针对1JC厚板表面质量在线检测,提出了 一种2D/3D复合成像的表面检测系统.应用2D和3D成像系统分别实现厚板上下表面无深度变化及有深度变化缺陷的检测和识别,系统均采用深度学习算法进行分类模型的训练和应用.该技术已在宝钢厚板1JC精整线得到成功验证,现场一年多的应用表明,该技术可以有效实现精整线厚板表面缺陷的在线检测和识别.

针对线材表面质量信息无法按支追溯到坯、对分析异常原因及改进均带来很大局限、造成各种质量异议的问题,结合机组现场条件,自主开发了小方坯端面手写字符自动识别系统.该系统以机器视觉技术为基础,配置高性能双侧照明光源和面扫描CCD图像传感器,获得高质量的方坯端面图像;系统检测识别软件采用快速高效的边缘检测算法和深度学习算法,实时识别出端面手写字符,并将检测结果发送至服务器.系统采用C/S模式为网络架构,实现检测数据在客户端和服务器之间的可靠传递.系统上线1年多的运行实绩证明:该系统在条钢部加热炉前位置,可长期连续工作,按支跟踪准确率达98％以上,实现小方坯的在线按支自动识别和跟踪,具有卓越的检测识别性能和良好的稳定性.

荧光磁粉成像能放大和强化小方坯表面细微缺陷的显现效果.将小方坯荧光磁粉成像和机器视觉检测技术结合起来,设计1套自动化检测系统,实现小方坯的表面磁化及磁粉液喷淋、紫外线照明和荧光图像采集和处理,最终完成缺陷的定位和识别.该系统已经在工业现场得到稳定应用,提高了缺陷检测的效率和精度.

焦类炭材料的颗粒尺寸分布是重要的质量指标之一.目前,主要采用人工筛分抽样统计焦类炭材料的尺寸分布的方法,统计结果粗糙,抽样效率低下,工人劳动强度大,无法自动化完成.基于上述问题,研发了用于实际生产线的焦类炭材料颗粒筛分装置及成像测试分析方法,根据颗粒度的判断规则,最终实现自动化粒径检测,实现对焦类炭材料粒度的及时反馈.运行证明,系统运行稳定、可靠,筛分准确率很高,大大提升了生产效率和产品质量.

针对热态无缝钢管表面质量在线检测,提出了一种分体式表面质量在线检测系统的架构,应用输送辊道参数,建立了成像系统的自动对中对焦模型,实现了各规格钢管表面的清晰成像.该技术已在宝钢无缝钢管厂得到成功应用,现场两年多时间的应用表明,该技术可以有效实现热态无缝钢管表面缺陷的在线检测.

介绍了集成学习算法的原理和应用,针对工业现场特别是带钢表面状态的特殊性,即具备正常带钢表面,又含有不影响使用的伪缺陷带钢表面以及含有真实缺陷的带钢表面这一复杂现象,提出了一种基于集成学习算法进行缺陷过滤并结合多尺度卷积、特征金字塔与视觉注意力机制和传统特征的深度学习网络算法模型.通过对比验证,集成学习算法具有较高的准确性和鲁棒性,能够满足工业现场需求.

Character recognition has always been a hot topic in the field of computer vision.However, it is often difficult to obtain high-precision results in the actual scene owing to factors such as lighting conditions and imaging angle.Aiming at the problem of handwritten billet identification in the steel industry, this paper proposes the use of the canny edge extraction method to enhance the contour characteristics of characters.This technique is combined with the object detection network to achieve the automatic identification of blank square numbers and solve the problem of automatic tracking of billet logistics in the production process.The proposed algorithm is applied to the site with more than 2 019 images containing characters in the test set.Results show that the proposed algorithm has good practical application potential.

