articles+ search results

CONVOLUTIONAL neural networks, SEMICONDUCTOR wafers, SURFACE defects, SEMICONDUCTOR manufacturing, ROBOTIC welding, and MANUFACTURING processes

Semiconductor wafer is widely used in welding robot, spray robot, unmanned material delivery vehicle, and detection station sensor. The defects of semiconductor wafer, such as stains, burrs, scratches, and holes generated in the manufacturing process, severely affect the quality of downstream products. Therefore, the inspection of wafer defect could not be neglected. Traditional semiconductor wafer defect inspection methods based on handcrafted features heavily rely on the expertise and are limited in some application scenario. In this article, a novel method based on deep convolutional neural networks for semiconductor wafer surface defect inspection is proposed. First, a new structure of feature pyramid networks with atrous convolution (FPNAC) is developed to extract the features and to generate feature maps. Second, the feature maps are fed into region proposal network (RPN) to generate region proposals. Finally, the region proposals are aligned to corresponding size as the inputs of deep multibranches neural network (DMBNN) consisting of three branches, to classify and segment the defects precisely. Experimental results demonstrate that the proposed method yields good comprehensive performance with mean pixel accuracy (MPA) 93.97% and mean intersection over union (MIoU) 83.58%. [ABSTRACT FROM AUTHOR]

CONVOLUTIONAL neural networks, METALLIC surfaces, METAL defects, INSPECTION & review, and PYRAMIDS

Visual surface defect inspection for metal part has become a rapidly developing research field within the last decade. But due to the variances of defect shapes and scales, the inspection of tiny and irregular shape defects has posed challenges on the robustness of the inspection model. In this context, a deep learning method based on the deformable convolution and concatenate feature pyramid (CFP) neural networks is proposed to improve the inspection. We design a deformable convolution layer in the neural networks as an attention mechanism to adaptively extract the features of defect shape and location, which enhances the inspection of the defects with large shape variances. We also merge the multiple hierarchical features collected from different deformable convolution layers by the CFP, which improves the inspection of tiny defects. The results show that the proposed method has a better generalization ability than traditional convolution neural networks. [ABSTRACT FROM AUTHOR]

MACHINING, SURFACE roughness measurement, GEOMETRIC surfaces, MILLING (Metalwork), AUTOMATION, MANUFACTURED products, and MACHINE parts

Machinery is one of the major fields that impact the growth of the country. Computer numerical control (CNC) plays a vital role in manufacturing machining parts. Each product manufactured from CNC has its performance index requirements based on its usage. The performance indexes, such as surface roughness and geometric tolerances, are important for any CNC machined product. The performance index values are affected by several controllable and uncontrollable parameters, among which the controllable parameters can be tuned to get the desired performance index values. Hence, in this article, we proposed an intelligent prediction model to predict the performance index values, such as surface roughness and geometric tolerances, considering the controllable machining parameters as the input for the model. The experimental results show that the performance index can be predicted effectively with various CNC machining parameters. Therefore, the model can be useful for the manufacturers for achieving the desired performance index values. [ABSTRACT FROM AUTHOR]

OPTICAL measurements, SURFACE topography, TACTILE sensors, OPTICAL interferometers, and SURFACE analysis

In comparison to tactile sensors, optical techniques can provide a fast, nondestructive profile/areal surface measurement solution. Nonetheless, high measurement noise, unmeasured points, and outliers are often observed in optical measurement, particularly for structured surfaces. To alleviate their detrimental impacts on the characterization of surface topography as well as the examination of micro/nanoscale geometries, a post processing filtering technique, i.e., the clustering filter, which is essentially an iterative process to find the aggregation center of a cluster of points, is implemented. The clustering filter is particularly useful for noises and outlier suppression for optical measurement of structured surfaces due to its edge-preserving capability. Five surface samples with structured features are measured by an in-house developed dispersive interferometer and a commercial white light interferometer, thereafter the measured surface data are filtered by the clustering filter. Both noise and outliers are suppressed, which not only facilitates the visualization and characterization of surface topography, but also enables the accurate evaluation of local functional geometries. [ABSTRACT FROM AUTHOR]

CONVOLUTIONAL neural networks, IMAGING systems, AUTOMATIC identification, SYSTEM identification, and RADARSAT satellites

