Safety maintenance of power equipment is of great importance in power grids,in which image-processing-based defect recognition is supposed to classify abnormal conditions during daily inspection.However,owing to the b...Safety maintenance of power equipment is of great importance in power grids,in which image-processing-based defect recognition is supposed to classify abnormal conditions during daily inspection.However,owing to the blurred features of defect images,the current defect recognition algorithm has poor fine-grained recognition ability.Visual attention can achieve fine-grained recognition with its abil-ity to model long-range dependencies while introducing extra computational complexity,especially for multi-head attention in vision transformer structures.Under these circumstances,this paper proposes a self-reduction multi-head attention module that can reduce computational complexity and be easily combined with a Convolutional Neural Network(CNN).In this manner,local and global fea-tures can be calculated simultaneously in our proposed structure,aiming to improve the defect recognition performance.Specifically,the proposed self-reduction multi-head attention can reduce redundant parameters,thereby solving the problem of limited computational resources.Experimental results were obtained based on the defect dataset collected from the substation.The results demonstrated the efficiency and superiority of the proposed method over other advanced algorithms.展开更多
The current casting surface defect detection algorithms suffer from poor small target defect recognition and imbalance between detection performance and detection time.An improved algorithmic framework for casting def...The current casting surface defect detection algorithms suffer from poor small target defect recognition and imbalance between detection performance and detection time.An improved algorithmic framework for casting defect detection was proposed based on the DEtection TRansformer(DETR)algorithm.The algorithm takes ResNet with an efficient channel attention(ECA)-Net module as the backbone network.In addition,based on the original algorithm architecture,dynamic anchor boxes,improved multi-scale deformable attention module,and SIoU loss function are introduced to improve the sensitivity of transformer structure to input location information and scale size,and the small target defect detection performance is effectively improved.The recognition performance of the algorithm in a self-built casting defect dataset was studied.The improved DETR algorithm has 97.561% accuracy in recognizing two defects,namely sandinclusion and notch,with the detection rate being improved by 65.854% and 17.073% compared with the original DETR and you only look once(Yolo)-V5,respectively.This algorithm verifies the applicability of the transformer architecture target detection algorithm for casting defect detection tasks and provides new ideas for detecting other similar application scenarios.展开更多
This work is aimed at developing an effective method for defect recognition in thermosonic imaging.The heat mechanism of thermosonic imaging is introduced,and the problem for defect recognition is discussed.For this p...This work is aimed at developing an effective method for defect recognition in thermosonic imaging.The heat mechanism of thermosonic imaging is introduced,and the problem for defect recognition is discussed.For this purpose,defect existing in the inner wall of a metal pipeline specimen and defects embedded in a carbon fiber reinforced plastic（CFRP） laminate are tested.The experimental data are processed by pulse phase thermography（PPT） method to show the phase images at different frequencies,and the characteristic of phase angle vs frequency curve of thermal anomalies and sound area is analyzed.A binary image,which is based on the characteristic value of defects,is obtained by a new recognition algorithm to show the defects.Results demonstrate good defect recognition performance for thermosonic imaging,and the reliability of this technique can be improved by the method.展开更多
A feasible approach for the recognition of silk fabric defects based on wavelet transform and SOM neural network is proposed in this paper, the indispensable processes of which are defect images denoising and enhancem...A feasible approach for the recognition of silk fabric defects based on wavelet transform and SOM neural network is proposed in this paper, the indispensable processes of which are defect images denoising and enhancement, image edge detection, feature extraction and defects identification. Both geometrical and textmal feature parmnete~ are extracted from the edge image and the enhanced defect image, and utilize SOM neural network to recognize the common defects which silk fabrics have, including warplacking, weft-lacking, double weft, loom bars, oil-stains. Experimental resets show the advantages with high identification correctness and high inspection speed.展开更多
