Lidar Scene Reconstruction Using Imagery and Frame Overlap from a Texel Camera

A small unmanned aerial vehicle (sUAV) can be used to reconstruct a large area of terrain by capturing 3D scans, consisting of LiDAR point clouds and an overlaid image, and combining the scans together. Forming a complete 3D scene using texel image data (fused LiDAR and digital image scans) from an entire flight can be computationally prohibitive on low-cost hardware, so combining registering scans is done using a streaming method which processes the scans in consecutive chunks. Depending on the flight pattern, matching points in the scene may be visible from scans which were not captured around the same time, so to improve the accuracy of the final reconstruction, reconstructing a complete scene must take into overlapping scans from any time in the flight. A modification is described which finds overlap in scans which are not adjacent in time to correct for error. This algorithm also addresses the ”loop-closing” problem, which occurs when the sensor returns to the starting point of a survey. In this work, flight data gathered from an sUAV with low-cost, commercial off-the-shelf sensors is used to demonstrate how 3D scene reconstruction using this algorithm can correct for errors. Examples of the resulting TDSMs are presented. The presented research can be applied to a variety of applications utilizing Texel cameras for terrain reconstruction, such as using multiple sUAVs capturing the same area of terrain simultaneously, and allowing for accurate reconstruction without requiring a strictly defined flight path to capture Texel images consecutively.