无人机航拍图像的输电线识别方法

发布时间：2018-01-02 15:30

本文关键词：无人机航拍图像的输电线识别方法　出处：《广西大学》2016年硕士论文　论文类型：学位论文

【摘要】：输电线路通常分布于环境复杂的环境中,采用无人机进行输电线巡检时,其航拍图像中的输电线通常会受到如建筑、道路等背景直线边缘的干扰,为了可以准确的识别输电线,提出了一套输电线识别方法。本文以PixHawk开源飞控为核心搭建了轴距为650毫米的无人机航拍系统,并可携带航拍设备,可以实现实时图像传输和数据传输。本文提出的输电线识别方法以此无人机采集的图像为研究对象,方法分3步：第一步对航拍图像进行预处理；第二步线增强并通过Hough变换对航拍图像进行直线检测得到直线的参数；第三步根据输电线簇参数的关联性并利用Hough变换对直线的参数点进行正弦曲线检测将输电线从干扰直线中提取出来。对航拍图像的预处理主要是为了解决航拍图像畸变和光照不均的问题。利用非线性畸变模型对图像进行非线性畸变矫正,并利用插值法进行灰度重建,还原图像。采用直方图均衡法对航拍图像进行光学矫正,解决图像光照不均的问题。最终,得到没有畸变且对比度统一的图像。提取直线像素包括边缘检测方法和线增强方法。边缘检测方法主要有：Robert算法、Prewitt算法、Sobel算法、Krisch算法、LoG算法和Canny算法。线增强的方法有：Radio算法和Hessian算法。通过实验对比得出结论—Hessian算法可以更有效的将输电线像素提取出来。采用Hough变换进行直线检测得到直线在极坐标中的参数点。本文提出了一种利用参数关联性进行输电线提取的方法。首先通过输电线所在平面坐标系与图像坐标系之间的关系推导出图像中输电线参数关系,然后利用此关系将输电线从具有干扰直线的背景中提取出来.最后对本文方法进行实验验证,由实验结果可见采用本文方法可以排除背景直线的干扰。证明本文提出的方法可以抵抗干扰并提高输电线识别的鲁棒性。
[Abstract]:The transmission line is usually distributed in the environment of complex environment, using UAV transmission line inspection, the aerial image in transmission lines is affected by such as buildings, roads and other background interference edge line, in order to identify the accurate transmission line, and put forward a set of transmission line identification method. In this paper, the UAV aerial system PixHawk flight control source as the core to build a wheelbase of 650 mm, and can carry aerial equipment, can achieve real-time image transmission and data transmission. The transmission wire recognition method is proposed in this paper to no image acquisition machine as research object, the method consists of 3 steps: the first step of aerial image preprocessing; second step line enhancement the aerial image line detection to obtain the linear parameters by Hough transform; the third step according to the relevance of the transmission line parameters and cluster using Hough transform for line parameters are Chord curve detection will be extracted from the transmission line. The line interference pretreatment on aerial image is mainly to solve the aerial image distortion and the problem of uneven illumination. The image nonlinear distortion correction using nonlinear distortion model, and interpolation method using gray reconstruction, to restore the image. The aerial image of optical correction using histogram balancing method, solve the problem of uneven illumination image. Finally, no image distortion and contrast. Uniform line extraction including pixel edge detection method and line enhancement method. Edge detection methods mainly include: Robert algorithm, Prewitt algorithm, Sobel algorithm, Krisch algorithm, LoG algorithm and Canny algorithm. The method of line enhancement A: Radio algorithm and Hessian algorithm. By comparing the conclusion Hessian algorithm can more effectively the transmission line pixel is extracted. By using Hough transform for straight The straight line detection parameters are obtained in polar coordinates. This paper presents a method for extraction of transmission lines using a parameter correlation. The relationship between the first transmission line plane coordinates and the image coordinates of the push wire transport parameters derived in the image, and then use this relationship will be extracted from the transmission line with interference the line in the background. Finally, experiments are carried out to validate this method, interference according to the results by this method can eliminate the background line. Results showed that the proposed method can resist the interference and improve the robustness of transmission line identification.

【学位授予单位】：广西大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TM75

【相似文献】