一种基于时空MRF的工业CT图像序列分割方法

提出了一种基于时空马尔科夫随机场的工业CT图像序列分割算法。此算法根据工业CT图像序列信息连续性的特点,建立时空Markov随机场,并且构造相应的混合高斯统计模型能量函数,利用条件迭代算法（ICM）实现最大后验概率（MAP）估计。仿真试验表明,该方法能够较好地实现工业CT图像序列的分割。     

边缘检测在工业CT图像中的应用

提出一种基于边缘提取的工业CT图像几何尺寸的自动测量方法。采用一种快速全自动的图像分割算法来提取边缘,并在此边缘上实现对工件CT图像的内部结构尺寸和缺陷尺寸的自动测量。该方法已成功应用于工业CT图像处理软件中,并具有较高的精确度和重复性。     

精密零件的工业CT图像测量方法

利用工业CT实现精密零件内部尺寸的高精度测量已逐渐成为产品质量控制的重要技术手段,而CT图像分割和数据拟合是工业CT图像测量的重要环节和难点问题。基于此,采用联合小波变换和RSF模型的图像分割方法对精密零件CT图像轮廓进行精确分割,得到感兴趣轮廓数据,接着采用最小二乘拟合方法得到圆和直线方程,进而测量得到感兴趣对象的直径和角度;同时,对于零件壁厚的测量,采用最小距离搜索法获取壁厚值。多组精密零件的CT图像测量结果表明,该方法测量精度总体优于通用测量软件VG Studio Max的精度,具有一定工程应用价值。     

电子束焊缝超声波C扫描与工业CT检测方法测试结果比较

以传动齿轮中电子束焊缝为对象,采用超声波C扫描检测以及高能加速器工业CT检测两种无损检测方法对其进行检测。结果表明,超声波C扫描检测技术对熔深、焊偏、未熔及延迟裂纹等缺陷检出率和灵敏度高,但检测结果易受工件结构、缺陷形状影响,缺陷尺寸由超声波探头声束直径和图像边界分割算法决定;高能加速器工业CT检测系统对小于0.5mm以下缺陷的定量精度优于超声波C扫描检测,不受工件结构及缺陷形状特征的限制,分辨率高,检测体积型缺陷优势大。     

基于DSnet网络的零件边缘轮廓提取北大核心

针对传统边缘轮廓提取算法极易受到光照强度等因素的影响,出现噪声点和零件边缘轮廓点缺失问题,提出一种基于DSnet分割网络及最外围约束算法结合的零件边缘轮廓提取算法。首先,利用DSnet分割网络对零件原始图像进行分割操作,避免影响因素干扰的同时,获得分割边界明显的零件分割图像,再根据零件分割边界像素点的特点,采用最外围约束算法进行边缘轮廓提取,得到零件边缘轮廓图像。实验结果表明:提出的方法能在光照强度不均匀等因素的干扰下准确提取出完整的、单像素的零件边缘轮廓,而且提取精度的各项指标,RMSE达到了11.18,PSNR达到了27.76,SSIM达到了0.989 3,满足边缘轮廓提取精度的要求。     

基于二元函数拟合的X射线焊缝图像缺陷分割方法

由于存在焊缝图像噪声强、不清晰、对比度低的问题,导致图像分割效果差,文中提出一种基于二元函数拟合的X射线焊缝图像缺陷分割方法。通过正弦变换函数对原始焊缝图像增强处理,使用B样条曲线拟合图像内的灰度曲线,计算高斯曲率与平均曲率得到焊缝表面图像边缘特征,通过二元函数得到不同类型的焊缝边缘数据,结合焊缝图像的表决图,完成对焊缝图像缺陷完美分割。试验结果表明,该方法分割精度高,且在缺陷类别识别和检测效果图上都要高于卷积神经网络算法、目标检测算法、多视觉成像算法的,证明所提方法分割效果好,有实际的应用价值。     

小波多尺度C—V模型的红外图像分割

由于红外图像大多具有目标边缘模糊、对比度低、噪声多的特点,传统的C-V模型在进行分割时,易产生过分割且计算量大.针对这一问题,本文提出了一种兼顾目标同质性信息的小波多尺度C-V模型红外图像分割方法,该方法根据图像目标和背景同质性信息所占面积比例关系自适应地调节模型的参数.首先对红外图像做小波分解,从分解的顶层低频图像开始,利用改进能量函数,寻找目标边界,通过插值由粗尺度到细尺度逐层演化,最终获得原始图像的分割.该方法与原有方法相比,大大减少数据的计算量,有效地提高了分割速度,改善了分割效果.仿真实验证明了算法的有效性.     

