A 6000 Hz Computer Vision System for Real-Time Wing Beat Analysis of Drosophila

Abstract

This paper reports the design and application of a novel high speed computer vision system for real-time analysis of the wing kinematics of tethered flying fruit flies (D. melanogaster). The system uses a camera with dynamic regions of interest (ROI) to increase temporal resolution from localized sampling, without loss of spatial resolution. An extended Kalman filter is employed to fit an a priori kinematic model to past wing position measurements, allowing the position of the next ROI to be predicted and providing a real-time readout of kinematic data. Using this approach, we sampled the wing position at 6250 Hz with a precision of 1deg, using a ROI of approximately 3600 pixels. This is more than four times faster than other computer vision based tracking system to date. Beyond the study of insect flight control, this paper demonstrates a novel approach to track complex and fast moving structures in real time applications, a challenge often faced in micro and nano technologies