Real-time Neural Rendering of LiDAR Point Clouds
This addresses the need for high-quality, real-time visualization of LiDAR data in applications like autonomous driving or mapping, though it is incremental as it builds on existing neural rendering techniques.
The paper tackles the problem of rendering photorealistic images from static LiDAR point clouds, which often have artifacts, by proposing an efficient method that uses a U-Net and depth-based filtering to achieve real-time rendering rates and outperform state-of-the-art in speed and quality.
Static LiDAR scanners produce accurate, dense, colored point clouds, but often contain obtrusive artifacts which makes them ill-suited for direct display. We propose an efficient method to render photorealistic images of such scans without any expensive preprocessing or training of a scene-specific model. A naive projection of the point cloud to the output view using 1x1 pixels is fast and retains the available detail, but also results in unintelligible renderings as background points leak in between the foreground pixels. The key insight is that these projections can be transformed into a realistic result using a deep convolutional model in the form of a U-Net, and a depth-based heuristic that prefilters the data. The U-Net also handles LiDAR-specific problems such as missing parts due to occlusion, color inconsistencies and varying point densities. We also describe a method to generate synthetic training data to deal with imperfectly-aligned ground truth images. Our method achieves real-time rendering rates using an off-the-shelf GPU and outperforms the state-of-the-art in both speed and quality.