基于工业CT图像的零件尺寸测量方法

为了获得高精度工业CT尺寸测量结果，针对工业结构中常见的圆柱体提出了一种亚体素CT图像尺寸测量方法。首先使用MarchingCubes方法检测边缘点和法线，然后顶点沿着法线方向移动到梯度最大值处，最后采用点云分割方法获得圆柱体直径。在仿真数据中验证了该方法的测量精度优于全局表面测定方法的测量精度，具有一定的工程应用价值。     

基于CT切片图像的虚拟测量

工业CT作为一种先进的无损检测手段,可以有效地检测零件的内外部结构情况,对扫描得到的工业CT图像进行二值化,将图像目标和背景分离,然后进行轮廓跟踪,识别出图像的外轮廓和内轮廓,并将内外轮廓数据拟合成样条曲线,最后进行虚拟测量。结果表明,该方法可以测量内外轮廓沿x,y方向以及任意方向的尺寸。     

一类窄焊缝的结构光图像特征提取方法

提出一种新颖的窄焊缝的特征提取与图像处理方法,有效地克服了环境的各种干扰,可靠地提取出了焊缝的特征点.首先设计了一套基于线结构光的显微视觉系统,可靠地检测出了结构光焊缝图像;接着提出一种基于阈值变换的行像素点统计算法和自适应形态学收缩算法可靠地分割出结构光条纹图像;然后采用过度参数直线拟合和B样条曲线拟合算法准确地得到了结构光条纹的中心线.最后,提出一种基于距离搜索的算法,得到了焊缝的特征点.结果表明,窄焊缝结构光图像特征提取方法可行.     

ICT在弹药检测中的应用

探讨了采用ICT技术对弹药进行无损检测的可行性，介绍了ICT系统的组成，重点讨论了提高ICT成像质量所采取的措施，给出了弹药中缺陷的CT扫描图像。结果表明ICT技术具有其它常规检测方法无法比拟的优越性，对弹药有很好的检测效果。     

Flaw detection in radiographic weld images using morphological approach

It is necessary to detect suspected defect regions in the radiographic weld images to find the type of flaw and its causative factors. This requires processing of radiographic images by a suitable approach. This paper presents an approach to process these radiographic weld images of the weld specimens considering morphological aspects of the image. The proposed approach first determines the flaw boundaries by applying the Canny operator after choosing an appropriate threshold value. The boundaries are then fixed using a morphological image processing approach i.e. dilating few similar boundaries and eroding some irrelevant boundaries decided on the basis of pixel characteristics. The flaws detected by this approach are categorized according to their properties.     

微间隙焊缝磁光成像小波多尺度融合检测

针对微间隙(小于0.1 mm)对接焊缝,通过对焊件施加感应磁场,并利用法拉第旋光原理构成的磁光传感器,获取焊缝磁光图像.为了获取焊缝的准确位置,研究一种焊缝磁光图像的小波多尺度信息融合边缘检测方法.对焊缝磁光图像进行3层小波分解,获得包含焊缝边缘信息的小波高频图像.根据各尺度高频信息包含的细节丰富度,融合各尺度高频图像,然后用小波模局部极大值对融合图像进行边缘检测,得到抗噪性和连续性好、定位精度高的焊缝边缘,最后进行焊缝跟踪试验.结果表明,磁光图像小波多尺度信息融合是一种有效的焊缝边缘提取方法,适用于磁光成像传感的微间隙焊缝跟踪图像处理过程.     

