WO2016201947A1 - 一种轮形铸造产品缺陷自动检测方法 - Google Patents
一种轮形铸造产品缺陷自动检测方法 Download PDFInfo
- Publication number
- WO2016201947A1 WO2016201947A1 PCT/CN2015/099632 CN2015099632W WO2016201947A1 WO 2016201947 A1 WO2016201947 A1 WO 2016201947A1 CN 2015099632 W CN2015099632 W CN 2015099632W WO 2016201947 A1 WO2016201947 A1 WO 2016201947A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image
- defect
- sample
- hub
- samples
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
- G06F18/24143—Distances to neighbourhood prototypes, e.g. restricted Coulomb energy networks [RCEN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2431—Multiple classes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
- G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
- G06V10/449—Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters
- G06V10/451—Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters with interaction between the filter responses, e.g. cortical complex cells
- G06V10/454—Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30116—Casting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30164—Workpiece; Machine component
Definitions
- the present invention relates to the field of computer vision, and in particular, to an automatic detection method for defects in a wheel casting product.
- a wheel-shaped product formed by casting processing usually causes various forms of casting defects to be generated inside the casting due to influences of process design, materials, equipment, and the like.
- the International Standards Committee ASTM has developed a defect reference image for the international standard for casting inspection and testing, and based on the outer contour of the defect pattern, the ratio of the area occupied by the defect per unit area of the product, determines the quality level of the defect affecting the product.
- the detection ⁇ usually images the casting product by X-ray equipment imaging, and compares the defects appearing with the standard defect image. When the actual defect size exceeds the user-selected quality level, that is, the defect area exceeds a certain size, The wheeled product was judged to be unqualified.
- the manual detection is an operation of the operator to determine whether the X-ray inspection apparatus detects a casting defect, and whether the size, shape, and the like of the defect are within the quality level range of the qualified product.
- the problem of this method is that the detection person is long. The work is easy to fatigue, the detection and detection efficiency is low, the judgment criteria vary from person to person, and the judgment result is subjective.
- the image processing method based on the underlying image feature determines whether a casting defect is displayed in the product image by image analysis processing methods such as denoising, image enhancement, region division, edge extraction, closed contour, target segmentation, target filling, etc. Whether the defect target has occurred, so that the first step of automatic detection is completed, and then the area calculation of the target, the peripheral measurement of the contour, and the unit product are performed according to the filled target image area. The calculation of the proportion of the defect area on the area, etc., completes the judgment of the defect level, and finally determines whether the product meets the quality standard of the product design. This method relies entirely on the underlying features of the image.
- the method is still based on the underlying features of the image, and the target recognition is performed on the images taken at different angles, and then the image fusion technology is used to improve the accuracy of the target recognition.
- the method is increased.
- Hardware cost Operation still depends on a large number of parameter settings and adjustments.
- the method of image processing technology for identifying casting defects in a single image is not mature, the above method is still an unreliable technical method.
- a method for automatically detecting defects in a wheel casting product includes the following steps:
- the output mode of the defect recognition detection is based on the user's request, or an image is displayed, or an alarm signal is given.
- step S2 sample preprocessing stage includes the following steps
- the current point pixel value is set to this result, which makes the image more prominent for analysis
- step S2-3 The image obtained in step S2-3 is normalized to speed up the convergence of the training network.
- the step S3 divides the sample into three categories: spokes, rims and axles, and the specific processing steps include:
- S3- 1 A small MxM image into which each type of sample is divided, and M may take 80, 100 or 120. Divide small image samples into positive and negative samples based on whether they contain defects;
- step S3-1 In the process of training the convolutional neural network model, in order to increase the robustness of the hub detector, the sample obtained in step S3-1 is randomly and slightly scaled ([0.96, 1.08] times ), contrast conversion (contrast coefficient [0.8, 1.2]) and rotation transformation ([-60, +60] degrees, 5 degrees each time);
- S3-3 Each time a small batch of samples is called, a small batch number can select 64 samples, randomly flip the sample horizontally, add Gaussian random noise, and randomly select NxN from the transformed small sample image.
- the region is used as a training sample of the convolutional neural network.
- a small image of 100x100 can randomly take a region of 96x96 to increase the diversity of the sample and improve the generalization ability of the trained convolutional neural network model;
- the process of S4 training to generate the hub defect detection neural network model, ie the defect detector includes: [0028] S4-1.
- the BP algorithm is used to train the rim defect detection neural network model and the spoke defect detection nerve Network model and wheel network defect detection neural network model
- S4-2 Training, the learning rate is set to 0.01; [0030] S4-3. Entering a small batch of samples per iteration, 64 samples can be input, and the parameters are updated with an average error; [0031] S4-4. Design a network model.
- the hub defect detection neural network model is a multi-layer convolutional neural network model.