Detection of wood plate surface defects using image processing is a complicated problem in the forest industry as the image of the wood surface contains different kinds of defects. In order to obtain complete defect i...Detection of wood plate surface defects using image processing is a complicated problem in the forest industry as the image of the wood surface contains different kinds of defects. In order to obtain complete defect images, we used convex optimization(CO) with different weights as a pretreatment method for smoothing and the Otsu segmentation method to obtain the target defect area images. Structural similarity(SSIM) results between original image and defect image were calculated to evaluate the performance of segmentation with different convex optimization weights. The geometric and intensity features of defects were extracted before constructing a classification and regression tree(CART) classifier. The average accuracy of the classifier is 94.1% with four types of defects on Xylosma congestum wood plate surface: pinhole, crack,live knot and dead knot. Experimental results showed that CO can save the edge of target defects maximally, SSIM can select the appropriate weight for CO, and the CART classifier appears to have the advantages of good adaptability and high classification accuracy.展开更多
Small-section hydraulic tunnels are characterized by small spaces and various section forms,under complex environments,which makes it difficult to carry out an inspection by the mobile acquisition equipment.To resolve...Small-section hydraulic tunnels are characterized by small spaces and various section forms,under complex environments,which makes it difficult to carry out an inspection by the mobile acquisition equipment.To resolve these problems,an arbitrarily adjustable camera module deployment method and the corresponding automatic image acquisition equipment with multi-area array cameras are proposed and developed.Such method enables the acquisition of full-length surface images of the hydraulic tunnels with different cross-section forms and diameters by a one-way travel,and the overlap rate and accuracy of the acquired image sets meet the requirements of three-dimensional reconstruction and panoramic image generation.In addition,to improve the speed and accuracy of traditional algorithms for tunnel surface defects detection,this paper proposes an improved YOLOv5s-DECA model.The algorithm introduces DenseNet to optimize the backbone feature extraction network and incorporates an efficient channel attention ECA module to make a better extraction of features of defects.The experimental results show that mAP,and F1-score of YOLOv5-DECA are 73.4%and 74.6%,respectively,which are better than the common model in terms of accuracy and robustness.The proposed YOLOv5-DECA has great detection performance for targets with variable shapes and can solve the problem of classification imbalance in surface defects.Then,by combining YOLOv5-DECA with the direction search algorithm,a“point-ring-section”method is established to allow rapid identification of common surface defects by detecting them layer by layer with the bottom image of the stitched panorama as the seed.The presented method in this paper effectively solves the problem that a single image fails to show the overall distribution of the defects and their accurate positioning in a whole large tunnel section and the effective features of defects in an excessively large panoramic image size are difficult to be captured by the neural network.Field applications demonstrated that the presented method is adequate for high-precision and intelligent surface defect detection and positioning for different small-section hydraulic tunnels such as circular,arch-wall,and box-shaped hydraulic tunnels.展开更多
Solar cell defects exhibit significant variations and multiple types,with some defect data being difficult to acquire or having small scales,posing challenges in terms of small sample and small target in defect detect...Solar cell defects exhibit significant variations and multiple types,with some defect data being difficult to acquire or having small scales,posing challenges in terms of small sample and small target in defect detection for solar cells.In order to address this issue,this paper proposes a multi-step approach for detecting the complex defects of solar cells.First,individual cell plates are extracted from electroluminescence images for block-by-block detection.Then,StyleGAN2-Ada is utilized for generative adversarial networks data augmentation to expand the number of defect samples in small sample defects.Finally,the fake dataset is combined with real dataset,and the improved YOLOv5 model is trained on this mixed dataset.Experimental results demonstrate that the proposed method achieves a superior performance in detecting the defects with small sample and small target,with the final recall rate reaching 99.7%,an increase of 3.9% compared with the unimproved model.Additionally,the precision and mean average precision are increased by 3.4% and 3.5%,respectively.Moreover,the experiments demonstrate that the improved network training on the mixed dataset can effectively enhance the detection performance of the model.The combination of these approaches significantly improves the network’s ability to detect solar cell defects.展开更多
基金supported in part by Major Program of the National Natural Science Foundation of China under Grant 62127803.
