Defect detection of uneven brightness in low-contrast images using basis image representation

In this paper, we propose a machine vision approach for detecting local irregular brightness in low-contrast surface images and, especially, focus on mura (brightness non-uniformity) defects in liquid crystal display (LCD) panels. A mura defect embedded in a low-contrast surface image shows no distinct intensity from its surrounding region, and even worse, the sensed image may also present uneven illumination on the surface. All these make the mura defect detection in low-contrast surface images extremely difficult.A set of basis images derived from defect-free surface images are used to represent the general appearance of a clear surface. An image to be inspected is then constructed as a linear combination of the basis images, and the coefficients of the combination form the feature vector for discriminating mura defects from clear surfaces. In order to find minimum number of basis images for efficient and effective representation, the basis images are designed such that they are both statistically independent and spatially exclusive. An independent component analysis-based model that finds both the maximum negentropy for statistical independency and minimum spatial correlation for spatial redundancy is proposed to extract the representative basis images. Experimental results have shown that the proposed method can effectively detect various mura defects in low-contrast LCD panel images.     

Anisotropic diffusion with generalized diffusion coefficient function for defect detection in low-contrast surface images

In this paper, an anisotropic diffusion model with a generalized diffusion coefficient function is presented for defect detection in low-contrast surface images and, especially, aims at material surfaces found in liquid crystal display (LCD) manufacturing. A defect embedded in a low-contrast surface image is extremely difficult to detect, because the intensity difference between the unevenly illuminated background and the defective region is hardly observable and no clear edges are present between the defect and its surroundings.The proposed anisotropic diffusion model provides a generalized diffusion mechanism that can flexibly change the curve of the diffusion coefficient function. It adaptively carries out a smoothing process for faultless areas and performs a sharpening process for defect areas in an image. An entropy criterion is proposed as the performance measure of the diffused image and then a stochastic evolutionary computation algorithm, particle swarm optimization (PSO), is applied to automatically determine the best parameter values of the generalized diffusion coefficient function. Experimental results have shown that the proposed method can effectively and efficiently detect small defects in various low-contrast surface images.     

An independent component analysis-based filter design for defect detection in low-contrast surface images

In this paper, we propose a convolution filtering scheme for detecting small defects in low-contrast uniform surface images and, especially, focus on the applications for backlight panels and glass substrates found in liquid crystal display (LCD) manufacturing. A defect embedded in a low-contrast surface image shows no distinct intensity from its surrounding region, and even worse, the sensed image may present uneven brightness on the surface. All these make the defect detection in low-contrast surface images extremely difficult.In this study, a constrained independent component analysis (ICA) model is proposed to design an optimal filter with the objective that the convolution filter will generate the most representative source intensity of the background surface without noise. The prior constraint incorporated in the ICA model confines the source values of all training image patches of a defect-free image within a small interval of control limits. In the inspection process, the same control parameter used in the constraint is also applied to set up the thresholds that make impulse responses of all pixels in faultless regions within the control limits, and those in defective regions outside the control limits. A stochastic evolutionary computation algorithm, particle swarm optimization (PSO), is applied to solve for the constrained ICA model. Experimental results have shown that the proposed method can effectively detect small defects in low-contrast backlight panels and LCD glass substrate images.     