Palmprints are a recognizable biometric feature that contains multimodal information. Contact-based high-resolution palmprints are the most commonly used image in automatic palmprint identification systems with high-level security requirements because of its high matching accuracy and robustness. Currently, the traditional contact-based high-resolution palmprint acquisition system only captures the total internal reflection (TIR) image, which is similar to fingerprint acquisition systems. However, it is worth noting that the palm is much larger than the finger, and different regions of the palm cannot be easily pressed to the same plane. We observed that the images obtained by these existing contact-based high-resolution palmprint image acquisition systems often have poor integrity and sharpness, especially in the center of the palm. Considering this problem, we proposed an improved contact-based high-resolution palmprint image acquisition system. In addition to the TIR image, it can also simultaneously acquire a diffuse reflection (DR) image. A high-resolution DR image has low contrast and is thus not suitable for direct identification. However, the high integrity and sharing of structural information with the TIR image make it possible to complement the TIR image. Our strategy is to transfer the DR image to the TIR domain through a convolution neural network (CNN) and then fuse it with the corresponding TIR image to obtain a higher quality TIR domain image. The partial-to-“full” identification experimental results show that the rank-1 accuracy of the fused TIR (FTIR) image is improved by approximately 23% compared with the traditional TIR image, which proves the effectiveness of our strategy. [ABSTRACT FROM AUTHOR]

LOCKER rooms, DEEP learning, AUDIO equipment in automobiles, SURFACE preparation, COMPUTER vision, and PAINT

The appearance quality assessment based on defect inspection for painted car-body surfaces is an essential work to monitor and analyze the level of paint appearance quality. In the industrial application, there are some challenges, such as the huge and stereo skeleton of car bodies, a variety of irregular local surface areas, low visibility of defects due to tiny real size, and specular car-body surface. To overcome these problems, a lightweight online appearance quality assessment system (OAQAS) based on parallel deep learning is proposed, it includes two parts: 1) a vision inspection subsystem with distributed multi-camera image acquisition module and 2) an appearance quality evaluation subsystem (AQES) based on parallel TinyDefectRNet for evaluating the proposed painted surface grinding difficulty criteria. TinyDefectRNet is able to inspect relatively accurate defect size, although it is trained on a coarsely annotated data set. The OAQAS is implemented in an actual painting production line of a car factory, and the application results show that our OAQAS is far superior to the manual inspection in evaluation accuracy and time consumption. Moreover, our system is lightweight so that it is easy to be plugged into existing painting production lines without rebuilding or changing the inspection room. [ABSTRACT FROM AUTHOR]

INTERMODULATION, PASSIVE components, TEST methods, and SURFACE preparation

An empirical modeling of contact nonlinearity-induced intermodulation (IM) effect on the coaxial connector is presented in this article. The IM weights on inner and outer conductors are clarified using the measurement method. The contact degeneration-induced IM evolution is quantized by considering the contact coupling effect between the inner and outer conductors. This article demonstrated a set of test methods to quantify the oxide-induced nonlinearity with contact degeneration effects, and these methods can evaluate the contact IM products and further predict the low IM lifetime of passive devices. [ABSTRACT FROM AUTHOR]

LITHIUM-ion batteries, QUALITY assurance, AUTOMATIC optical inspection, ELECTRODES, WAGES, and SUPPORT vector machines

This paper presents a novel method for lithium-ion battery electrode (LIBE) surface quality assurance. First, based on machine vision, an automatic optical inspection system is developed to check defects on LIBE. In addition, a background normalization algorithm is put forward to preprocess the large-scale LIBE with inhomogeneous thickness in uneven illumination. With the help of the auto-concentration compensation algorithm, flaws can be extracted precisely. Moreover, after characterizing the defects, features machine applied partiality parameter automatic adjustment method and partiality decision rules are exerted for defects accurate classification, which provides near-optimal performance and reduces the complexity of tuning parameters. The proposed method is computationally efficient and satisfies real-time online inspection requirement. Experimental results verify the effectiveness and performance of the proposed method according to the inspection speed and recognition rate. [ABSTRACT FROM AUTHOR]

QUARTZ crystal microbalances, POLYTEF, BORON nitride, SOUND waves, SCANNING electron microscopy, and SIGNAL-to-noise ratio

