计算机图像处理技术应用于晶振元件缺陷检测

为了对工业现场中晶振元件的缺陷进行检测,本文采用一种基于计算机图像处理的方法,由传统目测方法得到缺陷晶振的一般特征,然后对计算机采集到的图像按照缺陷分类,对每一种缺陷进行图像预处理和分析,在图像的分割中采用最优阈值统计算法,实验结果表明该算法能够达到预期的识别目的.     

光学薄膜表面微细缺陷在线检测方法研究

为了对光学薄膜进行精密光学检测,需要采用数个具有高解析度的线阵CCD相机进行光学薄膜缺陷图像的同步采集。基于计算机视觉的理论,提出了一种有效的自动质量监测方案,实现对光学薄膜表面微细缺陷项目的检测,并根据检测结果驱动打标机对光学薄膜缺陷进行自动标识和定位,该方法具有高精度、实时、在线和非接触的显著优势。探索性实验结果表明,采用该方法可获得光学薄膜表面缺陷的清晰图像,缺陷的细节能够得到很好的展现,并且能够完成缺陷特征的提取与识别。     

一种新的轮胎激光散斑检测缺陷识别方法

在轮胎缺陷识别系统中,提出将轮胎ESPI缺陷图像灰度纹理特征和一种改进的Hu不变矩特征的组合作为特征因子,结合支持向量机的方法应用于轮胎ESPI缺陷图像的分类识别。理论分析和仿真实验表明使用Hu不变矩以及Hu不变矩和灰度纹理特征结合分类效果较好,总体识别率达到了96.7%。     

基于改进型PCNN图像分割技术的金属表面检测

对金属表面缺陷中的轧入氧化皮、斑点和划痕3种缺陷检测设计了一种有效方法。首先将目标图片进行灰度值归一化增强图像,提高其视觉效果方便后续图像处理。然后应用改进型脉冲耦合神经网络图像分割技术将目标图片分割得到二值图像,因为三种缺陷各有其独有的特征,据此将二值图像进行分析对比,最后对已经检测出的三种缺陷采取不同的方法标注缺陷位置。在分割方法上与Gabor分割方法进行对比实验,实验结果表明所设计的方法对三种缺陷的检测是十分迅捷且有效的,是可以用于现代化工厂缺陷检测的。     

钢板缺陷的判别及电脑三维图像重构

这里论述了在中厚钢板超声探伤系统中，钢板缺陷的确定以及钢板缺陷位置的确定。根据计算机的检测结果，详细地介绍了钢板缺陷的电脑三维图像重构技术及其实现方法和应用。     

孵化鸡蛋胚体缺陷在线图像检测系统

主要研究孵化鸡蛋胚体缺陷在线图像检测系统的开发.该系统采用工业高速CCD摄像机拍摄孵化鸡蛋胚体缺陷内部图像,通过高速的图像采集卡获取图像,运用计算机对图像进行处理,实现了鸡蛋胚体(活胚、弱胚、死胚、污染胚)缺陷的自动在线图像检测.运用彩色图像预处理、直方图信息、RGB和HSI颜色模型等相关知识,完成了对孵化鸡蛋胚体缺陷图像特征信息的提取;运用模式识别知识,实现了孵化鸡蛋胚体缺陷的判别,也相应地提高了检测准确性.实验结果表明,该系统改变了传统的人工操作模式,避免了因人而异的检测结果,减小了检测分级误差,提高了检测速度和准确率.     

基于图像处理的石英晶体片分选系统

利用图像处理的方法识别石英晶体片缺陷,进而按照品质实现自动分选。石英晶体片经光学显微镜放大后,经CCD摄像头转换成为计算机图像,其缺陷被量化。按照缺陷特征设置了分类阈值,图像处理软件按照阈值进行分选,并将分选结果传送到控制器,控制分选机构将其放入分类容器中。阐述了系统的组成、工作原理及控制器的部分软件流程,最后给出了石英晶体片分选系统目前达到的整体性能。     

炭素制品缺陷的X射线自动检测技术研究

针对炭素制品X光图像的特点,对其缺陷的提取与识别技术进行了研究,给出了目标边界提取算法和基于小波变换的图像增强算法,实现了图像的背景去除及增强处理。在此基础上,为排除噪声干扰的影响,采用数学形态学和迭代阈值分割相结合的方法从背景去除后的图像中提取出缺陷区域,取得了良好的效果。对缺陷特征选择及识别方法进行了研究,设计了基于遗传策略的特征选择和基于BP神经网络的缺陷识别算法,计算表明:缺陷正确识别率可达95%以上。采用上述技术开发完成了一套炭素制品缺陷X射线自动检测系统。     

