Image and Sparse Laser Fusion for Dense Scene Reconstruction

This paper is concerned with reconstructing the metric geometry of a scene imaged with a single camera and a scanning laser. Our aim is to assign each image pixel with a range value using both image appearance and sparse laser data. We pose the problem as an optimisation of a cost function encapsulating a spatially varying smoothness cost and measurement compatibility. In particular we introduce a second order smoothness term. We derive cues for discontinuities in range from changes in image appearance and reflect this in the objective function. We show that our formulation distills down to solving a large linear system which can be solved swiftly using direct methods. Results are presented and analysed using synthetic cases to demonstrate salient behaviours and on real data to highlight real-world applicability.