微小运动放大及其在水电工程监测中的应用

发布时间：2018-06-28 14:13

本文选题：欧拉视频放大 + 空域分解　；参考：《三峡大学》2016年硕士论文

【摘要】：微小运动放大技术是一种放大影像中肉眼不可见或难以察觉的微弱变化的技术,该技术在很多领域都有着潜在的应用价值。欧拉视频放大技术是近年新提出的、非常有吸引力的微小运动放大技术,其核心包括空域分解、时域带通滤波、运动放大以及重建等,现有的欧拉放大技术有线性欧拉视频放大和相位欧拉视频放大技术。本文对现有的欧拉视频放大技术进行了深入的研究,并对其中的关键技术进行分析。针对现有的欧拉视频放大方法计算量大,需要花费大量的时间对视频进行处理的问题;以及现有的欧拉视频放大方法对视频图像序列中的每一个像素点进行处理,放大后视频的噪声会随着放大倍数的变大而增加,加上不能直观地在视频上选取感兴趣区域的问题,本文提出了改进方案;最后将改进的方法应用于对实际的水电工程监测。具体研究内容包括以下几方面:1)本文首先对颜色空间的处理以及多分辨率分解方面进行了探讨,提出了基于YIQ颜色空间中的Y通道进行运动放大处理的加速方法,从而减小计算量以达到提高算法速度的目的。实验结果表明,该加速方法使得视频的放大处理速度有了显著性的改善。2)针对视频的噪声会随着放大倍数的变大而增加,并且不能直观地在视频中选取感兴趣区域的问题,本文提出通过前景检测的方法将视频包含运动信息的区域保留出来,再进行形态学滤波,最后进行运动放大处理与融合。对于运动区域的前景约束,本文比较了几种前景检测算法,最后选用检测性能相对较好、并且与欧拉视频放大方法同样基于像素点的Vibe算法作为本文前景约束的方法。实验表明,对于运动放大结果相对原图像的峰值信噪比PSNR值,基于运动区域前景约束的方法比原始的欧拉视频放大方法的要大,因此保真效果相对较好,并且改进的方法能够针对性地对运动区域进行放大处理,不但提高了运动放大速度,而且使得被监测目标更易凸显出运动特征,以便于人们直接进行观测。3)探讨了将欧拉视频放大技术引入对水电工程项目的监测,为水电工程的智能监测提供了一个全新的研究视角。首先,给出了监测系统的组成框架,并进行了简要的分析。其次,分析了数据采集的难度及其重要性,并对数据采集以及处理进行了简要的分析。最后,将采集的数据进行试验分析,分别探讨了对普通桥梁、含拉索的大型桥梁受横风影响或是车辆通过时的情况以及大坝建设和维护时监测的结果,此外,对隔河岩大坝视频进行试验分析,并对试验结果进行了一定的探讨。试验表明,通过欧拉视频放大处理以后,可以明显地观察到桥梁、桥梁拉索、起重机以及大坝的运动情况,这对于辅助判断出其形变特性,以便采取进一步的措施进行预警处理以及危险排除是非常有价值的,同时也为监测目标的安全检测提供了全新的参考依据。综上所述,本论文系统地研究了欧拉视频放大技术,分别提出了基于Y通道的加速方法和基于运动区域前景约束的运动放大方法。最后将欧拉视频放大技术引入对水电工程项目的监测,并进行了相应的试验分析及讨论。
[Abstract]:The technology of micro motion amplification is a technique which is not visible to the naked eye in the enlarged image. This technique has potential applications in many fields. Euler video amplification is a new and attractive micro motion amplification technology proposed in recent years. Its core includes space domain decomposition and time-domain bandpass filtering. The existing Euler amplification technologies are linear Euler video amplification and phase Euler video amplification. The existing Euler video amplification technology is studied in this paper, and the key technologies are analyzed in this paper. The existing Euler video amplification methods need to spend a large amount of time. The problem of video processing, and the existing Euler video amplification method to deal with each pixel in the video sequence, the noise of the video will increase with the enlargement of the magnification, and the problem of selecting the region of interest can not be intuitively selected in the video. The improvement scheme is proposed in this paper. Finally, the modified scheme will be modified. The proposed method is applied to the actual monitoring of the hydropower project. The specific research contents include the following aspects: 1) first, the paper discusses the processing of color space and the multiresolution decomposition, and puts forward the speed adding method based on the Y channel in the YIQ color space for motion amplification, thus reducing the amount of calculation to improve the calculation. The experimental results show that the acceleration method makes the speed of the video magnification improved.2.) the noise of the video will increase with the enlargement of the magnification, and can not directly select the region of interest in the video. This paper proposes that the video include the motion by the method of foreground detection. The region of the information is preserved, then the morphological filtering is carried out, and then the motion magnification processing and fusion are carried out. For the foreground constraint of the motion area, several foreground detection algorithms are compared. Finally, the detection performance is relatively better, and the Vibe algorithm based on the Euler video amplification method is based on the pixel point as the foreground constraint. The experiment shows that, for the peak signal to noise ratio (PSNR) value of the motion amplification result relative to the original image, the method based on the motion region foreground constraint is larger than the original Euler video amplification method, so the fidelity effect is relatively good, and the improved method can amplify the moving area and improve the movement. Large speed, and make the monitored target more easy to highlight the motion characteristics, in order to facilitate people to observe the.3 directly) to explore the introduction of Euler video amplification technology to the monitoring of hydropower projects, provide a new research perspective for the intelligent monitoring of hydropower projects. First, the framework of the monitoring system is given and the system is carried out. Secondly, the difficulty and importance of data acquisition are analyzed, and the data collection and processing are briefly analyzed. Finally, the collected data are tested and analyzed to discuss the effect of the cross wind on the common bridge, the large bridge with the cable, when the vehicle passes through, and the construction and maintenance of the dam. In addition, the video of the Geheyan dam is tested and analyzed, and the test results are discussed. The experiment shows that the movement of bridges, bridge cables, cranes and dams can be observed obviously after Euler video amplification, which can help judge its deformation characteristics so as to take the first step. The steps of early warning and danger removal are of great value. At the same time, it also provides a new reference for monitoring the safety of monitoring targets. In this paper, the Euler video amplification technology is systematically studied in this paper, and the acceleration method based on the Y channel and the movement generosity based on the motion region foreground constraints are proposed. Finally, Euler video amplification technology is introduced into the monitoring of hydropower projects, and corresponding experimental analysis and discussion are carried out.
【学位授予单位】：三峡大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TV698.1;TP391.41

【相似文献】