钢板表面低对比度微小缺陷图像增强和分割

目的 环境干扰及光学元件不稳定等因素往往会造成钢板表面图像照度不均,钢板表面的微小缺陷具有图像灰度不均、对比度低、形态微小等特点,给后续图像分析和缺陷识别带来因难。为此,提出一种钢板表面低对比度微小缺陷图像增强和分割算法,以消除照度不均并突出缺陷信息,从而有效分割缺陷目标。方法 采用小波-同态滤波算法进行图像增强处理,即先利用小波变换对图像进行分解,再基于同态滤波对小波低频系数进行图像灰度修正,同时对高频系数进行高通滤波,然后将处理后的小波低频系数和高频系数进行重构得到增强的图像,从而达到消除照度不均、增强缺陷细节信息的目的。最后利用最大类间方差法（Otsu法）确定自适应阈值提供给Canny算子进行边缘检测。结果 采用本文算法对钢板表面多类型低对比度表面微小缺陷进行研究,有效消除了光照不均;单一的Otsu阈值分割和Canny算子难以有效检测这些缺陷,而本文Otsu-Canny算法的正确检测率达96%。结论 采用小波-同态滤波进行图像增强处理后,再利用Otsu-Canny算法对钢板表面多类型、低对比度的微小缺陷进行边缘检测取得了良好效果。     

基于轻量化重构网络的表面缺陷视觉检测

基于深度学习的方法在某些工业产品的表面缺陷识别和分类方面表现出优异的性能,然而大多数工业产品缺陷样本稀缺,而且特征差异大,导致这类需要大量缺陷样本训练的检测方法难以适用.提出一种基于重构网络的无监督缺陷检测算法,仅使用容易大量获得的无缺陷样本数据实现对异常缺陷的检测.提出的算法包括两个阶段:图像重构网络训练阶段和表面缺陷区域检测阶段.训练阶段通过一种轻量化结构的全卷积自编码器设计重构网络,仅使用少量正常样本进行训练,使得重构网络能够生成无缺陷重构图像,进一步提出一种结合结构性损失和L1损失的函数作为重构网络的损失函数,解决自编码器检测算法对不规则纹理表面缺陷检测效果较差的问题;缺陷检测阶段以重构图像与待测图像的残差作为缺陷的可能区域,通过常规图像操作即可实现缺陷的定位.对所提出的重构网络的无监督缺陷检测算法的网络结构、训练像素块大小、损失函数系数等影响因素进行了详细的实验分析,并在多个缺陷图像样本集上与其他同类算法做了对比,结果表明重构网络的无监督缺陷检测算法有较强的鲁棒性和准确性.由于重构网络的无监督缺陷检测算法的轻量化结构,检测1 024×1 024像素图像仅仅耗时2.82 ms,...     

Automatic mura detection based on thresholding the fused normalized first and second derivatives in four directions

Abstract— The size of flat‐panel liquid‐crystal displays is getting larger; as a result, it is becoming harder to inspect for defects and may require a human visual inspector to judge the severity of the defects on the final product. Recently, mura phenomenon, which is defined as a visual blemish with non‐uniform shapes and boundaries, is becoming a serious unpleasant effect which needs to be detected and inspected in orderto standardize the LCD's quality. Hence, an automation process based on machine vision has proven to be a good choice to facilitate and stabilize the process. An effective general algorithm for detecting different types of mura defects with various contrast, shape, and direction, based on the fusion of the normalized magnitude of first‐ and second‐order derivative responses in four directions, is proposed. The experiments applied on various types of pseudo‐mura with different shapes show an efficient detection rate of more than 90%.     

A new algorithm on the automatic TFT‐LCD mura defects inspection based on an effective background reconstruction

In this study, an automatic detection method for mura defects is developed based on an accurate reconstruction of the background and precise evaluation of the mura index level. To achieve this, an effective background reconstruction method is first developed to represent the brightness intensity of the display panel. As a result, any nonuniform brightness of the background can be removed effectively. Furthermore, the associated mura level is quantified based on the sensitivity of the human eye in order to alternatively grade the liquid‐crystal display panels. The main focus of this study is on the reconstruction of the background from the display under test image. The proposed method takes full advantage of the following three existing methods: low‐pass filtering, discrete cosine transform, and polynomial surface fitting. By applying the method to several case studies, we have shown that it is more effective compared with other existing methods in detecting various types of mura defects.     

