Glass Defects Research Articles

Advanced cover glass defect detection and classification based on multi-DNN model

Demands for display panels and relevant technologies are rapidly increasing with the recent advances in smart mobile devices. Many manufacturers have begun slowly investing in fully automated inspection systems that enable consistent and objective inspections. Carrying out such undertaking aims to satisfy high user requirements concerning quality while coping with high volumes. Cover glass is one of the important items for inspection because users directly interact with it. Despite the extensive use of typical machine vision-based solutions in this field, many manufacturers continue relying on human-based judgment because of a deficient understanding of defects or poor confidence in algorithms. To overcome these problems, this study proposes a deep-learning neural network (DLNN)-based defect inspection system. The DLNN has advantages over traditional computer vision- or human-based inspection in terms of flexibility and performance. We introduce a weighted multi-DLNN inspection system capable of efficiently utilizing multi-channel measurement data, with a detection rate of up to 99% and a false pass rate below 1%.

Application of improved symmetric incremental learning algorithm in glass defect classification

This paper focuses on the application and development of image recognition technology in glass defect detection industry, and studies the basic theory and implementation method of support vector machine. Aiming at the main shortcomings of incremental learning algorithm in current support vector machines, an improved symmetric incremental learning algorithm (S-ISVM) is designed. Support vector machines are selected as training tools and classification tools, and the training samples and test samples are used as prediction samples respectively. The feasibility of the algorithm and the correctness and validity of data processing methods are verified. The research results have broad prospects for development and practical value in the field of glass industry production.

A Fuzzy Support Vector Machine-Enhanced Convolutional Neural Network for Recognition of Glass Defects

The existence of glass defects seriously affects the quality of glass products. Traditional glass defect recognition methods based on convolutional neural networks (CNNs) suffer from a long training time and low recognition accuracy. By introducing convolutional auto-encoder (CAE) into a CNN, we propose an auto-encoding convolutional neural network, and pre-train the convolution kernel by CAE, thereby reducing the training time caused by randomly initializing the convolution kernel. At the same time, in order to cope with the over-fitting problem caused by a small sample of glass defect dataset, we employ a fuzzy support vector machine (FSVM) instead of the Softmax classifier to classify glass defects. Furthermore, we propose a multi-channel auto-encoding convolutional neural network model to deal with misidentification of inclusions and tumor type defects due to small differences in feature space. The model takes a defect image and its enhanced image as input, and averages the output of each channel as the final output, thus achieving accurate recognition of glass defects. Experimental results show that the glass defect recognition method based on auto-encoding convolutional neural network can greatly reduce the training time while maintaining the same classification accuracy rate. The glass defect recognition method based on multi-channel auto-encoding convolutional neural network achieves a recognition rate of 95% for both inclusion and tumor type defects. As a result, and the overall recognition rate is increased from 92.6% using a single channel to 97%.

Floating glass on-line detection system based on defect feature maximum value method

This paper designs a glass defect online detection system based on ARM and DSP dual-core processor. The system is based on ARM LPC2200 and TMS320C6416 DSP chip, and the peripheral functional modules form the hardware. This paper also designs the extraction and analysis algorithm of defect information. This algorithm uses the defect feature maximum value method to complete the detection of the size, area, location and properties of glass defects. The experimental results have broad development prospects and practical value in the field of glass industry production..

Two-dimensional phase sensitive detector and its application to demodulating amplitude modulated image

<sec> Two-dimensional spatial modulation and demodulation technology can improve the weak signal detection capability of photoelectric detection system in a stronger noise background. In this paper, a two-dimensional phase-sensitive detector for the high-precision demodulation of 2D spatial amplitude-modulated signal is proposed. In this paper, we introduce the principle of extracting modulating signals from 2D amplitude modulated images by using 2D phase-sensitive detector, and simulate its ability to suppressing noise and extracting signal from the amplitude-modulated images buried in noise. In order to eliminate the influence of grid image generated by metal wire mesh sandwiched between two layers of glass on the detection of shielding glass defects, the methods of filtering in the frequency domain, rectifying plus filtering and two-dimensional phase sensitive detector are used to demodulate the mesh amplitude-modulated image, and the effects of extracting defects and suppressing noise are compared with each other. </sec><sec> The principle and experimental results of defect detection of ordinary glass by using external carrier are also provided. The simulation results and the detection results show that the two-dimensional phase-sensitive detector can be used to demodulate spatial two-dimensional amplitude-modulated image produced by optical modulators to extract two-dimensional measurement signals. The 2D phase-sensitive detector can greatly improve the signal-to-noise ratio of the output image, increase detection accuracy and the ability to extract modulating signals from the amplitude-modulated image buried in noise. </sec>

Enabling Computational Design of ZIFs Using ReaxFF.