There has been tremendous interest to employ 2-D materials as sensing medium in acoustic wave sensors such as quartz crystal microbalance (QCM). We report a method to transfer polymer-free 2-D materials onto QCM through a tailor-made automated process system and resultant enhancement in the sensing characteristics of QCM. A customized polytetrafluoroethylene (PTFE) structure was designed for the automated transfer of a 2-D heterostructure material made of hexagonal boron nitride (hBN)/graphene onto the QCM. The design ensured that the 2-D material adhered to the most sensitive region of the QCM, at the center of the electrode surface. The transferred 2-D material was characterized using Raman spectroscopy and scanning electron microscopy. Also, the $S_{11}$ characteristics of the QCM were measured before and after the transfer process using a network analyzer. We observed increment in return loss and loaded $Q$ -factor ($Q_{L}$) in the presence of the 2-D material. In order to investigate the sensing characteristics of the 2-D material integrated QCM, relative humidity (RH) was taken as an example measurand, and experiments were carried out to study the adsorption and desorption of moisture. The 2-D material integrated QCM showed good sensitivity with excellent signal-to-noise ratio for RH variations as evidenced from the significant changes in return loss and conductance measurements. On the other hand, the bare QCM exhibited poor signal-to-noise ratio and inconsistent response to RH changes. [ABSTRACT FROM AUTHOR]

SURFACE defects, ALGORITHMS, ATTENTION, COMPUTER vision, and SURFACE preparation

Surface defects of explosive cartridge in the automatic sorting process are of a small area, irregular shape, and random distribution, and all problematic characteristics that hinder surface defect detection. To address these issues, a new multidefect detection method has been proposed in this paper based on a combination of an improved visual attention model and image partitioning-weighted eigenvalue (IP-WEV). First, image preprocessing is carried out by a background estimation algorithm. Then, a new fusion operator based on defects discrimination is implemented in a visual attention model to integrate intensity, orientation, and edge conspicuity into a saliency graph, in which a saliency effect of defects is considered during the fusion process. Third, a saliency map is divided into image blocks based on the image variance. This allows for the extraction of image blocks including defects, the calculation of the weighted eigenvalue, and the determination of regions containing multidefect. The IP-WEV is used to make a decision for multidefect. The experimental results show this method’s detection accuracy as 98.2%, with a less computation time and quickly detection speed. Therefore, this method could be adopted for on-line detection systems for explosive cartridge surface defects. [ABSTRACT FROM AUTHOR]

ARTIFICIAL neural networks, SURFACE defects, and QUALITY control

Surface flaw inspection is of great importance for quality control in the field of manufacture. In this paper, a novel surface flaw inspection algorithm is proposed based on adaptive multiscale image collection (AMIC) using convolutional neural networks. First, the inspection networks are pretrained with ImageNet data set. Second, the AMIC is established, which consists of adaptive multiscale image extraction and with-contour local extraction from training images. Through the AMIC, the training data set is greatly augmented, and labels of images can be accomplished automatically without artificial consumption. Then, transfer learning is performed with the AMIC established from training data set. Finally, an automatic surface flaw inspection instrument for large-volume metal components embedded with the proposed inspection algorithm is designed. Experiments with small metal components are performed to analyze the influence of parameters, and comparative experiments are carried out. The inspecting precisions for indentation, scratch, and pitted surface of the proposed method are 97.3%, 99.5%, and 100%, respectively. The experimental results demonstrate the effectiveness of the proposed method in the detection of various surface flaws. [ABSTRACT FROM AUTHOR]

LIPASES, MILKFAT, FAT content of milk, ERROR analysis in mathematics, and DETECTORS

In this paper, a novel sensor, coated with lipase and polymethyl methacrylate, is proposed to detect the fat content of milk. Phase angle of the sensor impedance is changed with the milk samples of a different fat percent. The change of the phase angle is due to the hydrolysis of triglycerides, which are the key element of the milk fat. Hydrolysis process is catalyzed by the presence of external lipase, which is immobilized on the sensor surface. The sensitivity of the sensor is experimentally found as 0.5°/% fat. Milk fat is varied from 1.5% to 20%. A rechargeable battery operated hand-held instrument is proposed, which gives the output as low-fat milk (LFM), high-fat milk (HFM), or very high-fat milk (VHFM) according to the fat percentage of the milk. A study is carried out for power consumption by the instrument. Error analysis, specifications, and performance of the proposed instrument are also studied. [ABSTRACT FROM AUTHOR]

CONVOLUTIONAL neural networks, IMAGING systems, AUTOMATIC identification, SYSTEM identification, and RADARSAT satellites