Traditional pattern recognition methods are widely used to detect defects in the industry, however most of existing methods are not universal for all kinds of defects on bars. This paper proposes a method which combines Rectified Linear Units (ReLU) and Batch Normalization(BN) in Sacked Denoising Autoencoders(SDA) for surface defect detection of bars. Gradient diffusion often occurs in traditional SDA, which leading to inefficient learning. In order to solve gradient diffusion, we replace the Sigmod activation function with ReLU function. The activation value calculated by ReLU is often oversparing which leading to loss of features. For solving the oversparing of ReLU, we add BN layer into SDA to normalize each batch. Finally we obtain network weights through unsupervised pre-training and supervised fine-tuning. We train two models which one for prediction and the other for reconstruction. Experiment results show that the proposed method can achieve an average accuracy rate of 99.1% on our data set. Compared with the traditional pattern recognition method, traditional SDA and Fisher criterion-based stacked denoising autoencoders(FCSDA), our method shows higher accuracy and TPR. Moreover, due to the addition of BN, the time complexity of our method is significantly lower than the SDA and FCSDA.

振动是影响视觉检测系统成像效果的最重要因素之一,被测对象及检测系统本身的振动对成像质量有着重要影响.在基于机器视觉的表面检测系统中,选择合适成像位置或增加稳定辊,并通过专用机械结构支撑检测系统,将检测系统与生产机组进行隔离,有效避免了生产过程中的振动对成像的影响.表面检测系统防振设计方法已在宝钢多套表面检测系统中得到应用,为表面检测系统在现场的稳定运行起到了非常重要的作用.

提出一种基于保相位变换(PHOT)的纹理抑制方法对花纹板表面进行处理,抑制规律纹理,突出非规律纹理即缺陷区域.首先对花纹板图像进行预处理,主要包括降采样、亮度补偿及水痕标记,对预处理后的图像进行FFT变换,幅度均一化及FFT反变换,得到仅保留相位信息的频域图像.对得到频域图像求梯度以放大缺陷和正常区域的差异,并采用马氏距离对梯度图进行二值化操作.最后对二值化图像进行包括缺陷重构和去水痕在内的后处理操作,从而实现花纹板表面缺陷检测.经测试,该算法对花纹板表面存在的花纹残缺、花纹错乱、翘皮和油斑四种缺陷都有较好的检出效果.

针对镀锡带钢表面差厚打印线在表面检测系统中的识别和剔除,以机组生产信息进行差厚打印线信息解析,结合图像处理功能有效地将差厚打印线与表面缺陷进行区分和识别.首先,生产系统将含有打印线模式的相关信息通过L3传送给带钢表面检测系统;然后,检测系统解析打印线模式,并综合图像处理结果进行打印线过滤.该技术已在宝钢镀锡带钢表面检测系统中得到成功应用,现场3年多时间的应用表明,该技术可以有效实现差厚打印线在表面检测系统中的检测.

The uniform of multiple CCD imaging is a key factor for surface inspection system's performance,with non-uniform input image,which will bring difficulties to the defect detection and analysis. There are many factors that make in terms of uniform imaging systems is difficult. By analyzing the impact of multi-line scan camera input uniform image factor,a targeted solution has been proposed. Eventually,a solution including software and hardware component has been formed. After applying the solution,which can be very intuitive and efficient way to improve the uniform input image level of the system. The method provides an efficient way for multi-camera imaging system in actual applications.

介绍了数学形态学的原理和应用,针对中厚板表面图像存在的低对比度、背景复杂和氧化皮干扰等问题,分析表面裂纹的形态特性,提出了基于多重形态滤波和图像块累积直方图分析的表面裂纹在线检测方法.通过与其他形态学方法的对比试验,表明该方法不仅能有效地消除噪音干扰、更为准确地检测裂纹,而且能够满足实际生产的实时性需求.

For high-speed, high-resolution images of steel strip surface online inspection, a machine vision system based on TDI (Time Delayed Integration) imaging technology was researched and developed. TDI CCD sensor was used and cooperated with adaptive illumination LED light in this system. The acquired images of steel strip were sent to the image splitter through the fiber and then real-time processed by the Image Processing Units (IPU). Surface defects on tin strip were detected in this system. Application and results show effectiveness for tin strip surface defect inspection of the system (SIS) based on TDI (Time Delayed Integration) imaging technology.