Infrared thermal image segmentations employing the multilayer level set method for non-destructive evaluation of layered structures

This paper describes the multilayer level set method for identification of surface defects within a material. This method relies upon the examination of temperature variations within the material. Even though several image processing techniques have used thermal images for detection of surface defects. Detecting and locating surface defects from thermal images is difficult. Mumford and Shah proposed to divide an image in a set of homogeneous sub-regions such that the energy contained in the image can be minimized. Based on this minimization of the energy, the multilayer level set method implicitly presents the regional boundaries as several nested level lines. By increasing iterations and pre-selected level values, these lines evolve close to the level boundaries based on the energy minimization. In this paper, two kinds of tests are employed to evaluate the performance of the algorithm: the first is that the artificial defects are buried behind the tiles; the second that the artificial defectors are buried behind and near the surface of a structure covered with carbon fiber reinforced plastic (CFRP). With a set of halogen lights used to heat the structure, a thermal camera with temperature resolution 0.1 °C is employed to record the temperature changes. The experimental results show that, according to the predefined level values, the multilayer level set method can successfully detect regional boundaries of the buried defects by identifying temperature changes within their neighborhoods using infrared thermal images.     

高温合金的晶界提取及特征定量表征方法

晶粒尺寸在金属的拉伸强度、韧性等力学性能方面起着决定性作用，为了定量评估金属中的晶粒，必须在金相图像中确定晶界。针对高温合金高倍显微图像中晶界模糊或断裂、析出相、抛光颗粒杂质干扰等现象，提出一种晶界提取的方法。通过拉普拉斯锐化和直方图均衡化等处理来增强图像对比度，再用Otsu法进行分割，将得到的晶界图求最大连通域去除独立的点，最后作平滑细化操作得到完整的晶界并计算出晶粒的特征参数。对GH4149、GH706、GH738金相图片进行处理，结果表明，该算法效果接近传统软件提取加人为后续加工的效果，省去了重复工作，具有一定的可靠性。     

Characterisation of pitting corrosion by white light interferometry

An improved method is described for obtaining detailed information on the size and shape of many corrosion pits by examination with White Light Interferometry. The entire surface is first imaged at low resolution by automatically stitching individual images into one composite image. Specially written software analyses the image and determines the location of all corrosion pits. The program then guides the white light interferometer in making high resolution topographic images of each pit. The images are analyzed to give any desired data about the pit geometry, such as depth, maximum width, projected area, volume, etc. Curves showing the depth distribution for each pit can be generated along with detailed topographic images of all pits. The technique does not work for pits that undercut the surface, and there is a practical size limit of about 20 cm². The method is therefore best suited for studying the early stages of pit propagation.     

Self-similar growth of multiple compound layers in binary diffusion couples with application to the “multi-foil” method

The diffusion-controlled growth of multiple compound phases is studied with the nonlinear Kirkendall effect included. Previous work [1] has analyzed the growth of one compound layer. In that analysis, the nonlinear, time-dependent diffusion equation with two free boundaries is reduced by a self-similar transformation into an ordinary differential equation, which is then solved numerically by a shooting method. This work extends that analysis to arbitrary N layers. The method of extracting intrinsic diffusion coefficients from only the positions of interfaces is expanded to two layers. In addition, the asymptotic solution valid for small concentration gradients is applied to the “multi-foil” method of measuring intrinsic diffusion coefficients and yields an analytic expression for the displacement curve. It is found that two vertex positions of the triangular displacement curve are sufficient to calculate the intrinsic diffusion coefficients.     

基于聚焦合成的砂轮表面三维重构方法

提出一种基于聚焦合成的重构砂轮三维地貌的新方法。利用显微镜物镜焦深范围有限,观察砂轮样本时不能聚焦清晰图像的特点,先采集同一区域的一系列聚焦在不同深度的图像,用图像聚焦评价函数进行图像叠合、比较,得到图像中不同位置的深度信息,最后通过高斯插值法得到砂轮表面的三维地貌。并根据所得到的三维地貌数据,对砂轮磨刃高度分布进行了计算。