Abstract: Since visible light or single-band image stitching is mainly used for visual effects and mapping, the stitching of unmanned aerial vehicle(UAV) hyperspectral images has higher requirements, and the spectral information after stitching needs to be used. In order to solve the unique problem of UAV hyperspectral image stitching, a UAV hyperspectral image stitching optimization method based on Shape Preservation and Human Vision (SP-HVS) was proposed, aiming to obtain a complete hyperspectral data cube in the study area. Through the experiments of UAV hyperspectral images in farmland, rivers and forest areas, the results of the proposed algorithm are more in line with the visual characteristics of the human eye. By calculating the spectral angle filling and spectral correlation of the obtained hyperspectral data cube, the spectral fidelity of the proposed algorithm at the sampling points on the splicing line reaches 99.2%. This research provides a novel approach for high-precision UAV hyperspectral image stitching, offering significant practical value.