文摘Safety maintenance of power equipment is of great importance in power grids,in which image-processing-based defect recognition is supposed to classify abnormal conditions during daily inspection.However,owing to the blurred features of defect images,the current defect recognition algorithm has poor fine-grained recognition ability.Visual attention can achieve fine-grained recognition with its abil-ity to model long-range dependencies while introducing extra computational complexity,especially for multi-head attention in vision transformer structures.Under these circumstances,this paper proposes a self-reduction multi-head attention module that can reduce computational complexity and be easily combined with a Convolutional Neural Network(CNN).In this manner,local and global fea-tures can be calculated simultaneously in our proposed structure,aiming to improve the defect recognition performance.Specifically,the proposed self-reduction multi-head attention can reduce redundant parameters,thereby solving the problem of limited computational resources.Experimental results were obtained based on the defect dataset collected from the substation.The results demonstrated the efficiency and superiority of the proposed method over other advanced algorithms.
基金the support of National Natural Science Foundation of China(No.51405002)Anhui Provincial Natural Science Foundation(No.2108085ME173)+2 种基金open funds from Anhui Province Key Laboratory of Metallurgical Engineering&Resources Recycling(No.SKF20-05)Opening Project of Engineering Technology Research Center of Anhui Education Department for Energy Saving and Pollutant Control in metallurgical processOpening Project of Anhui Engineering Laboratory for Intelligent Applications and Security of Industrial Internet(Grant No.IASII21-03)for financial support.
文摘The current casting surface defect detection algorithms suffer from poor small target defect recognition and imbalance between detection performance and detection time.An improved algorithmic framework for casting defect detection was proposed based on the DEtection TRansformer(DETR)algorithm.The algorithm takes ResNet with an efficient channel attention(ECA)-Net module as the backbone network.In addition,based on the original algorithm architecture,dynamic anchor boxes,improved multi-scale deformable attention module,and SIoU loss function are introduced to improve the sensitivity of transformer structure to input location information and scale size,and the small target defect detection performance is effectively improved.The recognition performance of the algorithm in a self-built casting defect dataset was studied.The improved DETR algorithm has 97.561% accuracy in recognizing two defects,namely sandinclusion and notch,with the detection rate being improved by 65.854% and 17.073% compared with the original DETR and you only look once(Yolo)-V5,respectively.This algorithm verifies the applicability of the transformer architecture target detection algorithm for casting defect detection tasks and provides new ideas for detecting other similar application scenarios.
基金Joint Funds of the National Natural Science Foundationof China (61079020)
文摘This work is aimed at developing an effective method for defect recognition in thermosonic imaging.The heat mechanism of thermosonic imaging is introduced,and the problem for defect recognition is discussed.For this purpose,defect existing in the inner wall of a metal pipeline specimen and defects embedded in a carbon fiber reinforced plastic（CFRP） laminate are tested.The experimental data are processed by pulse phase thermography（PPT） method to show the phase images at different frequencies,and the characteristic of phase angle vs frequency curve of thermal anomalies and sound area is analyzed.A binary image,which is based on the characteristic value of defects,is obtained by a new recognition algorithm to show the defects.Results demonstrate good defect recognition performance for thermosonic imaging,and the reliability of this technique can be improved by the method.
基金Ministry of Commerce of the People's Republic of China (PRC)
文摘A feasible approach for the recognition of silk fabric defects based on wavelet transform and SOM neural network is proposed in this paper, the indispensable processes of which are defect images denoising and enhancement, image edge detection, feature extraction and defects identification. Both geometrical and textmal feature parmnete~ are extracted from the edge image and the enhanced defect image, and utilize SOM neural network to recognize the common defects which silk fabrics have, including warplacking, weft-lacking, double weft, loom bars, oil-stains. Experimental resets show the advantages with high identification correctness and high inspection speed.