1D-based defect detection in patterned TFT-LCD panels using characteristic fractal dimension and correlations

Thin film transistor-liquid crystal display (TFT-LCD) is a major technology for flat panel display used in a wide range of electronic devices. As the TFT-LCD panel becomes dense, small defects can only be observed at an extremely high resolution. For fast imaging of a large-sized TFT-LCD panel at a high resolution, a one-dimensional (1D) line scan system is demanded. A TFT-LCD panel image at a fine resolution presents very complicated repetitive patterns, which increases the difficulty of the defect detection task. In this paper, we propose a 1D self-comparison defect detection scheme that directly works on the 1D line images of a TFT-LCD panel. The proposed method first uses the fractal transformation to enhance the periodicity and regularity of a 1D gray-level line image, and then divides the resulting fractal signal into small segments, each of the length of the repeated period. By calculating each divided segment’s normalized cross correlation with its neighboring segments and comparing the resulting correlation value with a predetermined threshold, the segments containing a defect can be effectively identified. Since the proposed method does not require a reference template, it is invariant to changes in illumination and image translation. Experimental results on a number of microdefects in patterned TFT-LCD panel surfaces show that the proposed method can well detect various ill-defined defects and is computationally very efficient.     

基于改进相干增强扩散与纹理能量测度和高斯混合模型的导光板表面缺陷检测方法

针对液晶屏(LCD)导光板表面缺陷检测方法存在漏检率和误检率较高,对产品表面复杂渐变的纹理结构适应性差的问题,提出一种基于改进相干增强扩散(ICED)与纹理能量测度和高斯混合模型(TEM-GMM)的LCD导光板表面缺陷检测方法。首先,构建ICED模型,基于结构张量引入平均曲率流扩散(MCF)滤波,使得相干增强扩散(CED)模型对缺陷的细线状纹理有良好的边缘保持效果,并利用相干性得到缺陷纹理增强和背景纹理抑制的滤波后图像;然后,根据Laws纹理能量测度(TEM)提取图像纹理特征,将图像的背景纹理特征作为离线阶段高斯混合模型(GMM)的训练数据,使用期望最大化(EM)算法估计GMM参数;最后,计算待检测图像各像素的后验概率,并将其作为在线检测阶段缺陷像素的判断依据。实验结果表明,该检测方法在导光颗粒随机、规则两种分布的缺陷图像测试数据组上的漏检率和误检率分别为3.27%、4.32%和3.59%、4.87%。所提检测方法适用范围广,可有效检测出LCD导光板表面划痕、异物、脏污和压伤等类型的缺陷。     

基于改进YOLOv5s的大尺寸导光板缺陷检测

导光板(LGP)是液晶显示器(LCD)背光模组的主要部件. 导光板的缺陷将直接影响液晶显示器的显示效果. 针对导光板图像纹理背景复杂、低对比度、缺陷尺寸小等问题, 本文提出了一种用于大尺寸导光板缺陷检测的AYOLOv5s网络. 首先, 将导光板图像进行分图处理, 然后在主干部分和特征融合部分集成Transformer和注意力机制coordinate attention, 并选择Meta-ACON激活函数. 最后, 基于自建数据集LGPDD进行了大量实验. 实验结果表明, LGP缺陷检测算法的平均精度(mAP)可以达到99.20%, 并且FPS可达77, 可以实现在12 s/pcs内对尺寸为17英寸的导光板中的亮点、划伤、异物、磕碰伤、脏污等缺陷具有较好的实际检测效果.     

基于反向P-M扩散的钢轨表面缺陷视觉检测

研制了一种基于反向P-M（Perona-Malik）扩散的钢轨表面缺陷视觉检测装置,该装置可 自动获取钢轨表面图像,并实现实时检测与定位钢轨表面缺陷. 钢轨图像具有光 照变化、反射不均、特征少等特点,为了在运动过程中 从复杂的钢轨表面图像提取缺陷,首先将图像进行反向P-M扩散,然后将扩散后的图像与原图像进 行差分,从而减小了上述因素的影响,最后将差分图像进行二值化操作,根据 缺陷边缘特性和面积进行滤波,分割出缺陷图像. 实验仿真和现场测试结果表明,该方法能很好地识别块状缺陷和线状缺陷,并且检测速度、精度、识别 率和误检率都能很好地满足要求.     

