Autonomous Viewpoint Selection based on Aesthetic Evaluation of a Scene with Kullback-Leilber Divergence Minimization

Abstract

In this paper, we propose an optimal viewpoint selection system for monitoring robots to search for the optimal viewpoint of a scene with the highest aesthetic property. We first put forward a novel aesthetic evaluation method by the use of Kullback-Leilber divergence, considering the directional information of each target according to some famous composition rules in the field of photography. Then based on the evalution results, we propose a viewpoint selection method by Kullback-Leilber Divergence Minimization. At last, the effectiveness of our optimal viewpoint selection system is confirmed with experiments.