Change detection is a highly valuable research branch in the field of computer vision.In order to improve the monito-ring efficiency of production and construction activities along the Changjaing River,a building change detection system along the Chanjiang River was proposed on the basis of image registration,combined with self-attention and feature extraction.It integrated image processing,image registration and change detection,achieving automatic operation across overall processes.The results show that the designed system minimizes the registration accuracy error within two pixels,achieving pixel-level accuracy.Moreover,the system achieves a result of more than 75%in terms of F1 score on the proposed change detection test dataset,demonstrating its high accuracy.The research can be further applied in the shoreline management of large reservoirs,improving the efficiency of governance of reservoirs.
维普
万方数据