基于形态学的PCB缺陷快速检测技术

提出了一种基于数学形态学的PCB自动缺陷检测算法。 在对测试图像进行距离变换时,将参考图腐蚀后的边缘作为感兴趣区域使用,边缘上的每个点都具有线路边界的相应距离信息。对边缘上的距离图像进行直方图分析后得出合格线路的距离信息,以该距离为参照,可以快速地检测出各种缺陷。结合轮廓特征的对比,其能够进行准确的缺陷类型识别。实验证明,所提算法能够快速检测出PCB图像中的各种缺陷,并能进行准确的自动分类识别。     

A fast regularity measure for surface defect detection

In this paper, we propose a fast regularity measure for defect detection in non-textured and homogeneously textured surfaces, with specific emphasis on ill-defined subtle defects. A small neighborhood window of proper size is first chosen and they slide over the entire inspection image in a pixel-by-pixel basis. The regularity measure for each image patch enclosed in the window is then derived from the eigenvalues of the covariance matrix formed by the variance–covariance of the x- and y-coordinates with the pixel gray levels as the weights for all pixel points in the window. The two eigenvalues of the weighted covariance matrix will be approximately the same when the image patch contains only a homogeneous region, whereas the two eigenvalues will be relatively different if the image patch in the window contains a defect. The smaller eigenvalue of the covariance matrix is then used as the regularity measure. The integral image technique is introduced to the computation of the regularity measure so that it is invariant to the neighborhood window size. The proposed method uses only one single discrimination feature for defect detection. It avoids the use of complicated classifiers in a high-dimensional feature space, and requires no learning process from a set of defective and defect-free training samples. Experimental results on a variety of material surfaces found in industry, including textured images of plastic surfaces and leather and non-textured images of backside solar wafers and LCD backlight panels, have shown the effectiveness of the proposed regularity measure for surface defect detection. It is computationally very fast, and takes only 0.032 s for a 400 × 400 image on a Pentium 3.00?GHz personal computer. In a test set of 73 backside solar wafer images involving 53 defect-free and 20 defective samples, the proposed regularity measure can correctly identify all the test images.     

液晶显示屏背光源模组表面缺陷自动光学检测系统设计

详细介绍了自动光学检测技术在液晶显示屏背光源模组表面缺陷在线检测中的应用,分析并比较了背光源模组缺陷自动光学在线检测中的成像技术、检测系统的组成、结构原理与设计方法,阐述了检测结果为不良品的返修方法。给出了背光源模组表面缺陷常见缺陷的种类和缺陷分类判断准则,把种类繁多的背光源模组表面缺陷分为画面缺陷、外观缺陷与异常缺陷;根据背光源模组缺陷形成的原因、种类,设计了背光源模组缺陷点灯检测和非点灯检测两种自动光学检测方案,所设计的自动光学检测方案对背光源模组组装产业开发缺陷检测系统具有有益的参考价值。     

手机屏幕图像缺陷的实时检测

当前液晶屏类产品图像缺陷主要依靠人工检测,该文算法实现了图像缺陷的自动实时检测。首先通过DirectShow技术从图像采集卡缓存区快速获取实时图像;其次,将多帧实时图像加权平均,剔除坏帧,通过高斯金字塔采样去噪及分别从RGB三通道递归迭代获得分割阈值,提取屏幕矩形外框,自动校正手机姿势,继而提取ROI,完成图像的预处理;最后,利用Canny算法检测缺陷轮廓,结合Douglas-Peucker算法与弗里曼链码提取缺陷信息,最终检测手机屏幕图像缺陷：坏点数目,几何失真度,色差。算法实时、高效,依托相关国家标准,可广泛应用于液晶屏类产品的图像缺陷检测,具有一定的推广价值。     

