Point Cloud and Image Data Fusion Methods based on the Least Squares Method

To enhance sensing capabilities in terms of accuracy and efficiency, the fusion of multiple sensors possessing complementary characteristics is considered. Considering the absence of color information in 3D laser point cloud data and the lack of spatial distance information in images, this study proposes a measurement approach enabling the fusion of LiDAR and camera data. The objective is to address the challenge of integrating image features and point cloud features in the spatial domain, thereby achieving real-time visualization of 3D coordinate data corresponding to the point cloud by selecting points on the image. Given the sparsity of the point cloud, interpolation is conducted using the least squares method. The accuracy of the least squares method is compared with that of the nearest point method through mean square error analysis, and the comprehensive experimental results indicate that the least squares method outperforms the nearest point method, yielding a significant improvement in accuracy by 10.288%, when compared to the inverse distance weighted, which shows an enhancement of 11.6%.