Strip Surface Defects Research Articles

Adaptive classifier for steel strip surface defects

Surface defects detection system has been receiving increased attention as its precision, speed and less cost. One of the most challenges is reacting to accuracy deterioration with time as aged equipment and changed processes. These variables will make a tiny change to the real world model but a big impact on the classification result. In this paper, we propose a new adaptive classifier with a Bayes kernel (BYEC) which update the model with small sample to it adaptive for accuracy deterioration. Firstly, abundant features were introduced to cover lots of information about the defects. Secondly, we constructed a series of SVMs with the random subspace of the features. Then, a Bayes classifier was trained as an evolutionary kernel to fuse the results from base SVMs. Finally, we proposed the method to update the Bayes evolutionary kernel. The proposed algorithm is experimentally compared with different algorithms, experimental results demonstrate that the proposed method can be updated with small sample and fit the changed model well. Robustness, low requirement for samples and adaptive is presented in the experiment.

An End‐to‐End Steel Strip Surface Defects Recognition System Based on Convolutional Neural Networks

Steel strip surface defects recognition is very important to steel strip production and quality control, which needs further improvement. In this paper, an end‐to‐end surface defects recognition system is proposed for steel strip surface inspection. This system is based on the symmetric surround saliency map for surface defects detection and deep convolutional neural networks (CNNs) which directly use the defect image as input and defect category as output for seven classes of steel strip defects classification. The CNNs are trained purely on raw defect images and learned defect features from the training of network, which avoiding the separation between feature extraction and image classification, so that forms an end‐to‐end defects recognition pipeline. To further illustrate the superiority of the defect recognition methods with CNNs, an authoritative and standard steel strip surface defect dataset − NEU is also used to evaluate the defect recognition effect using CNNs. Experimental results demonstrate that the proposed methods perform well in steel strip surface defect detection of different types and achieve a high recognition rate for defect images. In addition, a series of data augmentation methods are discussed to analyze its effect on avoiding over‐fitting for defects recognition.

Automated surface defect detection for cold-rolled steel strip based on wavelet anisotropic diffusion method

Taking advantage of digital image processing technology, automatic inspection for surface defects of cold-rolled steel strips promises real-time restriction and is preferable to human inspectors. Due to the complexity of surface texture, the images obtained from the existing online detection system cannot show the strip surface defects exactly, which becomes one of the important problems to be solved for the detection of surface defects of cold-rolled strip. A novel wavelet-based image filtering algorithm by virtue of anisotropic diffusion filtering is proposed in this paper. The algorithm is divided into four steps: wavelet decomposition, coefficient normalisation of wavelet diffusion, wavelet reconstruction, and edge detection. Experimental results indicated that the method could not only filter off the unnecessary texture background but also preserve the valuable information in detail effectively.

Simulation of the Mechanism of Insect Bionic Compound Eye Super-Resolution Restoration Technology of Copper Strip Surface Defect Images

In view of the problem that in traditional copper strip surface defect inspection process, the low resolution of the collected images will greatly decrease the accuracy of the detect defects, through the study of biological bionic imaging technology, combined with insect bionic compound eye imaging technology, this paper proposes a visual simulation insects bionic mechanism of the copper strip surface defect image super-resolution reconstruction technique. Through the study of biological bionic imaging technology, and take the advantage of insect compound eye visual imaging mechanism, this paper uses multiple linear array CCD image sensors to collect images aimed at getting defect images in all perspectives. The actual input images are restored by the method of super-resolution reconstruction using the sample library to improve the resolution of the image. Through the large amount of experiments of different copper strip surface defect images, and then compare the results, it can be seen that taking this papers method to dispose the defect images, will improve the images PSNR value, and has greatly improved the images quality, which will do good to improve the accuracy of the copper strip surface defect detection.

Recognition and Control of the Influence Factors on the Surface Defects of Cold Rolled Strip with Emulsion Lubrication

In cold rolling, good surface quality should be maintained with the requirement of constant increasing productivity. Oil-in-water (O/W) emulsions are widely used in cold rolled strips due to their good characteristics as lubricants and coolants. The strip surface defect is one of the central quality problems in the rolling process. A poor strip surface quality after cold rolling may have a large impact on the downstream process, like annealing, galvanization and painting. The surface condition could be deteriorated in many forms affected by various factors. However, few reports focused on the effect of the lubricating property of emulsion on the final surface quality. In this paper, the general surface defects of cold rolled strips such as watermark defects, black spot defects, heat scratches and corrosion defects were analyzed by scan electron microscope (SEM) and energy spectrum analysis (EDS) and the morphology and composition of the defects have been researched. The reason of the surface defects generated is analyzed, and the factors which introduce the surface defects during rolling process are presented. Therefore, it has an important significance to identify the features of the surface defects and put forward the solution measures.

A noise robust method based on completed local binary patterns for hot-rolled steel strip surface defects

Automatic recognition method for hot-rolled steel strip surface defects is important to the steel surface inspection system. In order to improve the recognition rate, a new, simple, yet robust feature descriptor against noise named the adjacent evaluation completed local binary patterns (AECLBPs) is proposed for defect recognition. In the proposed approach, an adjacent evaluation window which is around the neighbor is constructed to modify the threshold scheme of the completed local binary pattern (CLBP). Experimental results demonstrate that the proposed approach presents the performance of defect recognition under the influence of the feature variations of the intra-class changes, the illumination and grayscale changes. Even in the toughest situation with additive Gaussian noise, the AECLBP can still achieve the moderate recognition accuracy. In addition, the strategy of using adjacent evaluation window can also be used in other methods of local binary pattern (LBP) variants.

