Depth and normal vector identification of an unknown slope from a UAV using a single camera
Zhichao Liu, Jianliang Wang, Poh Eng Kee, Suresh Sundaram · 2013
This paper presents a novel vision-based system to estimate the normal vector of an unknown slope and the range from a camera fixed on a UAV to the slope using a single camera. An exact point-based image moments model considering the camera's focal length is presented. Using the model, a fast estimator is designed to estimate the image flow with high precision. The continuous model is then discretized using Taylor series method. Finally, a particle filter is used to obtain a solution to the estimation problem. The whole system estimates simultaneously the normal vector of the unknown slope and the depth from the camera on the UAV to the slope.