Change Object Detection for Remote Sensing Image Based on Top-view Similarity
In this paper,a change object detection method based on top-view similarity for optical remote sensing im-ages is proposed.After registering the two optical remote sensing images taken in different periods in the same area,the object detection is carried out respectively to extract all the target areas of interest.The residual network is used to extract the top-view features of objects,which improves the robustness of the features.The change of objects is measured by calculating the top-view similarity between objects,and the distance criterion and similarity criterion are proposed to judge whether the object has changed,so as to realize the change object detection.The experimental results show that the pro-posed method has better performance than the existing methods,and the average F1 can reach 0.837.
optical remote sensing imageobject of interestchanged object detectiontop-view similarity
NETL
NSTL
万方数据
