A multi-feature integrated visual attention model for matching-area suitability analysis in visual navigation

For matching-area suitability analysis of unmanned aerial vehicle (UAV) visual navigation, a multi-feature integrated visual attention model (MFI-VAM) was established by introducing invariant features, based on which the extraction method of suitable matching-area was proposed. Speeded-up robust features (SURF) were added into visual attention model. The conspicuity map of SURF channel is obtained by cross-scale feature maps integration. Based on multi-feature fusion of SURF, color, intensity and orientation, the MFI-VAM model was built. Salient locations in current map were chosen as suitable matching-areas based on this model. Simulation results show that the image registration error of extracted matching-area based on proposed method is small. This paper could provide new ideas and theoretical guidance for UAV autonomous navigation.