引用本文: 刘胜南, 付强, 冯楠, 张春华, 贺威. 面向扑翼飞行机器人的电子稳像算法设计[J]. 工程科学学报.
LIU Shengnan, FU Qiang, FENG Nan, ZHANG Chunhua, HE Wei. Design of an electronic image stabilization algorithm for flapping-wing flying robots[J]. Chinese Journal of Engineering.
 Citation: LIU Shengnan, FU Qiang, FENG Nan, ZHANG Chunhua, HE Wei. Design of an electronic image stabilization algorithm for flapping-wing flying robots[J]. Chinese Journal of Engineering.

## Design of an electronic image stabilization algorithm for flapping-wing flying robots

• 摘要: 在扑翼飞行机器人的飞行过程中，由于其特有的驱动方式，机身存在周期性的俯仰和滚转运动，导致航拍视频出现抖动，影响成像效果的清晰度和稳定性. 为了解决这一问题，本文提出了一种基于ORB和滑动均值滤波的电子稳像算法，用于在线消抖处理. 首先，针对扑翼飞行机器人航拍图像的抖动周期与机翼扑动周期相一致的特性，设计了一种估计算法来根据图像特征估计机翼扑动周期. 这一算法能够更准确地捕捉到抖动的周期性特征，为后续的稳像处理提供了重要参数. 其次，提出了一种与扑动周期相关联的运动滤波算法，能够根据不同飞行工况对滤波参数自适应地进行动态调整. 本文提出的算法优点在于能够根据扑翼飞行机器人实际飞行情况实时调整参数，从而更好地适应不同的飞行工况，进一步提高了稳像效果. 最后，为了验证算法的可行性和稳定性，将视觉成像装置搭载在扑翼飞行机器人上进行了飞行实验. 实验结果表明，相较于常用的电子稳像算法，本文所设计的算法在扑翼飞行机器人中表现出更好的稳像效果. 最后，总结了本文所提出的算法优点，并对未来研究方向做出了展望.

Abstract: During the flight of a flapping-wing flying robot, the unique flapping-wing propulsion mechanism causes periodic pitching and rolling motions of the body. In passing through different airspeeds and altitudes over different terrain, the wings cycle up-and-down through predefined stroke patterns to generate the aerodynamic force required for powered flight. However, this oscillatory flapping motion also causes the aircraft structure to pitch and roll periodically about its center of mass in an oscillatory manner. Consequently, substantial high-frequency jitter shakes aerial video footage captured by the onboard optical sensors. In particular, the rapid shaking that disrupts the image is synchronized with the characteristic wing beat rhythm. This jitter negatively affects the quality and usefulness of the acquired aerial video. Repeated up-and-down pitching and rolling displacements shake aerial footage, greatly reducing its clarity and stability. Without mitigation, the jitter severely affects imaging results, limiting the potential applications of such video. To solve this problem, this paper proposes an electronic image stabilization algorithm based on oriented fast and rotated brief (ORB) and sliding mean filtering for online debounce processing. First, because the jitter period of aerial images of a flapping-wing flying robot is consistent with the wing flapping period, we designed an estimation algorithm to estimate the wing flapping period based on image features. This algorithm enables us to more accurately capture the periodic characteristics of jitter and provides important parameters for subsequent image stabilization processing. Second, we proposed a motion filtering algorithm associated with the flapping period, which can adaptively and dynamically adjust filtering parameters adaptively according to different flight conditions. The algorithm proposed in this paper is advantageous because it can dynamically adjust parameters in real time based on the actual flight conditions of flapping-wing robots, thereby better adapting to different flight conditions and further improving the image stabilization effect. Third, to verify the feasibility and stability of the algorithm, this paper performed a flight experiment by mounting a visual imaging device on a flapping-wing flying robot. Experimental results show that the proposed algorithm shows better image stabilization effects than commonly used electronic image stabilization algorithms in flapping-wing flying robots. Finally, we summarized the advantages of the proposed algorithm and provided an outlook on the future research directions.

/

• 分享
• 用微信扫码二维码

分享至好友和朋友圈