- the hub defect detection neural network model consists of two parts: The first part is a multi-stage feature extractor, which alternately includes a convolutional layer, a downsampling layer and a local response normalization layer, performing convolution, downsampling and nonlinear transformation; The second part is a classifier, which is a fully connected neural network consisting of two fully connected layers connected to each other.
- a back-propagation algorithm is used to train a classifier that can correctly classify the image of the hub image extracted from the first part.
- the feature extraction of the hub defect detection neural network model designed in this scheme has two stages. The first stage is the extraction of low-level features, such as points and lines, and the second stage is a combination of low-level features to form high-level features through the training of back-propagation algorithms.
- wheel Hub Defect Detection Neural Network Model In the classification, the obtained wheel hub image is segmented into a small sample image of MxM. When M is taken as 100 ⁇ , take a small sample image of each 100 to take the upper left corner, lower left The angle, the upper right corner, the lower right corner and the center of the five 96x96 area images, the five area images are calculated by the convolutional neural network model to obtain five output values, and the five output values are averaged, and finally based on the average output value. It can be determined whether the area of the hub is defective.
- the range of network output values is [0,1], greater than the set threshold of 0.5 means that there is a defect in the corresponding area on the hub, and conversely, there is no defect in the corresponding area on the hub.
- On-line detection ⁇ according to the automatic detection process from the axle to the spoke to the rim, the detection system sequentially acquires the online image at different stations, and the specific process of the online automatic defect detection phase in the step S6 is as follows:
- S6-3 Acquiring the rim image of the hub, and preprocessing the acquired rim image by using the sample offline preprocessing method, calling the rim defect detection detector to perform defect detection on each rim image, and if a defect is found, Marking the area with defects, and naming the image with the detected time as the file name for saving;
- S6-4 The output mode of the defect recognition detection is based on the user's request, or an image is displayed, or an alarm signal is given.
- the present invention has the following advantages and beneficial effects:
- the convolutional neural network learns the characteristics of the hub defect from a large number of different spokes, rims, positive and negative samples of the axle, which is stronger than the features manually extracted from the image. Recognizability, can be classified;
- the hub defect detector obtained by the training has a certain degree in actual wheel defect detection.
- the robustness of the classification of wheel defect identification with almost no similar shape has accuracy.
- 1 is an algorithm module diagram of an automatic defect detection method for a wheel casting product according to the present invention
- FIG. 2 is a flow chart of a method for automatically detecting defects of a wheel casting product according to the present invention
- 3 is an internal connection and structure diagram of a hub defect detection neural network of the method of FIG. 2; [0047] FIG.
- FIG. 4 is an original view of the measured spoke sample
- Figure 5 is a gray histogram of Figure 4.
- FIG. 6 is an effect diagram after preprocessing the FIG. 4;
- FIG. 7 is a histogram of FIG. 6 and an indication of the largest and second largest peaks
- Figures 8-1, 8-2, and 8-3 are the results of defect detection for the spokes, rims, and axle images of the hub, respectively.
- the offline portion provides a working basis for the online portion
- the online portion is based on the method of forming the offline portion and the detector for the continuous online defect recognition detection.
- a neural network based wheeled product defect detection system and method includes the following steps:
- the output mode of the defect recognition detection is based on the user's request, or an image is displayed, or an alarm signal is given.
- FIG. 2 The specific working process of the hub defect detection method based on the convolutional neural network is shown in FIG. 2, which includes three main stages: sample preprocessing, offline training, and online detection.
- the sample preprocessing stage includes the following steps. Step:
- the current point pixel value is set to this result, which makes the image more prominent for analysis
- step S2-3 The image obtained in step S2-3 is normalized to speed up the convergence of the training network.
- the offline training phase first performs sample classification, that is, the sample is divided into three categories: spokes, rims, and axles, and the specific processing steps include:
- S3- 1 A small MxM image into which each type of sample is divided, and M may take 80, 100 or 120. Divide small image samples into positive and negative samples based on whether they contain defects;
- the sample obtained by S3-1 is randomly scaled by a small scale ([0.96, 1.08] times), contrast transformation (contrast coefficient [0.8, 1.2]), and rotational transformation ([-60, +60] degrees, each transformation 5 Degree);
- N can take 96 to increase the diversity of samples and improve the generalization ability of the trained convolutional neural network model.
- the training hub defect detection neural network model that is, the defect detector includes:
- S4-1 According to the rim sample, the spoke sample and the axle sample, the BP algorithm is used to train the rim defect detection neural network model, the spoke defect detection neural network model and the axle defect detection neural network model, and the minimum batch is used for each iteration. Ways to calculate network errors and update weights;
- the hub defect detection neural network model is a multi-layer convolutional neural network model.