Classical force fields have been broadly used in studies of metal-organic framework crystals. However, processes involving bond breaking or forming are prohibited due to the nonreactive nature of the potentials. With emerging trends in the study of zeolitic imidazolate frameworks (ZIFs) that include glass formation, defect engineering, and chemical stability, enhanced computational methods are needed for efficient computational screening of ZIF materials. Here, we present simulations of three ZIF compounds using a ReaxFF reactive force field. By simulating the melt-quench process of ZIF-4, ReaxFF can reproduce the atomic structure, density, thermal properties, and pore morphology of the glass formed ( agZIF-4), showing remarkable agreement with experimental and first-principles molecular dynamics results. The predictive capability of ReaxFF is further exemplified in the melting of ZIF-62, where the balancing of electronic and steric effects of benzimidazolate yields a lower Tm. On the basis of the electron-withdrawing effect of the -NO2 group, ReaxFF simulations predict that ZIF-77 has an even lower Tm in terms of Zn-N interaction, but its low chemical stability makes it unsuitable as a glass former. Because of its low computational cost and transferability, ReaxFF will enable the computational design of ZIF materials by accounting for properties associated with disorder/defects.

Parallelizing Hartley transform with Hadoop for fast detection of glass defects

SummaryGlass defect detection methods based on grating projection can effectively detect various glass defects. The Fourier transform in general can be used as an online processing method for detecting glass defects based on fringe images. Processing fringe images with Fourier transform needs a large amount of computation as Fourier transform is a complex computation method. In order to reduce the amount of computation, an improved fringe image processing method based on the Hartley transform is proposed in this paper. To further speed up the computation process, the Hartley transform is parallelized with Hadoop, which is a major computing technology in support of data intensive applications. Experimental results show that the parallel Hartley transform significantly reduces computation complexity in detection of glass defects.

Comparative assessment of stained‐glass windows materials by infrared thermography

AbstractThis paper reports the analyses of infrared thermography images of two stained‐glass windows with the objective of the in situ characterization of this type of artworks. The analyses were carried out by active thermography. The observations revealed that glasses absorbed the long‐wave IR radiation emitted by the halogen lamps and their apparent surface temperature progressively increased. After switching the spotlight off, they experienced a progressive decrease in temperature. Silver stained glasses presented the same thermographic behavior than uncolored glasses because silver nanoparticles were too small or the yellow layer was too thin to produce a different response than the base glass with the IR radiation. The apparent surface temperature of enamels and grisailles depended on their thickness and color. Lead cames maintained an almost constant surface apparent temperature, except those painted that behave in a similar way than enamels. Metallic tin‐lead welds experienced the most important variation in the surface apparent temperature in reflection mode due to the energy reflected by the surface of the weld. Glass defects such as big bubbles were also observed.

Furnace start-up defects: AZS exudation or corrosion?

Glass defects, such as knots, cords and cat scratches, are often attributed to fused cast AZS refractory. The mechanism behind these defect formations is thought to be either run down of viscous liquid containing high concentrations of Al2O3, Na2O and low concentrations of ZrO2 from superstructure AZS or disruption of the boundary layer on the glass contact AZS. When this viscous liquid does not completely dissolve in the glass melt, glass defects are formed, often containing ZrO2 inclusions. Such defects have been reported both shortly after start-up and during the life of the furnace campaign. This paper describes the difference between AZS exudation and corrosion as it pertains to defects during the beginning start-up of a newly lined soda-lime glass furnace. Exudation and corrosion/exudation tests following the ASTM C-1223-92 exudation procedures and modifications thereof were completed, and analyses of the exudate were compared with that of viscous knot start-up defects.

