부산대학교 도서관

The visual quality assessment for aerial cloud-removal image is a challenging task in the field of image quality assessment. Due to the unique nature of cloud removal in aerial images, the full reference assessment criteria cannot be directly applied. In this paper, we present a novel algorithm for assessing the quality of aerial cloud-removal images, which combines manipulation detection technology. This algorithm identifies the processing area of the cloud-removal image and segments it at the pixel level to obtain detection masks. The quantitative assessment of cloud-removal image quality is achieved by comparing the intersection of the detection mask with the actual processing area. The quality ranking experiment demonstrates that the proposed quality assessment method in this paper is more consistent with human subjective judgment compared to other quality algorithms.