基于SVM的玻璃瓶缺陷分类算法研究

针对玻璃瓶质量检测系统缺陷分类难的问题,选取气泡、结石、裂纹、污点、皱纹这五种常见的缺陷作为分类目标,从研究每种缺陷的图像特征入手,提出了七个统计特征作为分类器的输入特征向量,根据该分类问题的特点构建SVM分类器,采用现场采集的缺陷图像样本对SVM分类器进行训练和测试。实验结果表明:设计的SVM分类器识别率较高,适合玻璃瓶缺陷图像分类。     

列车车轮踏面缺陷的图像区域提取

建立了基于机器视觉的踏面缺陷检测系统,研究了该系统的踏面缺陷图像区域提取技术.采用基于平稳小波自适应阈值算法提取踏面区域;然后,根据踏面剥离缺陷图像特征,利用基于分块思想的粗定位和精定位组合的方法提取剥离图像区域;最后,根据踏面擦伤缺陷图像特征,利用基于踏面边缘线扫描搜索擦伤区域的方法提取擦伤图像区域.用两个实例验证了提出方法的有效性,实验结果表明:系统对剥离和擦伤两种缺陷的漏识率分别为8.3%和5.3%,误识率为5.1%,从而为后续特征提取和缺陷识别奠定了基础.     

大口径精密光学元件表面疵病检测系统研究

根据大口径精密光学元件表面疵病检测的工程特点,提出了一种基于优化模糊相似度算法的案例推理疵病识别法,解决了疵病图像获取中的摄像机平面移动定位、自动调焦和图像拼接等自动检测技术问题,研制了基于机器视觉的大口径精密光学元件表面疵病检测系统.实验结果表明,以上算法和技术是正确的,达到了理想的效果,可推广应用到其他材质的精密表面缺陷检测中.     

Comparison of systems for ultrasonic nondestructive testing using antenna arrays or phased antenna arrays

This paper considers the features of forming the images of reflectors using phased antenna array technology and the images obtained by the C-SAFT method by echo signals measured in by double-scanning mode. It is shown that in some cases the images obtained by the phased antenna array technology are less informative, while the images obtained by the C-SAFT method have a higher frontal resolution over the entire area of image restoration; the partial images restored at different positions of the antenna array can be coherently combined to obtain a high and homogeneous resolution throughout the entire volume of thickwalled articles and increasing the signal/noise ratio. For the sake of brevity, the registration of echo signals in the double-scanning mode and restoration of the images of reflectors by the C-SAFT method will be called digital focusing by an antenna array (DFA). The ability to restore the partial images of reflectors by once measured echo signals according to many acoustic schemes with their consequent combination into one high-quality image should provide reliable automation of the process of recognition and dimensioning of scatterers. Another advantage of the images obtained by the C-SAFT method with three-dimensional focusing is the possibility of restoring images in the same coordinate system when using antenna arrays on prisms of different configurations. This facilitates the joint analysis of the images. Speeds of forming the images by the technology of phased-antenna arrays and the images obtained by the C-SAFT method are comparable. If the testing technique is based on the use of nonlinear effects, then in this case PA flaw detectors have a distinct advantage over DFA flaw detectors. However, within the framework of linear acoustics, PA flaw detectors have no fundamental advantages over DFA flaw detectors. It is fairer to say that PA flaw detectors have drawbacks. This article shows images that illustrate the features of the images that were obtained by PA and DFA flaw detectors.     

A new signal processing technique for detecting flaw echoes close to the material surface in ultrasonic NDT

A new signal processing technique for separating the reflection of a flaw near a surface from the surface reflection itself is proposed. The method is based on ultrasonic scattering energy function model of flaw close to the material surface. Flaw sensitive component in the model, which corresponding to the low frequency signal, is remained by using a low pass digital filter. In order to make the recognition procedure easier and enhance the recognition effect, a reference signal is introduced to avoid the influence of the surface echo. Thus, the surface flaw is easily and clearly recognized. The good performance of the approach is experimentally verified in laboratory on a steel sample with different man-made surface flaws.     

火炮身管疵病深度测量系统

为了准确获取火炮身管疵病深度参数,提出了一种新的基于等效多基线立体成像的疵病深度测量方法,并采用单摄像机建立了疵病深度自动测量系统。首先,对摄像机采用Tsai两步法进行标校。接着,采用等效多基线立体成像法获取疵病图像。在此基础上,利用标校数据和图像的位置参数,通过疵病特征提取和疵病图像立体匹配处理,获取疵病深度图像的输出,从而得到准确的疵病深度值,实现对火炮身管疵病深度参数的精确测量。测试结果表明:该疵病测量系统得到的疵病深度测量绝对误差0.1 mm,相对误差5%,完全满足火炮身管疵病深度测量的需要,能够为火炮鉴定试验和伴装保障提供重要的技术支持。     