基金supported by the Fund of Forestry 948project(2015-4-52)the Fundamental Research Funds for the Central Universities(2572017DB05)the Natural Science Foundation of Heilongjiang Province(C2017005)
文摘Detection of wood plate surface defects using image processing is a complicated problem in the forest industry as the image of the wood surface contains different kinds of defects. In order to obtain complete defect images, we used convex optimization(CO) with different weights as a pretreatment method for smoothing and the Otsu segmentation method to obtain the target defect area images. Structural similarity(SSIM) results between original image and defect image were calculated to evaluate the performance of segmentation with different convex optimization weights. The geometric and intensity features of defects were extracted before constructing a classification and regression tree(CART) classifier. The average accuracy of the classifier is 94.1% with four types of defects on Xylosma congestum wood plate surface: pinhole, crack,live knot and dead knot. Experimental results showed that CO can save the edge of target defects maximally, SSIM can select the appropriate weight for CO, and the CART classifier appears to have the advantages of good adaptability and high classification accuracy.
基金funded by the Hunan Provincial Natural Science Foundation Project(Grant No.2023JJ30672)the Science and Technology Research and Development Program Project of China Railway Group Limited(Grant No.2021-Special-08(A))+1 种基金the Science and Technology Research and Development Plan Project of China National Railway Group Co.Ltd.(Grant No.L2022G003)the Open Foundation of National Engineering Laboratory for High-speed Railway Construction(No.HSR202005).
文摘Small-section hydraulic tunnels are characterized by small spaces and various section forms,under complex environments,which makes it difficult to carry out an inspection by the mobile acquisition equipment.To resolve these problems,an arbitrarily adjustable camera module deployment method and the corresponding automatic image acquisition equipment with multi-area array cameras are proposed and developed.Such method enables the acquisition of full-length surface images of the hydraulic tunnels with different cross-section forms and diameters by a one-way travel,and the overlap rate and accuracy of the acquired image sets meet the requirements of three-dimensional reconstruction and panoramic image generation.In addition,to improve the speed and accuracy of traditional algorithms for tunnel surface defects detection,this paper proposes an improved YOLOv5s-DECA model.The algorithm introduces DenseNet to optimize the backbone feature extraction network and incorporates an efficient channel attention ECA module to make a better extraction of features of defects.The experimental results show that mAP,and F1-score of YOLOv5-DECA are 73.4%and 74.6%,respectively,which are better than the common model in terms of accuracy and robustness.The proposed YOLOv5-DECA has great detection performance for targets with variable shapes and can solve the problem of classification imbalance in surface defects.Then,by combining YOLOv5-DECA with the direction search algorithm,a“point-ring-section”method is established to allow rapid identification of common surface defects by detecting them layer by layer with the bottom image of the stitched panorama as the seed.The presented method in this paper effectively solves the problem that a single image fails to show the overall distribution of the defects and their accurate positioning in a whole large tunnel section and the effective features of defects in an excessively large panoramic image size are difficult to be captured by the neural network.Field applications demonstrated that the presented method is adequate for high-precision and intelligent surface defect detection and positioning for different small-section hydraulic tunnels such as circular,arch-wall,and box-shaped hydraulic tunnels.
文摘Solar cell defects exhibit significant variations and multiple types,with some defect data being difficult to acquire or having small scales,posing challenges in terms of small sample and small target in defect detection for solar cells.In order to address this issue,this paper proposes a multi-step approach for detecting the complex defects of solar cells.First,individual cell plates are extracted from electroluminescence images for block-by-block detection.Then,StyleGAN2-Ada is utilized for generative adversarial networks data augmentation to expand the number of defect samples in small sample defects.Finally,the fake dataset is combined with real dataset,and the improved YOLOv5 model is trained on this mixed dataset.Experimental results demonstrate that the proposed method achieves a superior performance in detecting the defects with small sample and small target,with the final recall rate reaching 99.7%,an increase of 3.9% compared with the unimproved model.Additionally,the precision and mean average precision are increased by 3.4% and 3.5%,respectively.Moreover,the experiments demonstrate that the improved network training on the mixed dataset can effectively enhance the detection performance of the model.The combination of these approaches significantly improves the network’s ability to detect solar cell defects.
