【作者】 张英杰；

【导师】 张学成；

【作者基本信息】 吉林大学 ， 机械制造及其自动化， 2006， 硕士

【摘要】 针对传统应变测量的不足,研究和探讨运用视频图像处理与分析技术进行应变精密测量问题。文中主要内容和目标是用视频图像处理和分析的方法完成应变精密测量时的实时采集跟踪和亚像素边界提取问题,并主要以金属材料的拉伸试验为对象进行验证。提出了基于视频图像分析和处理的测量方案,对金属材料试件,在不同环境下进行了测量。建立了图像的数学模型,研究了图像的采集、边界提取和定位、边界点的拟合、系统的标定等问题。通过对不同条件下相同试件图像的分析和处理,找出了最佳实验条件,使试件边界更加清晰,测量精度更高。通过对各种算法的比较,得出亚像素(此处为一维灰度矩)算法能较精确地计算出试件的长度,且使用简单,运算速度快;通过对不同滤波方法的比较,找出了用递推平均滤波法和最小二乘直线拟合滤波法,使测量精度得以提高;通过改变摄像机的频率来改变采集速度,得出15fps的采集频率能满足试验要求;同时本文还分析了产生误差和影响精度的因素,并提出有效的解决方案。实验证明,本文提出应变精密测量的理论方法是有效的,可行的。更多还原

【Abstract】 The material science is one of the most important technique courses of theworld development nowadays. It is basic condition of the material sciencetechnique development that the physics function of the material experiment. Strainis the material physics to experiment medium important and basic parameter,usually adopting the mechanical extensometer to clip and hold the specimen,measuring at the time of inflicting the load to specimen. For the rigid material,strain can use the traditional mechanical extensometer to clip and hold thespecimen or the thin extensometer to measure. However, these devices can’t use forsuch as fiber, thin film, foam or soft material of experiment, because their weightand clipping method will affect to the result of experiment and split the point.Under the condition of many, It needs to measure to know the material functionthat super big strain scope keeps go to split. Limited to the route of travel,above-mentioned extensometer account to need to be dismantled before rupture.Under some particular environment condition of experiment, for example the heatcondition, the mechanical extensometer is restricted namely.For overcoming the above-mentioned restriction, adapting the new lately-develop demand of situation, this text study a kind of technique making use of thevideo image to measure object, That is to say the technical method of whole imageand high precision non-contact strain measurement. Develop a high performancenon-contact extensometer with measurement performance. The main work of thistext is to research following several contents.1. Measure principle and system constructionsThe hardware system with video image processing technology used for strainprecise measurement consists of the light system, CCD camera, PCI card, computerand output equipments. Figure 1 show the system concrete construction. Thesoftware system consists of three models that is getting image edge, markingsystem and real time calculating the geometry parameters. Figure 2 show thesystem concrete construction. Work principle is: first make use of CCD camera totake a specimen picture, then transport the picture into the computer by PCI card,finally get the specimen edge by digital image processing technology. Thespecimen edge in picture is proportional to the actual one, and the ratio is aconstant K determined by CCD camera magnified ratio. Calculating the straingeometry parameters and comparing with the actual size.2. Digital image processing and measuringThe basic characteristic of the image is the edge. In image processing, edge isimportant to both image division and measurement. Meanwhile, edge measurementis the key technology in non-contact measurement. Its results have directly impacton precision. This text adopts rough edge detection, together with precise edgedetection to process the image. First, search the rough edge by phasing inLight system Specimen CCD camera PCI card Output equipments Computer Fig1 Hardware of strain measuring systemCapture image Input image Detect edge Mark system Output parameters Measure image Fig2 Software of strain measuring systemdispersion, the precision of location attain pixel level. For attaining the higherprecision in searching the rough edge, we make use of subpixel arithmetic ---onedimension gray moment operators. It breaks the restrict of the CCD camera physicsresolution, and makes the edge location precision attain subpixel level.3. Characteristic measurementIf you want characteristic measurement result of strain, according toε =Δl/ l0, namely obtain the variety quantity of length, you must follow the edgeprocess. We obtain a frame of image first, according to subpixel edge location, youcan get picture elements every edge takes up, then use the system ratio K to get thegeometry value of measured specimen, this is the main measuring method.Initialize the speed of test-machine, according to the variety quantity of specimenlength and the speed, calculate the pixel dispersion of close frame of image.According to this method, make sure the searching area of measured specimen. Inthis way we complete real-time edge location.4. The factors of influencing the measuring precisionIn this text we find the factors of influencing the measuring precision byexperiment. Though measuring precision of the image measurement systemdepends mainly on resolution ratio of the digital camera, we must consider manyfactors which affect the precision in order to improve precision. The factors are asfollows: Error of oneself of CCD, Error of geometrical aberrance of imagingsystem, Noise of imaging system, Noise of light ken, calibration error, arithmeticerror and influence of vibrancy. By analyzing the factors of error, we choose thebest experiment condition. The error decreased to some extent by recursion averagefiltering and least square method.5. ConclusionThe key problem of the video image measuring system is the systematicprecision, Higher precision means better measuring result, higher accuracy. Inorder to improve precision of the system, we do many experiments. Throughcomparing different experiment condition and analyzing experiment result, wechoose the best experiment condition as the follow study condition. In order toimprove precision, we deal with measuring results of many pictures equally andfinalize the results. Order to test the correctness, we choose different lengthspecimen as target object. Experiment results show that the method on video imagemeasurement of strain precision measurement is correct and feasible.更多还原