Robust optical flow estimation to enhance behavioral research on ants

Abstract

The ant behavior is an intrinsically fascinating topic. The movement of ants provides a rich source of social behavior for evaluating motion estimation methods. Studies on ants motion usually involve researchers watching video recordings and manually scoring each ant's movements. The traditional observation method is challenging because many ants can interact with each other, interfering with perceiving motion. Pixel-level motion estimation can perform this more robustly and accurately. This paper focuses on optical flow estimation to observe ant movements from an imaging system to perceive the motion through the sequence of images. It is known that the optical flow methods suffer from the issue of ill-defined edges and boundaries of the moving objects. An edge-preserving filter-based optical flow method is constructed to estimate the motion of ants in outdoor complex scenarios. The results reveal that the proposed method can improve the accuracy of motion estimation.