基于立体视觉的直升机机载旋翼共锥度测量系统研究

发布时间：2018-06-02 20:13

本文选题：直升机 + 旋翼共锥度　；参考：《南昌航空大学》2017年硕士论文

【摘要】：直升机旋翼共锥度是反映直升机旋翼动平衡特性的重要参数,它直接影响直升机的安全性、稳定性和机动性等关键指标,因此旋翼共锥度的精确测量为直升机旋翼的生产与维护提供重要依据。在基于立体视觉的直升机机载旋翼共锥度测量系统中,采集的旋翼图像往往具有大场景、背景复杂以及含有一定干扰噪声等特点,导致传统标记点检测算法定位慢、精度低,进而影响旋翼共锥度实时测量。针对上述问题,本文对旋翼图像中圆形标记点的快速精准定位方法进行了深入研究,主要工作内容和研究成果如下:(1)概述了基于立体视觉的直升机机载旋翼共锥度测量系统及相关理论知识。首先,介绍了直升机旋翼共锥度的测量原理和双目立体视觉原理;然后,开展了基于立体视觉的直升机机载旋翼共锥度测量系统设计,该系统包括硬件子系统和软件子系统,硬件系统包括:光电传感器模块、双目图像采集模块、PC/104嵌入式系统、供电模块和配置工具模块;软件系统包括:配置软件和旋翼共锥度测量软件;最后,深入分析了该系统中圆形标记点检测算法存在定位慢和精度低的原因,并针对该问题开展了圆形标记点定位方法的研究。(2)提出了一种基于CIC的圆形标记点快速定位算法(简称:CIC算法)。首先,利用Canny算子对旋翼图像进行边缘检测,再采用连通域搜索法提取轮廓信息;然后,依次采用连通域轮廓的圆度特征、惯性率特征和凹凸度特征筛选旋翼图像中圆形标记点,并进行验证;最后,采用最小二乘圆拟合法实现圆心的精确定位。通过大量仿真和真实环境中圆形标记点检测实验,结果表明,本算法能够在复杂背景下精准快速地定位圆形标记点。但是,当旋翼处于高速旋转以及相机焦距不准等因素影响时,造成图像中圆形标记点边缘模糊,导致部分圆形标记点漏检。(3)针对上述部分圆形标记点漏检的现象,改进了一种基于EDCIC的圆形标记点快速定位算法(简称:EDCIC算法)。首先,采用高斯滤波器对图像进行平滑,利用Sobel算子获取梯度等级图和梯度方向图,并利用梯度等级图搜索图像中的锚点;其次,采用Smart Routing算法连接锚点,得到边缘图像;再次,利用Helmholtz感知原理判决边缘线段,去除边缘图像中的杂线;最后,采用CIC算法对边缘图像进行圆形标记点检测。通过大量仿真和真实环境中圆形标记点检测实验,结果表明,本算法与传统圆形标记点检测算法相比,具有较高的检测速度、定位精度,并较好地解决了由于边缘模糊造成圆形标记点漏检的问题,可用于提高复杂背景下直升机旋翼共锥度的测量速度和精度。
[Abstract]:The common taper of helicopter rotor is an important parameter to reflect the dynamic balance characteristics of helicopter rotor, which directly affects the safety, stability and maneuverability of helicopter. Therefore, the accurate measurement of the common taper of rotor provides an important basis for the production and maintenance of helicopter rotor. In the helicopter airborne rotor common taper measurement system based on stereo vision, the captured rotor images often have the characteristics of large scene, complex background and certain interference noise, which leads to the slow positioning and low precision of the traditional marking point detection algorithm. Furthermore, the real time measurement of common taper of rotor is affected. In order to solve the above problems, the fast and accurate positioning method of circular marking points in rotor images is studied in this paper. The main work and research results are as follows: 1) A helicopter airborne rotor common taper measurement system based on stereo vision and related theoretical knowledge are summarized. Firstly, the measurement principle of common taper of helicopter rotor and the principle of binocular stereo vision are introduced. Then, the design of helicopter common taper measurement system based on stereo vision is carried out. The system includes hardware subsystem and software subsystem. The hardware system includes: photoelectric sensor module, binocular image acquisition module PC-104 embedded system, power supply module and configuration tool module; software system includes configuration software and rotor common taper measurement software; finally, In this paper, the reasons of slow location and low precision of circular marking point detection algorithm in this system are analyzed. In order to solve this problem, the research of circular marking point location method is carried out. (2) A rapid circular marking point location algorithm based on CIC is proposed (abbreviated as: CIC algorithm). Firstly, the edge of rotor image is detected by Canny operator, then the contour information is extracted by using the connected domain search method, and then the roundness feature of the connected domain contour is adopted in turn. The inertial rate feature and convexity feature are used to screen and verify the circular marking points in the rotor image. Finally, the least square circle fitting method is used to locate the center of the rotor accurately. Through a large number of simulation and real environment experiments, the results show that the algorithm can accurately and quickly locate circular marking points in complex background. However, when the rotor is under the influence of high speed rotation and the camera focal length is not correct, the edge of the circular marking point in the image is blurred, which leads to the missing detection of the partial circular marking point. This paper presents an improved circular marker point location algorithm based on EDCIC. Firstly, the image is smoothed by Gao Si filter, the gradient grads and gradient patterns are obtained by Sobel operator, and the anchor points in the images are searched by gradient gradients. Secondly, the edge images are obtained by using the Smart Routing algorithm to connect the anchor points. Thirdly, the edge line segment is determined by Helmholtz sensing principle, and the clutter line is removed from the edge image. Finally, CIC algorithm is used to detect the circular marking points of the edge image. Through a large number of experiments in simulation and real environment, the results show that the algorithm has higher detection speed and positioning accuracy than the traditional circular marker detection algorithm. It can be used to improve the measuring speed and precision of helicopter rotor taper in complex background.
【学位授予单位】：南昌航空大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：V275.1;TP391.41

【参考文献】