基于机器视觉的螺纹钢表面缺陷检测方法

螺纹钢是一种广泛应用的建筑材料,在轧制过程中如果不能及时发现其尺寸和表面缺陷,就会生产出大量废品,给企业带来损失.本文设计了一种基于视觉的螺纹钢表面缺陷检测方法.先利用仿射变换对图像中歪斜的螺纹钢进行校正,然后基于霍夫变换检测纵肋边缘直线位置的方法对螺纹钢正面、侧面图像进行区分.最后针对正面、侧面图像分别进行缺陷检测,快速准确地判别表面是否存在缺陷.实验表明所设计的方法具有较好的稳定性和实用性,能有效地解决人工检测过程中效率低、误检率高等问题.     

火电机组轴承表面细小缺陷深度检测方法仿真

采用当前方法检测火电机组轴承表面细小缺陷未对高效分离背景图像和缺陷特征,导致检测细小缺陷时,检测所用的时间较长,得到的检测结果与实际不符,存在检测效率低和误检率高的问题。提出火电机组轴承表面细小缺陷深度检测方法。通过形态学滤波算法去除火电机组轴承表面图像中存在的噪声,利用曲线拟合方法实现火电机组轴承表面图像的背景估计,通过最大熵分割法火电机组轴承图像进行二值化处理,使背景图像和缺陷特征高效分离;在此基础上,火电机组轴承表面缺陷目标,通过深度置信网络在逐层学习模型的基础上实现火电机组轴承表面细小缺陷的检测。仿真结果表明,所提方法的检测效率高、误检率低。     

精密光学元件表面疵病检测技术研究

依据光学元件表面疵病存在的对比度低、边界模糊等特性,以及现有的自动化检测算法,提出了一种基于机器视觉的数字化评价系统,解决了疵病图像获取中的相机定位、图像采集、疵病检测、图像拼接以及疵病统计等自动检测技术问题.实验结果表明:系统实现了精密光学元件表面疵病的自动化检测,能够有效分辨微米(μm)量级的疵病,具有良好的疵病识别性能.     

Fabric defect detection using local contrast deviations

Defect inspection is a vital step for quality assurance in fabric production. The development of a fully automated fabric defect detection system requires robust and efficient fabric defect detection algorithms. The inspection of real fabric defects is particularly challenging due to delicate features of defects complicated by variations in weave textures and changes in environmental factors (e.g., illumination, noise, etc.). Based on characteristics of fabric structure, an approach of using local contrast deviation (LCD) is proposed for fabric defect detection in this paper. LCD is a parameter used to describe features of the contrast difference in four directions between the analyzed image and a defect-free image of the same fabric, and is used with a bilevel threshold function for defect segmentation. The validation tests on the developed algorithms were performed with fabric images from TILDA’s Textile Texture Database and captured by a line-scan camera on an inspection machine. The experimental results show that the proposed method has robustness and simplicity as opposed to the approach of using modified local binary patterns (LBP).     

基于机器学习算法的光伏组串故障诊断技术研究

常规缺陷检测方法中,主要依据光伏电站面板异常状态数据检测面板缺陷,检测结果存在着一定的随机性,导致缺陷检测结果不清晰。因此,利用了无人机影像技术,设计了光伏电站面板缺陷检测方法。提取出图像中的缺陷特征,结合无人机影像技术,通过灰度共生矩阵将缺陷图像与完整图像分割开来,识别可见光图像缺陷位置,并将缺陷图像放在光伏面板缺陷检测模型中进一步检测,使图像纹理特征与形状特征高度融合,从而实现光伏电站面板缺陷的精准检测。采用对比实验的方式,验证了该检测方法的检测置信度更高,检测精准度随之升高,能够应用于实际生活中。