- the hub defect detection neural network model consists of two parts: The first part is a multi-stage feature extractor, which alternately includes a convolutional layer, a downsampling layer and a local response normalization layer, performing convolution, downsampling and nonlinear transformation; The second part is a classifier, which is a fully connected neural network consisting of two fully connected layers connected to each other.
- a back-propagation algorithm is used to train a classifier that can correctly classify the image of the hub image extracted from the first part.
- the feature extraction of the hub defect detection neural network model designed in this scheme has two stages. The first stage is the extraction of low-level features, such as points and lines. The second stage is a combination of low-level features to form high-level features through the training of back-propagation algorithms.
- the acquired hub image is segmented into a small sample image of MxM, and M is taken as 100.
- M is taken as 100.
- Take a small sample image of each MxM take the upper left corner, the lower left corner, the upper right corner, the lower right corner, and the center five 96x96 area images, and calculate the five output values from the five regional images by the convolutional neural network model, and The five output values are averaged, and finally, based on the average output value, it can be determined whether the area of the hub is defective.
- the range of network output values is [0,1], and greater than the set threshold value of 0.5 indicates that there is a defect in the corresponding area on the hub, and conversely, there is no defect in the corresponding area on the hub;
- On-line detection ⁇ according to the automatic detection process from the axle to the spoke to the rim, the detection system sequentially acquires online images at different stations, the specific steps are as follows:
- S6-2 Acquiring the spoke image of the hub, and pretreating the acquired spoke image by using the off-line pre-processing method of the sample, and calling the spoke defect detection detector to perform defect detection on each spoke image, and if a defect is found, Marking the area with defects, and naming the image with the detected time as the file name for saving;
- S6-3 Acquiring the rim image of the hub, and using the sample offline preprocessing method to obtain the image of the rim Line preprocessing, calling the rim defect detection detector to perform defect detection on each rim image, and if there is a defect, marking the defective area, and naming the image with the detected time as the file name for saving;
- S6-4 Defect recognition detection output mode according to user requirements, or display images, or give alarm signals.
- FIG. 3 is a schematic view showing the complete structure of a hub defect detection convolutional neural network of the present invention.
- the output function of all convolutional neurons is the RELU function.
- the pooling mode of all pooling layers is the maximum pooling, and the final output layer uses the Softmax layer.
- the present invention collects an image of the hub in an actual hub production line and detects the acquired image of the hub using the hub defect detection method.
