View Loss Evaluation and Keyframe Reselection for Active Aerial Visual Reconstruction

In this paper, we propose a keyframe reselection method based on the field-of-view loss evaluation for active visual reconstruction with photography constraints using UAVs. To obtain the pose and position shift of the actual active vision base coordinate to the planned ones, a particle filter optimization method that utilizes the quartered depth histogram is presented. To evaluate the view loss of the captured images relative to the photogrammetric-constraint planning, a set of quantitative criteria is conducted. The keyframe reselection algorithm is improved based on the above works, which aims to keep images that are more in line with the photogrammetric constraints. The experimental results show that the proposed method can effectively enhance the effect of reconstruction.