Innovative approach for identifying root causes of glass defects in sterile drug product manufacturing

In sterile drug product manufacturing, scratched and broken glass containers (i.e., vials) cause product losses, glass particles, equipment contamination and additional cleaning efforts. However, mechanical resistance and exposure of vials to mechanical stress are not sufficiently understood, and no systematic approach for reducing glass-related losses is established. Manufacturers may tackle glass-related losses more rationally if (i) frequencies for inflicting disqualifying damages to drug product containers are known for given forces, (ii) actual exposure in industrial filling lines is quantified and (iii) process enhancements are derived based on collected information.In this work, an innovative approach for exploiting these opportunities, identifying glass defect root causes and reducing glass defects is provided. Devices for quantifying (i) damaging frequencies and (ii) actual exposure are presented and then applied in an industrial case study on sterile drug product manufacturing; finally, (iii) process enhancements are derived and implemented. In the case study, frequencies for scratching vials at given forces as well as breaking forces have been determined. Peak exposure in the investigated filling line was detected at 6 N. As a result of the case study, key machine parts were identified and adjusted.

Short-time Fourier transform method for online glass ream inspection

In the float glass defect inspection system based on machine vision, the common defects can be detected and classified correctly. However, the online glass ream inspection is still a problem to be solved, and the accuracy cannot fulfill the requirement of glass grading. As the glass ream is a rather slight optical distortion in most cases, it should be amplified so that it can be checked out easily. A Moire method based on the theory of superposition shadow is proposed, which can magnify the distortion effectively without increasing the system complexity, and a STFT method is presented to analyze the distortion according to the characteristics of Moire fringe pattern. With this method, the vision interference angle of glass ribbon can be calculated out accurately in real time, which makes it suitable for online glass ream inspection.

Conjugate gradient neural network‐based online recognition of glass defects

SummaryOnline recognition of glass defects plays an important role in improving the quality of glass products. This paper presents a conjugate gradient neural network‐based method, which combines phase map and the defective image for recognition of glass defects. The boundary coordinates of the connected defect region are calculated and used to extract the defect region in the defective image correspondingly. The piecewise linear gray‐level transformation is designed to reduce the noise and to enhance the signal‐to‐noise ratio of the defective image. The second iteration segmentation based on gray range is applied to calculate the low and high thresholds, and the ternary‐valued defective image is acquired. The seven features calculated by Hu invariant moment and four features extracted from the ternary‐valued defective image are used as inputs of the conjugate gradient neural network to recognize the defect type. Experimental results show that the accuracy of the recognition reaches up to 93%. Copyright © 2016 John Wiley & Sons, Ltd.

The online measurement of optical distortion for glass defect based on the grating projection method

In order to measure the optical distortion of float glass defect and obtain more accurate online measurement result, a novel online measuring method based on digital raster projection is proposed in the present paper. In this novel method, the deformation degree of optical distortion is represented as the diopter; thus a mathematical model is derived to describe the relationship between diopter and phase of fringe image. The image processing algorithm based on one-dimensional Fourier transform and elimination method of edge effect based on spliced removal method are used to process the fringe image. Finally, a diopter distribution of fringe image is obtained. The result shows that this novel method is able to fulfill the online measurement of the optical distortion with a higher accuracy of diopter which is up to 0.04D.

Direct and indirect evidence for free oxygen (O2−) in MO-silicate glasses and melts (M = Mg, Ca, Pb)