Fault diagnosis of rotating machine by isometric feature mapping

Principal component analysis (PCA) and linear discriminate analysis (LDA) are well-known linear dimensionality reductions for fault classification. However, since they are linear methods, they perform not well for high-dimensional data that has the nonlinear geometric structure. As kernel extension of PCA, Kernel PCA is used for nonlinear fault classification. However, the performance of Kernel PCA largely depends on its kernel function which can only be empirically selected from finite candidates. Thus, a novel rotating machine fault diagnosis approach based on geometrically motivated nonlinear dimensionality reduction named isometric feature mapping (Isomap) is proposed. The approach can effectively extract the intrinsic nonlinear manifold features embedded in high-dimensional fault data sets. Experimental results with rotor and rolling bearing data show that the proposed approach overcomes the flaw of conventional fault pattern recognition approaches and obviously improves the fault classification performance.     

局部方向模式在非接触掌纹识别中的应用

提高非接触掌纹识别系统的性能是目前掌纹识别领域一个具有实际意义的热点。针对非接触掌纹识别系统的鲁棒性问题,以掌纹图像的纹理特征为基础,提出一种基于局部方向模式(LDP)的掌纹识别方法,设计并实现了符合应用环境的嵌入式系统。LDP方法主要利用了Kirsch八方向算子的边缘响应值,从而获取图像的纹理方向模式特征。首先给出LDP算法的基本模型和流程,然后将非接触掌纹图像分成大小均匀的区块,利用LDP算法获取不同区块的纹理特征直方图向量,并进行融合形成总的模式特征,最后使用Chi距离测度进行匹配识别。在香港科技大学(HKUST)和自建的非接触掌纹图库上进行了实验测试,结果表明,该方法正确识别率可达97.824 4%和96.754 7%,相比其他典型和流行方法,最高可提升6.452 9%和5.995 6%。同时在室内环境下,利用自行设计的嵌入式原型装置进行了初步实际测试,结果表明,该方法正确识别率可达96.193 3%,具有可行性和有效性,提高了非接触掌纹识别系统的性能。     

Inspection of rotor disks of HPT and LPT of TK-10-4 gas-compressor units by the ultrasonic flaw detection method

The special features of diagnosing disks of gas turbines of gas-compressor units by the ultrasonic flaw detection method are considered. The efficiency of detecting crack-type defects in disks of HPT and LPT rotors under the first tooth of fir-tree slots of the disks in the area of trapezoidal grooves for tightening plates is shown. The application of the ultrasonic flaw detection method substantially increases the reliability of crack detection.     

Coherent Reconstruction of Flaw Images during Detection of Echo Signals in a Separate Mode

The possibility of using a scheme for separate detection of echo signals for obtaining high-quality flaw images is substantiated. A conventional algorithm for obtaining images by the projection method in the spectral space and an algorithm for obtaining images that is not limited by the approximation of a plane incident wave are considered. Coherent storage of images reconstructed from echo signals, which were detected in a separate scheme but at different positions of the radiator (transmitter), reduces the speckle-noise level and increases the resolution of flaw images. Positive and negative features of these algorithms are analyzed. Numerical and model experiments have demonstrated the feasibilities of different schemes of detecting echo signals for obtaining flaw images.     

基于混合模型与改进多分类马田系统的控制图模式识别

为提高生产过程中产品质量的智能监控水平，提出基于时间序列混合模型及改进多分类马田系统的控制图模式识别算法。选用时间序列混合模型对控制图实时数据进行特征提取；改进马田系统的阈值计算方法并制定多类判别准则，将表征的特征向量代入改进多分类马田系统分类器中进行特征约减及模式识别。最后，将该识别算法应用于控制图公开数据集及生产案例中，以验证算法的有效性，并与其他算法对比了分析，结果表明，基于时间序列混合模型及改进多分类马田系统算法能简化识别系统，识别精度高，是一种更为有效的控制图模式识别方法。     

Multiscale singular value manifold for rotating machinery fault diagnosis

Time-frequency distribution of vibration signal can be considered as an image that contains more information than signal in time domain. Manifold learning is a novel theory for image recognition that can be also applied to rotating machinery fault pattern recognition based on time-frequency distributions. However, the vibration signal of rotating machinery in fault condition contains cyclical transient impulses with different phrases which are detrimental to image recognition for time-frequency distribution. To eliminate the effects of phase differences and extract the inherent features of time-frequency distributions, a multiscale singular value manifold method is proposed. The obtained low-dimensional multiscale singular value manifold features can reveal the differences of different fault patterns and they are applicable to classification and diagnosis. Experimental verification proves that the performance of the proposed method is superior in rotating machinery fault diagnosis.