2vh4ltpu