부산대학교 도서관

Aiming at the problems of low accuracy of image filtering, low production efficiency and high input cost of mobile remote sensing satellite receiving and processing station, this paper proposed a fast remote sensing image segmentation method based on visual attention. By using remote sensing image fast segmentation equipment and software, and through visual attention screening strategy, remote sensing images can be quickly filtered based on artificial vision before advanced product production, so as to obtain slice areas of interest to interpreters, and discard useless or redundant targets, which will greatly reduce the amount of data in the subsequent standard product production process. The analysis results show that the proposed scheme can reduce the amount of data required to be calculated by the target detection model to one 512*512 image slice every 3 seconds, and the single service node resource can be used for near real time processing.