Palmprints are a recognizable biometric feature that contains multimodal information. Contact-based high-resolution palmprints are the most commonly used image in automatic palmprint identification systems with high-level security requirements because of its high matching accuracy and robustness. Currently, the traditional contact-based high-resolution palmprint acquisition system only captures the total internal reflection (TIR) image, which is similar to fingerprint acquisition systems. However, it is worth noting that the palm is much larger than the finger, and different regions of the palm cannot be easily pressed to the same plane. We observed that the images obtained by these existing contact-based high-resolution palmprint image acquisition systems often have poor integrity and sharpness, especially in the center of the palm. Considering this problem, we proposed an improved contact-based high-resolution palmprint image acquisition system. In addition to the TIR image, it can also simultaneously acquire a diffuse reflection (DR) image. A high-resolution DR image has low contrast and is thus not suitable for direct identification. However, the high integrity and sharing of structural information with the TIR image make it possible to complement the TIR image. Our strategy is to transfer the DR image to the TIR domain through a convolution neural network (CNN) and then fuse it with the corresponding TIR image to obtain a higher quality TIR domain image. The partial-to-“full” identification experimental results show that the rank-1 accuracy of the fused TIR (FTIR) image is improved by approximately 23% compared with the traditional TIR image, which proves the effectiveness of our strategy. [ABSTRACT FROM AUTHOR]

PHOTOMETRIC stereo, ALGORITHMS, IMAGE processing, COMPUTER vision, and SIMULATION methods & models

Fast inspection of a defect is a challenging task in mass production of curved surfaces, and photometric stereo (PS) utilizing multiple images from a single camera under a number of different illumination directions is a promising technique for this task due to its high sensitivity to surface normal perturbations. This paper adapts conventional PS and extends the technique to the inspection of non-Lambertian surfaces with high accuracy and efficiency. A data-driven PS is presented by establishing the Gaussian process (GP) model to represent the nonlinear reflectance behavior of various materials based on measured reflectance data sets. With the trained GP model, the surface normal can be estimated in two steps: prediction of bidirectional reflectance distribution function values under different light directions and the subsequent least-squares estimation of a surface normal. Comparison tests with other algorithms on the Mitsubishi Electric Research Laboratories data set and real workpieces with non-Lambertian materials show the superior accuracy and efficiency of the proposed method in surface normal estimation. After the surface normal of the workpiece is recovered, the defects can be detected by filtering out the perturbation of the surface normal. Experiments on steel and glossy polyester workpieces validate the efficacy of the proposed approach in detecting defects on curved non-Lambertian surfaces. [ABSTRACT FROM AUTHOR]

ELECTRICAL resistance tomography, ELECTRODES, TIKHONOV regularization, FOOD industry, and FOULING

This paper investigates the feasibility of using electrical resistance tomography (ERT) to visualize the surface deposit during a cleaning-in-place (CIP) process, which is a common problem in various industries. A pilot scale cleaning rig with a purposely designed ERT test section was used for monitoring the prerinse CIP stage of cleaning milk paste adhering to a test section. The ERT test section comprises a CIP-ERT multilinear sensor consisting of four planes of eight electrodes. The industrial tomography system, P2000 ERT system, was used to collect the voltage data that were processed in a MATLAB-based ERT toolkit. 3-D images were reconstructed using the L-curve Tikhonov method. A feasibility study was conducted to evaluate the performance of the CIP-ERT sensor. Image analysis showed that ERT is suitable for monitoring and visualizing the CIP process for the materials that foul similar to milk pastes. The success of this paper contributes to a time/energy efficient industrial process as the cleaning process can be visualized and over cleaning can be avoided. [ABSTRACT FROM AUTHOR]

RAILROAD track inspection, SURFACE defects, GAUSSIAN mixture models, MARKOV random fields, CURVATURE measurements, and IMAGE processing

Rails are among the most important components of railway transportation, and real-time defects detection of the railway is an important and challenging task because of intensity inhomogeneity, low contrast, and noise. This paper presents an automatic railway visual detection system (RVDS) for surface defects and focuses on several key issues of RVDS. First, in view of challenges such as complex condition and orbital reflectance inequality, we put forward a region-of-interest detection region extraction algorithm by vertical projection and gray contrast algorithm. In addition, a curvature filter equipped with implicit computing and surface preserving power is studied to eliminate noise and keep only the details. Then, an improved fast and robust Gaussian mixture model based on Markov random field is established for accurate and rapid surface defect segmentation. Additionally, an expectation–maximization algorithm is applied to optimize the parameters. The experimental results demonstrate that the proposed method performs well with both noisy and railway images, which enables identification and segmentation of the defects from rail surface, achieving detection performance with 92% precision and 88.8% recall rate on average, and is robust compared with the related well-established approaches. [ABSTRACT FROM AUTHOR]

ACOUSTIC emission, MONITORING of machinery, MANUFACTURING processes, PIEZOELECTRIC transducers, and DIGITAL filters (Mathematics)