Machine Vision Inspection of <i>In Vitro</i> Diagnostic Kits

A machine vision-based inspection system was proposed to inspect the defects of in-vitro diagnostic kits in the production line. The proposed system consisted of two sub-systems, which inspect the strip surface defects and strip assembly defects respectively. The procedure to inspect five types of major defects was determined by the application of image processing and analysis techniques such as image enhancement, edge detection, threshold segmentation, and morphology. The proposed system was implemented using 300 defect samples. Experimental results show the proposed system is effective and efficient.

Recognition and Classification of Hot Strip Surface Defect Based on Binary Tree SVM

The paper presents a new method which uses Binary Tree SVM in the automatic classification of surface defects for hot strip. Two types of Binary Tree SVMs are applied in defect classification. Compared with BP neural network and one-against-one SVM, the algorithm adopted in the paper greatly improved the accuracy of classification and decreased the classification time.

Image Classification for Steel Strip Surface Defects Based on Support Vector Machines

In order to realize less time consuming and on-line image classification for steel strip surface defects, an improved multiclass support vector machine (SVM) was proposed. The SVM used a novel algorithm and only constructed (k-1) two-class SVMs where K is the number of classes. In the testing phase, to identify the surface defects it used a new unidirectional acyclic graph which had internal (k-1) nodes and k leaves. Its testing time is less than traditional multiclass SVM method. The experiment results shows that this method is simple and less time consuming while preserving generalization ability and recognition accuracy toward steel strip surface defects.

基于超熵和模糊集理论的带钢表面缺陷分割

Because of the existence of transition zones in a cold rolling strip surface defect image,gray information and spatial structure information should be combined to segment images to obtain better image results.Therefore,the excess entropy of information entropy and fuzzy set theory were researched.As the excess entropy could be used to measure spatial structure of an image and the characteristic of image gray transition zone could be described well by the fuzzy set,an image threshold segmentation algorithm based on maximal fuzzy excess entropy was proposed.The fuzzy excess entropy was built by the combination of excess entropy and fuzzy set theory and the threshold was determined by the best membership function parameter combination according to the maximal fuzzy excess entropy value.Then,the image was segmented by using the threshold.Finally,the algorithm was compared with Ostu and 1D maximal fuzzy entropy segmentation algorithms.The experiment indicates that the proposed algorithm can extract the defect from a background exactly and can constrain the over-segmentation effectively.The quantificational evaluation of segmented image was performed by the wrong segmentation rate and effective information rate,and the results show that the effective information rate of the algorithm is higher than 82.7%,which is the maximal one among three methods.Meanwhile the wrong segmentation rate is below 2.1%.

An effective Filter Algorithm Approach to Steel Strip Surface Image

Due to the complexity of surface texture, the images obtained from the existing online detection system cannot show the strip surface defects exactly, which becomes one of the important problems to be solved for the detection of surface defects of cold-rolled strip. An innovative wavelet-based image filtering algorithm by virtue of anisotropic diffusion is therefore proposed. It decomposes the original image into the low and high-frequency components by wavelet transform, then the high-frequency components are regularized by wavelet diffusion coefficients and, finally, the filtered image is reconstructed by inverse wavelet transform. To achieve a satisfactory filtering result, the wavelet-based anisotropic diffusion is often performed iteratively. Experimental results indicated that this new algorithm could not only filter off the unnecessary texture background unnecessary texture background but also preserve the valuable information in detail effectively. With more favorable combinability in filtering, this algorithm will lay a solid foundation for the subsequent image processing, e.g. image edge detection, image auto-segment, etc.

Downcoiler surface fault prediction for a hot strip steel mill

This paper studies the means of detecting steel strip surface defects produced by a hot strip mill downcoiler. The analysis is based on the force feedback signals taken from the side-guide and the pinch-roll of the steel mill. The study compares the results from two approaches. One approach is to apply principal component analysis (PCA) on the autoregressive (AR) model spectrum, and the other is to apply PCA on the signal periodogram. The spectrum estimation processes help eliminate the noise effects, and the PCA then distinguishes the signals into different clusters. Some of these clusters will then represent the problematic strips. The analysis results on the data from China Steel Corporation (CSC) show that PCA in conjunction with the AR model is able to separate a major portion of the defective processes, and that PCA with the periodogram is not as effective. However, the two processes sometimes compensate for each other, and the combination of the methods may provide improved results.

Research on System of Quality Assessing and Defects' Information Managing of Strip Steel Surface

Under the demand of steel enterprise’s production and consumer’s usage, by way of making up and improving continuous function of steel strip surface defects inspection system on-line, the system of grade assessing of surface quality and defects’ information of strip steel has been set up based on quality statistical method. The system of steel strip surface quality grade assessing standards has been analyzed by many steel enterprises in different countries and gotten the limitations of some of the standards. According to the rules of the National Hardwood Lumber Association of America and the feature of Chinese steel enterprise production, the grade assessing standards of steel strip surface quality has been established. Based on character of steel strip surface defects’ information, through the analysis of the requirement, aim and function, system of grade assessing of surface quality and defects’ information managed of strip steel has been developed to fit the demand of the enterprises and consumers.

트리구조 신경망을 이용한 냉연 강판 표면 결함의 분류

A new tree-structured neural network classifier is proposed for the automatic real-time inspection of cold-rolled steel strip surface defects. The defects are classified into 3 groups such as area type, disk type, area & line type in the first stage of the tree-structured neural network. The defects are classified in more detail into 11 major defect types which are considered as serious defects in the second stage of neural network. The tree-structured neural network classifier consists of 4 different neural networks and optimum features are selected for each neural network classifier by using SFFS algorithm and correlation test. The developed classifier demonstrates very plausible result which is compatible with commercial products having high world-wide market shares.

Surface defects of strip rolled in 20-roll stands

Surface defects of strip rolled in 20-roll stands