摘要 |
Disclosed is a remote sensing image salient object change detection method, comprising: extracting a salient object area on a reference image; sampling the reference image and an input image, and using sampling points to approximately represent the salient object area; extracting a DAISY feature of the sampling points; looking for multiple candidate matching points in a sampling point set of the input image for sampling points of the reference image; searching for an optimal matching point in a corresponding candidate matching point set for a sampling point set in the salient object area, and using a distance between optimal matching point sets as a change feature corresponding to the salient object area; and determining whether an area corresponding to the salient object area in the input image is changed. The present invention filters a large amount of redundant information, and improves practicability of remote sensing image change detection, an area description capability, robustness to a view angle change and a registration error, and inter-class separability between a change class and a non-change class. The present invention can be widely used in many fields such as disaster monitoring, object reconnaissance and the like. |