Jitter error detection and compensation method based on H-infinity filter fusion for remote sensing satellite image

Satellite jitter is a common phenomenon in high-resolution satellite applications, leading to image blurring, geometric distortion, difficulties in image stitching, and reduced spatial resolution. Therefore, accurately detecting satellite jitter and compensating for it in images is crucial. In this paper, we propose a novel framework for multispectral satellite jitter detection and compensation, which integrates the results of two different detection methods to enhance detection accuracy and effectively compensate for the images. First, dense matching is performed between the bands of the multispectral imagery to obtain the relative registration error between two bands; Then, dense control point matching is conducted between the single-band image and the corresponding Ortho-Images. Next, a method based on the H-infinity filtering model is proposed to fuse the results of the two detection methods and compensate for the original image. Comparative experimental analysis shows that the compensation results after fusion exhibit significant improvements compared to those obtained using a single detection method.