A Redundant Data Elimination Method Based on Change Detection

In recent years, with the rapid development of aerospace technology, the variety of satellite payloads has increased, and the volume of remote sensing data has become increasingly massive. These changes have also put forward higher requirements for the functionality and performance of storage systems. Starting from the perspective of the content validity of image data, this paper proposes a redundant data elimination method based on change detection. By using a twin fully convolutional neural network to implement change detection for remote sensing images of the same scene at different times, and based on the change detection results, adopting certain strategies to eliminate unchanged areas, the method maximizes the elimination of redundant information in remote sensing data without losing effective information, thereby reducing the volume of stored data. The experimental results have verified the effectiveness of the proposed method.