Oxygen 1s XPS spectra of a Pb-silicate glass containing 76.6 mol% PbO provide the first accurate, direct measurement of free oxide ion (O2−) in these glasses. O2− constitutes 35 (±3) mol% of total oxygen, with NBO and BO constituting, respectively, 52 (±3) and 13 (±3) mol%. All 29Si NMR and O 1s XPS results for Pb-silicate glasses indicate mol% levels of O2− containing more than ~30 mol% PbO. The O2− abundances are consistent with equilibrium thermodynamic considerations where K ~ 12 for the mass action equation involving NBO, BO, and O2−. Raman and 17O NMR spectra of two CaMg-silicate glasses indicate ~10 (±4) mol% O2− in CaMgSiO4 glass and ~18 (±4) mol% O2− in a Ca0.36Mg0.36Si0.28O1.28 glass. Oxygen species abundances are calculated using experimental results from 13 separate 29Si NMR, 17O NMR, and Raman measurements of Mg-, Ca-, and CaMg-silicate glasses. All reveal mol% levels of O2− with ~1 to 2.6 mol% in metasilicate glass and ~5 to 10 mol% in orthosilicate glass. Recent Raman experimental results also indicate O2− in CaMg-silicate glasses at levels ranging from about 1 to 10 mol%. In all there are 23 separate 29Si NMR, 17O NMR, and Raman measurements indicating mol% levels of O2− in alkaline earth silicate glasses. Eight recent MD simulations of Mg, Ca, and CaMg-silicate glasses include 21 separate simulations over a wide compositional range. All indicate mol% levels of O2− in the glasses demonstrating that the MD simulations and experimental results on these systems are in accord. There are two fundamentally important implications of these studies. First, free oxygen (O2−) is an essential constituent of Pb, Mg, Ca, and CaMg binary silicate melts and glasses. It is not an “accidental” product associated with glass or melt defects. It is instead, a thermodynamically important constituent of these binary melts (and glasses). Second, where melts are equilibrated, the mass action equation relating BO, NBO, and O2− must hold across the entire Ca, Mg, CaMg, and Pb binary systems, thereby requiring the activities and mole fractions of all three species to be defined and finite in the melts. Free oxygen, however, may be too low to be detected in highly siliceous glasses using conventional spectroscopic techniques.

Illuminating source design of online visual inspection system for glass defects

为了实现生产线上玻璃质量的自动检测，采用机器视觉的方法对生产线上的玻璃进行缺陷检测。采集图像过程中光照方式对图像质量影响很大，为了避免由光照效果不佳对玻璃质量判断造成的干扰，保证视觉检测系统可识别不同种类的裂痕，研究了机器视觉系统中光源系统相对于生产线上玻璃的位置和光强度，解决上述问题造成的影响。分析光源系统对拍摄玻璃图像中缺陷部分和质量合格部分的相关影响，提出了一种可以检测出不同种类裂痕的光源照射方式，使其能够拍摄满足条件的高质量玻璃图像。通过样本批量测试实验证明：该光源照射方式对玻璃质量在线检测的识别准确率均为90%以上，所采集图像质量满足要求，能够准确检测出不合格产品的缺陷位置、面积大小及缺陷深度。

Cadmium Borate Glass as Host Media for Nickel Oxide Dopant

Nickel-doped cadmium borate glasses of various nickel oxide contents were prepared. UV-visible, and IR spectra of the prepared glasses were studied before and after exposure to gamma ray 8MR dose. Experimental optical data for the undoped glass reveal strong UV absorption originating from contaminated trace iron mainly (Fe3+) impurities present in the chemicals used for the preparation of this glass. NiO-doped glasses show characteristic absorption bands due mainly to octahedral coordination of divalent nickel (Ni2+) ions. Gamma irradiation produces induced bands generated from intrinsic defects from the undoped glass and extrinsic defects from collective sharing of trace iron impurities and nickel ions through possible photochemical processes. The changes in the spectroscopic data are discussed in relation to the state of nickel ions in this host glass. FTIR absorption spectra reveal combined presence of triangular and tetrahedral borate units in their specific wavenumbers. The effects of both NiO and gamma irradiation on the IR spectra were evaluated by the consideration of the shielding effect of the host base glass containing high content of heavy metal oxide (CdO)and the presence of low dopant ions which are situated in interstitial positions.

