A block-affine registration and cloud-controlled stitching method for airborne hyperspectral cameras

Airborne hyperspectral cameras with wide spatial coverage play a crucial role in remote sensing applications such as crop growth monitoring, soil fertility assessment, and forest health evaluation, demonstrating substantial practical value and application potential. However, the integration of multiple sensors for hyperspectral imaging poses challenges in achieving accurate hetero-spectral registration, while large-scale production inevitably introduces errors in stitching multiple flight strips. To address these issues, this study proposes a block-affine registration and cloud-controlled stitching method tailored for airborne hyperspectral imagery. For hetero-spectral registration, a block-based affine transformation strategy is employed, incorporating grid partitioning and local feature matching to compensate for local affine variations and mitigate geometric distortions, thereby enabling high-precision alignment. For flight strip stitching, a “cloud control” framework is introduced, which leverages existing geospatial data for feature matching with each strip. Control points are extracted to refine Position and Orientation Systems (POS) parameters, ultimately achieving high-accuracy alignment among flight strips. The proposed method was validated in the Yushu region of Jilin Province. Experimental results show that the localization accuracy is within three ground sample distances (GSD), satisfying the requirements of hyperspectral remote sensing applications and providing technical support for large-scale environmental monitoring.