- FIG. 4 is an image of a spoke sample taken with a defect on an actual production line. As can be seen from FIG. 4, the contrast between the original image and the background is not obvious.
- FIG. 5 is an image gray histogram before preprocessing the sample image image 4 in the embodiment of the present invention, from FIG.
- FIG. 6 is an image obtained by preprocessing the measured sample 4, and it can be seen from FIG. 6 that the image defect after preprocessing is more strongly contrasted with the background, and the defect is more obvious.
- FIGS. 8-1, 8-2, and 8-3 are respectively implementations of the present invention.
- the present invention has a relatively good recognition effect in practical engineering applications.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Quality & Reliability (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Biodiversity & Conservation Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
Description
Claims
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US15/736,518 US10803573B2 (en) | 2015-06-16 | 2015-12-29 | Method for automated detection of defects in cast wheel products |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510334740.3 | 2015-06-16 | ||
CN201510334740.3A CN105118044B (zh) | 2015-06-16 | 2015-06-16 | 一种轮形铸造产品缺陷自动检测方法 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2016201947A1 true WO2016201947A1 (zh) | 2016-12-22 |
Family
ID=54666019
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2015/099632 WO2016201947A1 (zh) | 2015-06-16 | 2015-12-29 | 一种轮形铸造产品缺陷自动检测方法 |
Country Status (3)
Country | Link |
---|---|
US (1) | US10803573B2 (zh) |
CN (1) | CN105118044B (zh) |
WO (1) | WO2016201947A1 (zh) |
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP3379491A1 (en) * | 2017-03-20 | 2018-09-26 | Rolls-Royce plc | Surface defect detection |
CN109781733A (zh) * | 2017-11-15 | 2019-05-21 | 欧姆龙株式会社 | 缺陷检查装置、缺陷检查方法及计算机可读存储介质 |
CN109840900A (zh) * | 2018-12-31 | 2019-06-04 | 常州轻工职业技术学院 | 一种应用于智能制造车间的故障在线检测系统及检测方法 |
CN109859169A (zh) * | 2019-01-03 | 2019-06-07 | 重庆信息通信研究院 | 一种基于卷积神经网络的tft-lcd模组白印检测方法 |
CN110020691A (zh) * | 2019-04-11 | 2019-07-16 | 重庆信息通信研究院 | 基于卷积神经网络对抗式训练的液晶屏幕缺陷检测方法 |
CN110942457A (zh) * | 2019-11-30 | 2020-03-31 | 天津大学 | 基于数字图像处理技术的太阳能电池板缺陷检测方法 |
CN111047550A (zh) * | 2019-11-04 | 2020-04-21 | 睿视智觉(厦门)科技有限公司 | 一种基于机器视觉的产品异常检测方法、装置及设备 |
CN111062919A (zh) * | 2019-12-12 | 2020-04-24 | 韦士肯(厦门)智能科技有限公司 | 一种轴承套圈外观缺陷检测方法 |
CN111179223A (zh) * | 2019-12-12 | 2020-05-19 | 天津大学 | 基于深度学习的工业自动化缺陷检测方法 |
CN111553929A (zh) * | 2020-05-12 | 2020-08-18 | 重庆邮电大学 | 基于融合网络的手机屏幕缺陷分割方法、装置及设备 |
CN111833328A (zh) * | 2020-07-14 | 2020-10-27 | 汪俊 | 基于深度学习的飞机发动机叶片表面缺陷检测方法 |
CN113008998A (zh) * | 2021-03-02 | 2021-06-22 | 南京审计大学 | 一种基于pcnn的隐蔽工程内部缺陷判断方法 |
CN113781585A (zh) * | 2021-08-27 | 2021-12-10 | 华中科技大学 | 一种增材制造零件表面缺陷在线检测方法及系统 |
CN114841937A (zh) * | 2022-04-21 | 2022-08-02 | 燕山大学 | 一种检测汽车轮毂表面缺陷检测方法 |
CN117058151A (zh) * | 2023-10-13 | 2023-11-14 | 山东骏程金属科技有限公司 | 一种基于图像分析的轮毂检测方法及检测系统 |
Families Citing this family (75)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105118044B (zh) | 2015-06-16 | 2017-11-07 | 华南理工大学 | 一种轮形铸造产品缺陷自动检测方法 |
CN106778604B (zh) * | 2015-12-15 | 2020-04-14 | 西安电子科技大学 | 基于匹配卷积神经网络的行人再识别方法 |
