Phase-based visual vibration detection against UAV shaking

Journal Information

J. Hefei Univ. Tech. (Nat. Sci.), Feb.2023, Vol.46 No.2: 180-188

DOI: 10.3969/j.issn.1003-5060.2023.02.007

Authors

ZHOU Yong $ ^{1,2} $, YANG Xuezhi $ ^{2,3} $, ZANG Zongdi $ ^{1,2} $, WANG Jincheng $ ^{1} $, WU Kewei $ ^{1,2} $

(1. School of Computer Science and Information Engineering, Hefei University of Technology, Hefei 230601, China; 2. Anhui Province Key Laboratory of Industry Safety and Emergency Technology, Hefei 230601, China; 3. School of Software, Hefei University of Technology, Hefei 230601, China)

Abstract and Keywords

Abstract: Aiming at the problem of inaccurate visual vibration detection of the unmanned aerial vehicle (UAV) platform, a vision-based vibration detection algorithm against the shaking of UAV is proposed to realize the vibration detection under the condition of unstable lens. The algorithm utilizes the locality of target vibration and the globality of the motion of UAV to extract the target vibration from the mixed signals of source video. Firstly, the vibration region of the target is separated from the background region according to the variation of the pixel brightness. Then the feature points are extracted from the background region and the motion matrix of it is solved. After that, the target region is corrected to remove the signal component of background, which is essentially the motion of UAV. Finally, the local phase sequence of the target vibration region is extracted and the vibration signal is constructed, and then the vibration frequency is detected. In order to verify the accuracy of the algorithm, a UAV is used to shoot multiple sets of vibration videos under different shaking amplitudes and detect their frequencies. Experimental results show that the proposed algorithm has high accuracy and robustness, and can effectively remove the shaking interference of UAV and accurately extract target vibration signals from video.

Keywords: vision-based measurement; vibration detection; unmanned aerial vehicle (UAV); steerable pyramid

作者信息

周勇 $ ^{1,2} $，杨学志 $ ^{2,3} $，臧宗迪 $ ^{1,2} $，王金诚 $ ^{1} $，吴克伟 $ ^{1,2} $

（1. 合肥工业大学计算机与信息学院，安徽合肥 230601；2. 工业安全与应急技术安徽省重点实验室，安徽合肥 230601；3. 合肥工业大学软件学院，安徽合肥 230601）

摘要和关键词

摘要: 针对无人机平台视觉振动检测不准确的问题,文章提出一种抗无人机晃动的视觉振动检测算法,用于实现无人机拍摄镜头不稳定情况下的振动检测。该算法利用目标振动的局部性和无人机晃动的全局性,从源视频的混合信号中提取出目标振动。首先根据像素点亮度变化的剧烈程度实现目标振动区域与背景区域的分离;其次对图像的背景区域进行特征点提取,并求解背景区域的运动矩阵;然后对目标区域进行运动矫正,去除其包含的背景信号分量,即晃动信号分量;最后提取目标振动区域的局部相位序列,构建振动信号,并检测其振动频率。为了验证该算法的准确性,使用无人机在不同晃动幅度下拍摄多组视频并检测其振动频率。实验结果表明,该算法具有较高的准确性与鲁棒性,能够有效去除无人机晃动干扰,准确提取出视频中的目标振动信号。

关键词: 视觉测量；振动检测；无人机；可控金字塔

Funding

安徽省科技重大专项资助项目（201903C08020010）；安徽高校协同创新资助项目（GXXT-2019-003）

合肥工业大学学报(自然科学版)

Navigation