A defect detection and measurement method was proposed for detecting dots and cracks on the surface of strip steel.The method includes the following steps: first,the digital image of the sample is filtered,binarized,thinned and deburred;then dot detection is conducted to the binary image and crack detection to the skeleton image;finally,after computation is made within the defect area,the number and area of dots and the number,length and direction of cracks are obtained.Test results proved the effectiveness of this method in detecting and measuring strip steel defects.

介绍一套基于光电检测技术的冷轧薄带钢针孔在线检测系统,该系统可有效识别直径≥15μm的针孔.高亮度LED光源照射在带钢上表面,光电接收器对透过针孔的光线进行汇聚,再被光电倍增管接收,转化为电信号,信号处理电路对电信号进行识别分类,最终由采集卡采集;边部针孔检测模块对边部进行精确跟踪和遮挡,保证系统在趋零盲区情形下实现边部针孔检测.运行证明,系统具有较高的实时性、可靠性和精确性.

An on-line strip surface inspection system based on machine vision technology has been developed and applied at the exit of cold continuous rolling mill,which is capable of working under severe conditions such as limited installation space,heavy pollution from hot oil and gas,high speed of 1 600m/min and so on.An image processing card base on FPGA technology is developed,which mainly accomplishes high speed image acquisition and processing.Compressed air is used to keep the critical image acquisition parts clean and water-cooled air conditioning is applied to control the camera temperature,both of which can improve the system reliability.The inspection system has been improved and optimized to run stable and effectively detect defects such as holes,shell and scratches.

The magnetic flux leakage method to inspect internal defects in steel sheet is based on ferro-magnetic material’s magnetic phenomena in magnetic fields.If there’s any defects in a ferro-magnetic material,magnetic flux leakage field will appear in the surface of the defects.By calculating the quantity of the magnetic flux leakage,defect’s size can be got.Testing results for internal defects in steel sheet is showed in this paper.In this test,the object detection is a piece of 0.13 mm thick steel sheet with pseudo defects,which are artificial holes and each hole’s diameter is 0.13 mm.The test showed that inspecting sensors could get good magnetic flux leakage signal when the steel sheet was deeply magnetized into saturated status.Two Japanese companies,TAIHEI and NIRECO,have developed magnetic flux leakage detecting devices with high inspecting accuracy,which are 0.001 4 mm3 and 0.000 373 mm3 respectively.And now they’ve already stepped into industrial production stage.By analysis,two factors,one is industrial site’s inspecting conditions and the other is inspecting accuracy,should be considered into to select suitable magnetic sensors.Meanwhile,DC magnetizing is the best way to magnetize steel sheets into deeply saturated status.As a result of above consideration ,the magnetic flux leakage signal of very small defects can be detected .After the signal is captured,the wavelet analysis method pre-processes the signal and then the size of defects can be calculated by statistical method or neural network method.In the end,the difficulties and technical trend for applications of magnetic flux leakage detecting technology in future has been discussed.

Diamond plate is one kind of steel products. It has a complex surface topography,resulting in an increase in the difficulty of applying surface quality detection system. In this paper,it attempts to analyse one of the most common defect on the surface of diamond plate-surface water marks,then gives a method of defect detection based on the texture background,to provide effective support for the deployment of the strip surface quality online detection system on the diamond plate. It presents a novel method to remove the diamond plate surface's water mark that combined with image's gray color parameter,gradient parameter and the relationship between the upper two. The method provides a good performance to make water marks excluded from the true defects.

针对宝钢某热轧产线在线检测需求及机组现场条件,自主研发了热轧带钢表面质量在线检测系统.该系统以机器视觉技术为基础,采用大功率高亮度LED光源配合高速线扫描CCD相机获取带钢表面图像,解决了高温环境下远距离清晰成像的问题.在空间尺寸受限的情况下,通过改变LED排布保证系统清晰成像所需的有效照明效果.系统中采用了高效的图像处理和目标识别算法,保证了高速图像处理及缺陷识别.现场应用证明:宝钢热轧带钢表面质量在线检测系统具有较高的实时性、可靠性和缺陷检出及识别率.