CN106127780B (zh) * | 2016-06-28 | 2019-01-18 | 华南理工大学 | 一种曲面表面缺陷自动检测方法及其装置 |
CN106355579A (zh) * | 2016-09-14 | 2017-01-25 | 南京理工大学 | 烟条表面褶皱的缺陷检测方法 |
CN108154072A (zh) * | 2016-12-02 | 2018-06-12 | 天津工业大学 | 基于深度卷积神经网络的航拍绝缘子故障自动检测 |
US10402688B2 (en) * | 2016-12-07 | 2019-09-03 | Kla-Tencor Corporation | Data augmentation for convolutional neural network-based defect inspection |
CN106875373B (zh) * | 2016-12-14 | 2019-12-20 | 惠州旭鑫智能技术有限公司 | 基于卷积神经网络剪枝算法的手机屏幕mura缺陷检测方法 |
CN106875381B (zh) * | 2017-01-17 | 2020-04-28 | 同济大学 | 一种基于深度学习的手机外壳缺陷检测方法 |
US10671873B2 (en) | 2017-03-10 | 2020-06-02 | Tusimple, Inc. | System and method for vehicle wheel detection |
WO2018165753A1 (en) * | 2017-03-14 | 2018-09-20 | University Of Manitoba | Structure defect detection using machine learning algorithms |
CN109215009B (zh) * | 2017-06-29 | 2023-05-12 | 上海金艺检测技术有限公司 | 基于深度卷积神经网络的连铸坯表面图像缺陷检测方法 |
CN108021938A (zh) * | 2017-11-29 | 2018-05-11 | 中冶南方工程技术有限公司 | 一种冷轧带钢表面缺陷在线检测方法以及检测系统 |
TWI653605B (zh) * | 2017-12-25 | 2019-03-11 | 由田新技股份有限公司 | 利用深度學習的自動光學檢測方法、設備、電腦程式、電腦可讀取之記錄媒體及其深度學習系統 |
CN108242054A (zh) * | 2018-01-09 | 2018-07-03 | 北京百度网讯科技有限公司 | 一种钢板缺陷检测方法、装置、设备和服务器 |
CN108230318B (zh) * | 2018-01-09 | 2019-04-16 | 北京百度网讯科技有限公司 | 钢包缺陷检测分类方法、装置、设备及计算机可读介质 |
CN108257121B (zh) * | 2018-01-09 | 2019-01-25 | 北京百度网讯科技有限公司 | 产品缺陷检测模型更新的方法、装置、存储介质及终端设备 |
CN108564104A (zh) * | 2018-01-09 | 2018-09-21 | 北京百度网讯科技有限公司 | 产品缺陷检测方法、装置、系统、服务器及存储介质 |
CN108389197B (zh) * | 2018-02-26 | 2022-02-08 | 上海赛特斯信息科技股份有限公司 | 基于深度学习的输电线路缺陷检测方法 |
CN108510472B (zh) * | 2018-03-08 | 2019-10-22 | 北京百度网讯科技有限公司 | 用于处理图像的方法和装置 |
CN115331198A (zh) * | 2018-03-09 | 2022-11-11 | 图森有限公司 | 用于车辆车轮检测的系统和方法 |
CN108460761A (zh) * | 2018-03-12 | 2018-08-28 | 北京百度网讯科技有限公司 | 用于生成信息的方法和装置 |
CN108615230A (zh) * | 2018-03-16 | 2018-10-02 | 北京邮电大学 | 一种轮毂表面异常检测方法和系统 |
CN108491841A (zh) * | 2018-03-21 | 2018-09-04 | 东南大学 | 一种汽车轮毂类型识别监控管理系统及方法 |
CN108665057A (zh) * | 2018-03-29 | 2018-10-16 | 东南大学 | 一种基于卷积神经网络的多生产点位轮毂图像分类方法 |
CN113610831B (zh) * | 2021-08-19 | 2022-03-11 | 江西应用技术职业学院 | 基于计算机图像技术和迁移学习的木材缺陷检测方法 |
CN109035232A (zh) * | 2018-07-24 | 2018-12-18 | 桂林电子科技大学 | 一种基于深度学习和机器视觉融合的夹片缺陷检测方法 |
WO2020024079A1 (zh) * | 2018-07-28 | 2020-02-06 | 合刃科技(深圳)有限公司 | 图像识别系统 |
CN114577812A (zh) * | 2018-08-15 | 2022-06-03 | 心鉴智控(深圳)科技有限公司 | 基于超小样本训练高效质检模型的方法及系统 |
CN109239075B (zh) * | 2018-08-27 | 2021-11-30 | 北京百度网讯科技有限公司 | 电池检测方法及装置 |
CN109509170B (zh) * | 2018-09-11 | 2021-08-17 | 韶关学院 | 一种压铸件缺陷检测方法及装置 |
CN109409404A (zh) * | 2018-09-13 | 2019-03-01 | 西南交通大学 | 一种基于深度学习的高铁接触网防风拉线故障检测方法 |
CN109615606B (zh) * | 2018-11-09 | 2023-01-06 | 华南理工大学 | 一种柔性ic基板点线面缺陷的快速分类方法 |
CN109712116B (zh) * | 2018-12-04 | 2023-03-28 | 华南理工大学 | 输电线路及其附件的故障识别方法 |
JP6908019B2 (ja) * | 2018-12-28 | 2021-07-21 | Tdk株式会社 | 画像生成装置及び外観検査装置 |
DE102019201557A1 (de) * | 2019-02-07 | 2020-08-13 | Zf Friedrichshafen Ag | Verfahren und Vorrichtung zum automatisierten Identifizieren eines Produktfehlers eines Produkts und/oder zum automatisierten Identifizieren einer Produktfehlerursache des Produktfehlers |
US12079267B2 (en) * | 2019-02-21 | 2024-09-03 | Canon Kabushiki Kaisha | Method and system for image searching and evaluation using tags |
CN111768357B (zh) * | 2019-03-29 | 2024-03-01 | 银河水滴科技(北京)有限公司 | 一种图像检测的方法及装置 |
