Lidar point clouds are powerful for creating high-resolution, 3D views of environments. However, one drawback is that they lack color information, making object identification challenging. To tackle this, lidar points can be colored by mapping them onto synchronized camera images. By projecting points onto images using extrinsic and intrinsic matrices, lidar sensor fusion can create a unified representation. This process involves transforming coordinates and adjusting colors to enhance the final point cloud output. While effective, the colorizing process can be slow; improvements include using numpy matrices for faster conversions and utilizing meshes to address occlusion issues.
https://medium.com/mindkosh/colorize-lidar-point-clouds-with-camera-images-4af69cb3efea