Towards the actualization of an air bag system for the UAV's crash, this paper proposes a method for estimating the UAV's crash site from the video sequence acquired by the camera attached to the UAV. The crash point can be considered to correspond to the divergence point of the optical flows. In the accumulator, the cells at which the optical flows (straight lines) pass through are incremented by one. After performing this process for all the optical flows, the cell with the largest vote is obtained as the crash point (divergence point) in the image plane. Experiments using a hand held camera show that the accuracy of estimating the crash site is increased as the camera approaches the target plane. Overall, the experimental results are promising.
Yusaku Hatano, Hiroshi Nagahashi, Chen Yi, Jun Ohya, "Estimating the UAVs' Crash Point Based on Optical Flows' Voting in the Image Plane" in Proc. IS&T Int’l. Symp. on Electronic Imaging: Computational Imaging XVI, 2018, pp 402-1 - 4026, https://doi.org/10.2352/ISSN.2470-1173.2018.15.COIMG-402