CN110033443B (zh) * | 2019-04-04 | 2021-09-03 | 武汉精立电子技术有限公司 | 一种显示面板缺陷检测方法 |
CN110264466B (zh) * | 2019-06-28 | 2021-08-06 | 广州市颐创信息科技有限公司 | 一种基于深度卷积神经网络的钢筋检测方法 |
CN110473178A (zh) * | 2019-07-30 | 2019-11-19 | 上海深视信息科技有限公司 | 一种基于多光源融合的外观缺陷检测方法及系统 |
CN110503638B (zh) * | 2019-08-15 | 2023-06-02 | 上海理工大学 | 螺旋胶质量在线检测方法 |
CN112288679B (zh) * | 2019-11-06 | 2024-03-29 | 中北大学 | 汽车轮毂x射线图像的缺陷区域交互式提取方法及系统 |
CN110927171A (zh) * | 2019-12-09 | 2020-03-27 | 中国科学院沈阳自动化研究所 | 一种基于机器视觉的轴承滚子倒角面缺陷检测方法 |
CN111080614A (zh) * | 2019-12-12 | 2020-04-28 | 哈尔滨市科佳通用机电股份有限公司 | 铁路货车轮轮辋及踏面的破损识别方法 |
CN111080636B (zh) * | 2019-12-23 | 2023-08-25 | 山东安亚金属科技有限公司 | 彩钢瓦表面缺陷的cnn语义分割自学习检测方法 |
CN111242890A (zh) * | 2019-12-24 | 2020-06-05 | 中国航空工业集团公司北京航空精密机械研究所 | 一种芳纶纸蜂窝涂胶缺陷检测方法 |
CN111161246B (zh) | 2019-12-30 | 2024-05-14 | 歌尔股份有限公司 | 一种产品缺陷检测方法、装置与系统 |
CN111260640B (zh) * | 2020-01-13 | 2023-03-31 | 重庆大学 | 基于cyclegan的树状生成器网络齿轮点蚀图像测量方法及装置 |
CN112766481B (zh) * | 2020-03-13 | 2023-11-24 | 腾讯科技(深圳)有限公司 | 神经网络模型的训练方法、装置及图像检测的方法 |
CN111798407B (zh) * | 2020-05-15 | 2024-05-21 | 国网浙江省电力有限公司嘉兴供电公司 | 一种基于神经网络模型的带电设备故障诊断方法 |
CN111709936B (zh) * | 2020-06-17 | 2023-10-24 | 广州麦仑智能科技有限公司 | 一种基于多级特征比对的令纸缺陷检测方法 |
CN112164144B (zh) * | 2020-09-24 | 2022-04-19 | 华中科技大学 | 一种结合d2算子和法向算子的铸件三维模型分类方法 |
CN112529893A (zh) * | 2020-12-22 | 2021-03-19 | 郑州金惠计算机系统工程有限公司 | 基于深度神经网络的轮毂表面瑕疵在线检测方法及系统 |
CN112598642B (zh) * | 2020-12-22 | 2024-05-10 | 苏州睿信诺智能科技有限公司 | 一种高速高精度视觉检测方法 |
CN112734742B (zh) * | 2021-01-19 | 2021-11-16 | 上海微亿智造科技有限公司 | 用于提高工业质检准确率的方法和系统 |
CN112801972A (zh) * | 2021-01-25 | 2021-05-14 | 武汉理工大学 | 一种桥梁缺陷检测方法、设备、系统及存储介质 |
CN112907519A (zh) * | 2021-01-29 | 2021-06-04 | 广州信邦智能装备股份有限公司 | 一种基于深度学习的金属曲面缺陷分析系统及方法 |
CN113344857B (zh) * | 2021-05-13 | 2022-05-03 | 深圳市华汉伟业科技有限公司 | 缺陷检测网络的训练方法、缺陷检测方法和存储介质 |
CN113762281A (zh) * | 2021-08-04 | 2021-12-07 | 浙江跃岭股份有限公司 | 一种基于机器学习的成型轮毂识别方法 |
CN113470018B (zh) * | 2021-09-01 | 2021-12-21 | 深圳市信润富联数字科技有限公司 | 轮毂缺陷识别方法、电子装置、装置及可读存储介质 |
CN113538433B (zh) * | 2021-09-17 | 2021-11-26 | 海门市创睿机械有限公司 | 基于人工智能的机械铸件缺陷检测方法及系统 |
CN116051446A (zh) * | 2021-10-26 | 2023-05-02 | 江苏时代新能源科技有限公司 | 一种极片打皱检测方法、系统、终端及存储介质 |
CN114066890A (zh) * | 2022-01-14 | 2022-02-18 | 成都数联云算科技有限公司 | 齿轮缺陷检测方法、装置、计算机设备及存储介质 |
CN114266773B (zh) * | 2022-03-02 | 2022-05-20 | 成都数联云算科技有限公司 | 显示面板缺陷定位方法、装置、设备及存储介质 |
CN114399504A (zh) * | 2022-03-25 | 2022-04-26 | 济南大学 | 一种基于工业互联网的智能轮辋检测与跟踪方法及系统 |
CN114757916A (zh) * | 2022-04-15 | 2022-07-15 | 西安交通大学 | 基于特征提取和bp网络的工业ct图像缺陷分类方法 |
CN114998259A (zh) * | 2022-06-02 | 2022-09-02 | 苏州香农科技有限公司 | 一种重力铸造轮毂的检测方法及装置 |
CN114708267B (zh) * | 2022-06-07 | 2022-09-13 | 浙江大学 | 一种输电线路上杆塔拉线腐蚀缺陷图像检测处理方法 |
CN114998593B (zh) * | 2022-06-21 | 2024-08-16 | 苏州数智源信息技术有限公司 | 一种小目标识别算法 |
CN115684175A (zh) * | 2022-10-20 | 2023-02-03 | 河南科技大学 | 一种轴承滚子端面外观分选方法及系统 |
CN115825118B (zh) * | 2022-11-18 | 2023-09-12 | 华中科技大学 | 一种铸件x射线探伤装备的自动评片集成系统及方法 |
CN117036344B (zh) * | 2023-10-07 | 2024-01-30 | 北京阿丘机器人科技有限公司 | 外观缺陷检测方法、装置、设备及存储介质 |
CN117197146A (zh) * | 2023-11-08 | 2023-12-08 | 北京航空航天大学江西研究院 | 一种铸件内部缺陷自动识别方法 |
CN117741090B (zh) * | 2024-02-19 | 2024-04-26 | 如皋市力霸铸造有限公司 | 一种变速箱精密铸件质量缺陷检测方法 |
CN118674721A (zh) * | 2024-08-23 | 2024-09-20 | 博通精密科技(浙江)有限公司 | 一种基于图像处理的无缝钢管生产用质量分析系统及方法 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20030194124A1 (en) * | 2002-04-12 | 2003-10-16 | The University Of Chicago | Massive training artificial neural network (MTANN) for detecting abnormalities in medical images |
