图像匹配技术在电力巡线故障检测中的应用研究

发布时间：2018-08-24 20:18

【摘要】：众所周知,随着科技的进步,图像处理越来越受到广大学者的重视。而图像匹配技术作为图像处理及计算机视觉的重要研究方向之一,在立体视觉、变化检测、遥感图像、目标识别与跟踪等各方面都有着广泛的应用。但是传统的匹配方法,如灰度图像匹配及特征图像匹配都有各自的弊端,致使精度和速度得不到统一。因此,本文针对目前图像匹配技术的缺点,提出行之有效的改进方法,并将其推广到电力线路故障检测中。具体研究内容如下:首先,分析人工蜂群算法存在的不足并进行改进。将最优引导及自适应修正率引入到人工蜂群算法中,提出自适应最优引导人工蜂群算法。使算法在引领蜂群朝着最优解方向移动的同时,自适应调节蜜蜂位置的变化程度,从而提高收敛速度。同时,经广泛采用的标准测试函数进行实验验证,均得到较好效果。其次,针对目前图像匹配技术速度与精度不能两全的现状,将灰度匹配算法与SIFT特征匹配算法相结合,提出粗搜索与外延窗细校正的图像精确匹配方法。粗搜索部分利用新提出的自适应最优引导人工蜂群算法代替传统灰度匹配算法的遍历性,以具有统计特性的灰色关联度作为蜂群算法的适应度函数。细校正以外延窗规则切割图像,利用SIFT算法进行匹配的精确校正。实现由粗定位到细校正的匹配方法,既保留粗搜索中蜂群算法与灰色关联度寻优的快速性,又达到外延窗结合SIFT算法性能匹配的准确度。最后,将粗搜索与外延窗细校正的图像精确匹配方法推广到电力巡线故障检测的实际应用中。分析电力线路的常见故障以及红外图像特点,利用粗搜索与外延窗细校正的图像精确匹配方法结合红外图像具有较好抗噪性的特点,对不同角度的电力线路图像进行故障检测。
[Abstract]:As we all know, with the progress of science and technology, image processing has been paid more and more attention by many scholars. As one of the important research directions of image processing and computer vision, image matching technology is widely used in stereo vision, change detection, remote sensing image, target recognition and tracking and so on. However, the traditional matching methods, such as gray image matching and feature image matching, have their own drawbacks, resulting in the inconsistency of accuracy and speed. Therefore, aiming at the shortcomings of the current image matching technology, this paper puts forward an effective improvement method and extends it to power line fault detection. The specific research contents are as follows: firstly, the shortcomings of artificial bee colony algorithm are analyzed and improved. The optimal bootstrap and adaptive correction rate are introduced into the artificial bee colony algorithm, and the adaptive optimal guidance artificial bee colony algorithm is proposed. The algorithm adaptively adjusts the degree of bee position change while leading the honeybee colony to move towards the optimal solution so as to improve the convergence rate. At the same time, the standard test function is widely used for experimental verification, and good results are obtained. Secondly, aiming at the present situation that the speed and precision of image matching technology can not be both perfect, combining gray level matching algorithm with SIFT feature matching algorithm, an image accurate matching method based on rough search and extension window fine correction is proposed. In the rough search part, the new adaptive optimal guided artificial bee colony algorithm is used to replace the ergodicity of the traditional gray level matching algorithm, and the grey correlation degree with statistical characteristics is used as the fitness function of the colony algorithm. The fine correction is based on the epitaxial window rule and the SIFT algorithm is used to correct the matching precision. The matching method from coarse location to fine correction not only preserves the rapidity of bee colony algorithm and grey correlation degree optimization in rough search, but also achieves the accuracy of performance matching between extension window and SIFT algorithm. Finally, the image accurate matching method of rough search and epitaxial window fine correction is extended to the practical application of power line inspection fault detection. The common faults of power lines and the characteristics of infrared images are analyzed. The fault detection of power line images from different angles is carried out by using the accurate matching method of rough search and extension window fine correction combined with the characteristics of better noise resistance of infrared images.
【学位授予单位】：东北石油大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TM755

【参考文献】