Aerial Image Series Quality Assessment

With the growing demand for geospatial data, the aerial imagery with high spatial, spectral, and temporal resolution achieves great development. It is imperative to evaluate whether the acquired images are qualified enough, since the further image mosaic asks for strict time consistency and a re-flight involves considerable resources. In this paper, we address the problem of quick aerial image series quality assessment. An image series quality analysis system is proposed, which includes single image quality assessment, image series quality assessment based on the image matching, and offering a visual matching result in real time for human validation when the computer achieves dubious results. For two images, the affine matrix is different for different parts of images, especially for images of wide field. Therefore we calculate transfer matrixes by using even-distributed control points from different image parts with the RANSAC technology, and use the image rotation angle for image mosaic for human validation. Extensive experiments conducted on aerial images show that the proposed method can obtain similar results with experts.