In the electric power industry

when using directly acquired hyperspectral point cloud images of transmission lines for modeling

issues such as low accuracy and misalignment distortion in point cloud stitching are prevalent.To address this

this paper proposes a hyperspectral transmission line multi-view point cloud stitching method based on oblique photography and line-surface feature matching.An unmanned aerial vehicle (UAV) platform equipped with a hyperspectral camera is utilized to conduct oblique photography

and hyperspectral transmission line point cloud data is obtained through data processing.The accuracy of the point cloud is improved through parameter setting

preliminary processing

and an IMU pre-integration model.The original point cloud data is preprocessed

and denoising is achieved using statistical filtering and radius filtering.Dimensionality reduction is performed to extract spectral features and construct multi-dimensional feature vectors

and coordinate registration is achieved based on control points.By combining point cloud geometry and hyperspectral features

a point line surface matching model is constructed

which not only enables transmission line fault location but also provides constraints for point cloud stitching.Based on feature matching

the point line surface ICP algorithm and global optimization method are employed to achieve multi-view point cloud stitching and construct a three-dimensional model of the transmission line.Experimental results show that through a joint constraint optimization and filtering strategy combining hyperspectral bands and spatial features of power lines

the layout of the transmission line can be accurately reconstructed

and the model details are well presented.