CN104458755A (zh) * | 2014-11-26 | 2015-03-25 | 吴晓军 | 一种基于机器视觉的多类型材质表面缺陷检测方法 |
CN104574408A (zh) * | 2015-01-16 | 2015-04-29 | 东华大学 | 基于形状特征提取的工业透明薄膜包装检测方法及装置 |
CN105118044A (zh) * | 2015-06-16 | 2015-12-02 | 华南理工大学 | 一种轮形铸造产品缺陷自动检测方法 |
Family Cites Families (39)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US3978712A (en) * | 1971-11-17 | 1976-09-07 | Scanning Systems, Inc. | Method and apparatus for testing wear, size and residual stress conditions |
CA2102140C (en) * | 1993-11-01 | 1999-02-02 | Grigory Izbinsky | Wayside monitoring of the angle-of-attack of railway vehicle wheelsets |
US5636026A (en) * | 1995-03-16 | 1997-06-03 | International Electronic Machines Corporation | Method and system for contactless measurement of railroad wheel characteristics |
US6262573B1 (en) * | 1999-09-17 | 2001-07-17 | General Electric Company | Electromagnetic system for railroad track crack detection and traction enhancement |
US6523411B1 (en) * | 2000-03-21 | 2003-02-25 | International Electronic Machines Corp. | Wheel inspection system |
US6675077B2 (en) * | 2000-10-11 | 2004-01-06 | Transportation Technology Center Inc. | Wheel-railhead force measurement system and method having cross-talk removed |
US6768551B2 (en) * | 2001-10-17 | 2004-07-27 | International Electronic Machines Corp. | Contactless wheel measurement system and method |
US7525667B2 (en) * | 2004-05-24 | 2009-04-28 | International Electronic Machines Corp. | Portable electronic measurement |
RO121293B1 (ro) * | 2004-09-30 | 2007-02-28 | Mb Telecom Ltd. - S.R.L. | Metodă şi sistem de control neintruziv |
US7889931B2 (en) * | 2004-10-22 | 2011-02-15 | Gb Investments, Inc. | Systems and methods for automated vehicle image acquisition, analysis, and reporting |
US8112325B2 (en) * | 2005-09-15 | 2012-02-07 | Manheim Investments, Inc. | Method and apparatus for automatically capturing multiple images of motor vehicles and other items for sale or auction |
US7564569B2 (en) * | 2005-09-16 | 2009-07-21 | International Electronic Machines Corp. | Optical wheel evaluation |
US20090138290A1 (en) * | 2006-09-26 | 2009-05-28 | Holden Johnny L | Insurance adjustment through digital imaging system and method |
CN100590428C (zh) * | 2006-12-28 | 2010-02-17 | 华南理工大学 | 基于图像理解的轮毂铸造缺陷自动检测方法及装置 |
US7864309B2 (en) * | 2007-05-04 | 2011-01-04 | Burke E. Porter Machinery Company | Non contact wheel alignment sensor and method |
US8014586B2 (en) * | 2007-05-24 | 2011-09-06 | Applied Vision Corporation | Apparatus and methods for container inspection |
US7684034B2 (en) * | 2007-05-24 | 2010-03-23 | Applied Vision Company, Llc | Apparatus and methods for container inspection |
KR20100138725A (ko) * | 2009-06-25 | 2010-12-31 | 삼성전자주식회사 | 가상 세계 처리 장치 및 방법 |
TW201131556A (en) * | 2010-03-10 | 2011-09-16 | Quanta Storage Inc | Apparatus and method for testing spoke sensor |
CN104160258A (zh) * | 2011-11-03 | 2014-11-19 | 尼恩麦提克斯有限公司 | 用于估计车辆轮胎的气压状态的系统和方法 |
WO2013069014A1 (en) * | 2011-11-08 | 2013-05-16 | Neomatix Ltd. | System, method and application for estimating pneumatic pressure state of vehicle tires using a handheld device |