Computational Modeling of Failure Patterns of Glass Panels with Imperfections Subjected to Air Blast

Glass fragments are a prime source of injury to occupants of buildings subjected to explosive events during which window glass breaks into either flying shards (annealed glass windows) or fragments (fully tempered glass). These account for injuries ranging from minor cuts to severe wounds. A successful blast-resistant glazing design requires balancing of the safety and security of the window panels with physical appearance and cost-effectiveness. Moreover, it should consider the principles of a reasonable degree of protection against explosive threats based on the proposed level of security and previous lessons learned. In this paper, explicit finite element method (FEM) with is used for modeling the window failure and investigate the failure patterns of annealed and fully tempered glass windows panels with embedded imperfections across the panel. For this purpose, the element-removal technique is utilized in accordance with the fracture micromechanics' principles and the brittle failure criteria of displacements; strains and stresses. The results of the analysis demonstrate that the response of the conventionally-framed window panels to blast loading cases depends upon the glazing technologies; blast load intensities; the glass types; and the location of imperfections. Through the application of element removal technique and stress redistributions at each infinitesimal time increment, the results of this paper further demonstrates that the existence of glass imperfections in the window panels will make them more vulnerable against the air blast overpressures for both annealed and fully-tempered glass panel cases: For fully tempered glass panels, the vulnerability arises from greater stress concentrations at the location of glass defects as well as higher brittleness of the panel itself when compared to the annealed glass panels. The results of this study show that the existence of defects along the edges will help in redistribution of the stresses in a way that the yield lines will pass through these imperfection locations. The general failure pattern for defective annealed glass panels subjected to small intensity air blast will be very similar to the failure pattern of homogeneous window panels yet with a faster pace. However, when subjected to a higher intensity air blast, the severity of the load will cause excessive rotation angles in the panel and subsequent computational divergence and instability. The panel failure will occur much faster and with different patterns compared to the case of annealed glass panels subjected to smaller intensity air blasts. KeywordsGlass; Air Blast; Safety; Window Panels; Fracture Mechanics; Crack Propagation

Study on inspection method of glass defect based on phase image processing

To solve the problem in projecting grating method, the paper presents an inspection method based on phase image processing for glass defects. In the proposed method, the wrapped phase difference is achieved according to the images of defect-free and defect-containing. The unwrapping phase algorithm, based on jump corrected is used to eliminate jump error. The segmentation of defect region is implemented by integrating grayscale mathematical morphology with high-low threshold segmentation, and the boundary coordinate of connected region is used to calculate the size and location of defect. The results demonstrate that the proposed method provides reliable identification of defects.

Iron modified structural and optical spectral properties of bismuth silicate glasses

Iron bismuth silicate glasses have been successfully synthesized by melt quenching technique. The amorphous nature of the glass samples is ascertained by the XRD patterns. The values of density, molar volume and crystalline volume have been measured and are found to decrease with increase in iron content. The glass transition temperature measured using Differential Scanning Calorimetry (DSC) also varies with increase in Fe2O3 content. The Raman and FTIR spectra of the studied glass system taken at room temperature suggests that Fe2O3 modifies the structure of bismuth silicate glasses and it acts as both network modifier as well as network former. Bismuth also plays the role of both network modifier (BiO6 octahedra) as well as network former (BiO3 pyramids) and SiO2 exists in SiO4 tetrahedral structural units with two non-bridging oxygens. The Hydrogenic excitonic model is found to be applicable to the studied glass compositions. The variation in Urbach energy value observed for the studied glass samples suggests the possibility of increase in the number of glass defects. The metallization criterion for the synthesized glass samples is determined and found to be in the range 0.30–0.38.

Defect inspection and extraction of the mobile phone cover glass based on the principal components analysis

Surface defect inspection is an important part of quality control in mobile phone cover glass manufacturing. The traditional method is usually carried out manually by experienced inspectors and thus lacks sufficient efficiency and accuracy. In this paper, an automatic defect inspection system based on the principal components analysis is proposed for five typical cover glass defects: scratch, crack, deformation, edge broken, and angle cutting. This inspection system is robust for the defect shapes and obtains high recognition accuracy. The inspection system includes three parts: pre-processing, PCA-based defect recognition, and defect edge extraction. After pre-processing, most of noise and outliers are eliminated and the pixels of defects in the image are well enhanced for inspection and recognition. The eigen-defect matrix is constructed to characterize the variation between defect images. Additionally, time consumption in constructing the eigen-defect matrix is also discussed. The experimental results show that the inspection system has achieved high accuracy for inspection and recognition.