Constructing a point cloud for a large geographic region, such as a state or country, can require multiple years of effort. Often several vendors will be used to acquire LiDAR data, and a single region may be captured by multiple LiDAR scans. A key challenge is maintaining consistency between these scans, which includes point density, number of returns, and intensity. Intensity in particular can be very different between scans, even in areas that are overlapping. The figure below illustrates this dilemna. Harmonizing the intensity between scans can be expensive and time consuming.
In this paper, we propose a novel method for point cloud harmonization based on deep neural networks. We evaluate our method quantitatively and qualitatively using a high quality real world LiDAR dataset. We compare our method to several baselines, including standard interpolation methods as well as histogram matching. We show that our method performs as well as the best baseline in areas with similar intensity distributions, and outperforms all baselines in areas with different intensity distributions.
Below are some visualizations of the evaluation tile we used to quantify our method’s performance. We evaluate our method on two verions of the dataset. One is the default version of DublinCity, while the other has a global shift applied across the x-axis. This simulates a shift in the physical brightness distribution. To measure the method’s performance, corruption is applied to each scan in the dataset, except the target scan. Our goal is to remove this corruption. Ideally, this should look just like the ground truth. For more info, see the paper.
No Shift dataset
With Shift Dataset