Researchers have evaluated a great number of monitoring techniques in order to control the surface condition of ground parts. Piezoelectric diaphragms of lead zirconate titanate are used in many fields, but these sensors are not common in the monitoring of the machining processes. This paper proposes a method for monitoring the workpiece surface condition (normal grinding and burn) by using a piezoelectric diaphragm and feature extraction techniques. A comparison is made with a conventional acoustic emission sensor, which is a traditional sensor in the monitoring of the machining processes. Grinding tests were performed in a surface-grinding machine with Society of Automotive Engineers (SAE) 1045 steel and cubic boron nitride (CBN) grinding wheel, where the signals were collected at 2 MHz. The workpieces were thoroughly analyzed through visual inspection, surface roughness and hardness measurements, and metallographic analyses. Study on the frequency content of both signals was carried out in order to select bands closely related to the workpiece surface condition. Digital filters were applied to the raw signals and features were extracted and analyzed. The root mean square values filtered in the selected bands for both sensors presented a better fitting to the linear regression, which is highly desirable for setting a threshold to detect burn and implementing into a monitoring system. Also, the basic damage index results show an excellent behavior for grinding burn monitoring for both sensors. The method was verified by using a different grinding wheel, which clearly shows its effectiveness and demonstrates the potential use of the low-cost piezoelectric diaphragm for grinding burn monitoring. [ABSTRACT FROM AUTHOR]

COORDINATE measuring machines, COMPUTER-aided design, OPTICAL properties of surfaces, SURFACES (Geometry), and COMPUTER simulation

One of the essential problems in the measurement of the freeform surfaces on a coordinate measuring machine is to design appropriate sampling plans to improve the industrial practice in terms of the tradeoff between the sampling accuracy and the efficiency. This paper presents a curve network sampling strategy to approximate the measured surface within a required accuracy while minimizing the cost and time for the measurement by adaptively deriving the optimal sampling locations. The method iteratively extracts two sets of iso-planar curves along two different directions on the parts to form a curve network, which is used to reconstruct the measured surfaces based on the Gordon surface fitting method. Two criteria are integrated to determine the locations of the sampled curves in the sampling process, including the surface complexity and the deviation of the reconstructed surfaces from the CAD model. Both the computer simulation and the actual measurement are conducted to verify the superior sampling efficiency of the proposed method to the conventional raster fashion sampling in measuring freeform surfaces. [ABSTRACT FROM AUTHOR]

PIEZORESISTIVE effect, CONDUCTING polymer composites, DEFORMATIONS (Mechanics), DISPLACEMENT (Mechanics), and PIEZO-electric detectors

To increase the linearity of the resistive sensor, a combined system composed of the change-ratio-decreasing subunit (CRDSU) and the change-ratio-increasing subunit (CRISU) is designed. In the CRDSU/CRISU, the change ratio of the output decreases/increases with the increase of the input. CRDSU and CRISU are used as the neighboring arms of the electrical bridge to convert the measured quantity to the output voltage. The combined system for the displacement sensor based on the piezoresistive effect of conductive polymer composite is developed as an example. The piezoresistive curve of the composite has a transition deformation. If the composite deformation is more/less than the transition deformation, the change ratio of the resistance decreases/increases with the increase of the displacement. To construct the combined system in which the varying tendencies for the change ratios of the resistances for CRDSU and CRISU are opposite, the initial deformations are applied on the subunits which are placed on the opposite sides of the measured workpiece. The experimental results verify the feasibility of using the combined system to improve the linearity. [ABSTRACT FROM PUBLISHER]

METROLOGY, DATA fusion (Statistics), MULTISENSOR data fusion, SIGNAL processing, and SENSOR networks

The combined use of multiple measurement sensors is considered as a promising solution in surface metrology. Such hybrid instruments require sophisticated data fusion process to achieve overall better measurement results. This paper presents a reconstruction–registration integrated data fusion method to address the difficulty in modeling and fusing multiscaled complex data sets. The method decomposes the data sets into different scales by fitting a common surface via reconstruction and registration process so that the modeling and fusion process are also decomposed, and are only performed among the fitting and matching residuals of the data sets. The quality of the fused results is improved based on weighted mean method with the aid of Gaussian process model by taking into account the associated errors of each data set. The validity of the proposed method is verified through a series of comparison tests with existing methods by both computer simulation and actual measurement. It is shown that both enhanced registration accuracy and fusion quality are achieved by the proposed method with acceptable computation cost. The method should improve the metrological performance of the multisensor instruments in measuring complex surfaces. [ABSTRACT FROM AUTHOR]