基于工业CT图像处理的材料损伤行为表征分析

发布时间：2018-04-10 23:33

本文选题：工业CT + 图像处理　；参考：《河南科技大学》2017年硕士论文

【摘要】：材料服役过程中产生的损伤是导致产品安全可靠性和服役寿命降低的关键因素。随着我国航空航天、国防军工、轨道交通等重大工程的实施,关键零部件服役条件日益苛刻,安全可靠性要求越来越高,开展材料损伤研究显得尤为重要。全面、准确获取损伤信息是研究损伤的前提,传统物理解剖法和数值模拟法因其盲目性、破坏性、不准确性导致无法全面获得材料内部损伤信息。工业CT因具有直观、无损、高分辨率等优势为材料工作者研究损伤提供了新的思路。然而,目前基于工业CT技术开展材料损伤研究的对象多为实验室条件下单一拉伸、压缩或弯曲加载环境产生的损伤,同时,在CT图像处理方法的探索及损伤行为的有效表征方面不够完善。本文以实际服役条件下产生的典型裂纹损伤为研究对象,通过工业CT系统对其进行360°扫描,获得了含有损伤信息的Top、Front、Right视图方向灰度切片图像。采用Matlab、ImageJ等软件,对各视图方向图像进行了详细、深入的分析,提出了具有针对性的图像处理方法,实现了基于图像处理的损伤行为全方位信息有效表征,并对结果进行了验证。结果表明:Top视图方向灰度切片图像最佳处理方法为:中值滤波(9×9滤波器)+log算子边缘检测;Front视图方向灰度切片图像最佳处理方法为:预处理(对比度增强(0.8)+中值滤波(7×7滤波器))+迭代法图像分割+log算子边缘检测。针对Right视图方向典型灰度切片图像,提出了基于图像各行Gaussian拟合统计分布特征的局部阈值算法,该算法的核心在于:以行为单位,基于Gaussian函数拟合曲线3?准则确定各行代表分割阈值,最后通过对图像每行进行分割而得到整体图像,同时对失真和噪声问题给出了解决方案,处理结果较为理想。将典型图像处理结果与实际情况进行对比,发现处理误差均控制在2%以内。其中,代表性图像Top522中裂纹区面积为9.62 mm2;连续而封闭的裂纹边缘周长为61.30 mm,裂纹区等效直径为29.68 mm;试样区等效直径为30.95mm,试样区边缘圆度为0.9。Right286图像中裂纹末端距试样上表面高度为30.24 mm,距凹槽底端的距离为23.13 mm。裂纹末端尾部距试样左边界为16.24mm;距右边界为14.45 mm。试样总体高度为39.60 mm。本文研究结果实现了工业CT图像的有效处理以及材料内部损伤全方位信息的定量表征,不仅可以弥补物理解剖法的盲目,又可为数值模拟结果的验证提供依据,对重大工程领域实际服役工况下的材料损伤分析具有重要指导意义。
[Abstract]:Materials produced in the process of injury is the key factor leading to reduce the safety and reliability of the product and service life. As China's aerospace, defense industry, the implementation of rail transportation and other major projects, key parts of the increasingly harsh conditions of service, safety and reliability of the increasingly high demand, to carry out the research on material damage is very important. A comprehensive, accurate injury information is the premise of the traditional physical injury, anatomical method and numerical simulation for the blind, destructive, inaccuracy can lead to full access to internal damage information materials. Industrial CT because of its intuitive, nondestructive, provides a new way for studying high resolution damage material. However, based on the current research object material damage to carry out industrial CT technology for the laboratory under a tensile, compression and bending loading environment damage, at the same time, in the CT image The exploration of effective characterization and damage behavior of physical method is not perfect. In this paper, the actual service conditions typical crack damage caused by the industry as the research object, CT system is 360 degrees on the scanning, the damage information containing Top, Front, Right view direction gray slice images. By using Matlab, ImageJ and other software. The image of the view direction of a detailed, in-depth analysis, proposed the image processing method, the damage behavior of image processing based on the characterization of a full range of information effectively, and the results were validated. The results show that the Top view direction gray slice image processing method is the best: the median filter (9 x 9 filter) +log edge detection; Front view direction gray slice image processing method for the optimal pretreatment (contrast enhancement (0.8) + (7 x 7 median filter filter)) + +log iterative method for image segmentation Edge detection. According to the Right view direction of typical gray images, we propose a local threshold algorithm for image feature line Gaussian fitting based on statistic distribution, the core of the algorithm is in units based on curve fitting function Gaussian 3? Criteria determine the row generation threshold by the end of the table, each row of the image segmentation and get the whole at the same time, image distortion and noise solutions were given, satisfactory result. Comparing the typical image processing results with the actual situation, found that the processing error is controlled within 2%. Among them, on behalf of crack image in the Top522 area of 9.62 mm2; continuous and closed crack edge perimeter is 61.30 mm, the crack area equivalent diameter is 29.68 mm; sample area equivalent diameter is 30.95mm, sample edge roundness for crack tip in the 0.9.Right286 image from the specimen surface height is 30 .24 mm, from the bottom of the groove distance is 23.13 mm. at the end of the tail from the left edge of crack for 16.24mm; from the right boundary of 14.45 mm. sample overall height is 39.60 mm. the results of this paper can achieve a quantitative sign of industrial CT image processing and effective material internal damage of a full range of information, not only can make up for the physical anatomy method blind, provide the basis for verification and numerical simulation results, the major project in the field of actual service material damage condition analysis has important guiding significance.

【学位授予单位】：河南科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TB302;TP391.41

【参考文献】