DE102012013626A1 (de) * | 2012-07-04 | 2014-01-09 | Hegenscheidt-Mfd Gmbh & Co. Kg | Verfahren und Vorrichtung zur Überprüfung von Eisenbahnrädern |
US9002095B2 (en) * | 2012-12-20 | 2015-04-07 | Wal-Mart Stores, Inc. | Faulty cart wheel detection |
US9285296B2 (en) * | 2013-01-02 | 2016-03-15 | The Boeing Company | Systems and methods for stand-off inspection of aircraft structures |
CN105122046B (zh) * | 2013-03-29 | 2018-01-23 | 斯奈克玛 | 用于检测物体上的缺陷方法 |
US9625402B2 (en) * | 2013-10-10 | 2017-04-18 | Xerox Corporation | Method and apparatus for detection of defective brakes |
US9400256B2 (en) * | 2013-12-06 | 2016-07-26 | Rolls-Royce Corporation | Thermographic inspection techniques |
CN107076645B (zh) * | 2014-11-11 | 2019-11-12 | 日本制铁株式会社 | 铁道车辆用车轮的轮缘磨损测定方法 |
WO2016115443A1 (en) * | 2015-01-16 | 2016-07-21 | International Electronic Machines Corp. | Abnormal vehicle dynamics detection |
GB2542115B (en) * | 2015-09-03 | 2017-11-15 | Rail Vision Europe Ltd | Rail track asset survey system |
US10480939B2 (en) * | 2016-01-15 | 2019-11-19 | Fugro Roadware Inc. | High speed stereoscopic pavement surface scanning system and method |
CN110139552B (zh) * | 2016-11-08 | 2023-08-15 | 道格图斯科技有限公司 | 机器人水果采摘系统 |
US10497108B1 (en) * | 2016-12-23 | 2019-12-03 | State Farm Mutual Automobile Insurance Company | Systems and methods for machine-assisted vehicle inspection |
US10580234B2 (en) * | 2017-01-20 | 2020-03-03 | Adesa, Inc. | Vehicle documentation system |
US20180222504A1 (en) * | 2017-02-08 | 2018-08-09 | Intel Corporation | Location based railway anomaly detection |
WO2018156703A1 (en) * | 2017-02-22 | 2018-08-30 | Tetra Tech, Inc. | Broken wheel detection system |
TWI628414B (zh) * | 2017-06-29 | 2018-07-01 | 峰安車業股份有限公司 | 影像檢測裝置以及影像檢測方法 |
US10564108B2 (en) * | 2017-07-03 | 2020-02-18 | Saudi Arabian Oil Company | Apparatus and method for nondestructively inspecting fiberglass and nonmetallic pipes |
US10607084B1 (en) * | 2019-10-24 | 2020-03-31 | Capital One Services, Llc | Visual inspection support using extended reality |
-
2015
- 2015-06-16 CN CN201510334740.3A patent/CN105118044B/zh not_active Expired - Fee Related
- 2015-12-29 US US15/736,518 patent/US10803573B2/en active Active
- 2015-12-29 WO PCT/CN2015/099632 patent/WO2016201947A1/zh active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20030194124A1 (en) * | 2002-04-12 | 2003-10-16 | The University Of Chicago | Massive training artificial neural network (MTANN) for detecting abnormalities in medical images |
CN104458755A (zh) * | 2014-11-26 | 2015-03-25 | 吴晓军 | 一种基于机器视觉的多类型材质表面缺陷检测方法 |
CN104574408A (zh) * | 2015-01-16 | 2015-04-29 | 东华大学 | 基于形状特征提取的工业透明薄膜包装检测方法及装置 |
CN105118044A (zh) * | 2015-06-16 | 2015-12-02 | 华南理工大学 | 一种轮形铸造产品缺陷自动检测方法 |
Cited By (30)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10467502B2 (en) | 2017-03-20 | 2019-11-05 | Rolls-Royce Plc | Surface defect detection |
EP3379491A1 (en) * | 2017-03-20 | 2018-09-26 | Rolls-Royce plc | Surface defect detection |
CN109781733A (zh) * | 2017-11-15 | 2019-05-21 | 欧姆龙株式会社 | 缺陷检查装置、缺陷检查方法及计算机可读存储介质 |
EP3486867A1 (en) * | 2017-11-15 | 2019-05-22 | Omron Corporation | Defect inspection apparatus, defect inspection method, and program thereof |
CN109781733B (zh) * | 2017-11-15 | 2022-02-22 | 欧姆龙株式会社 | 缺陷检查装置、缺陷检查方法及计算机可读存储介质 |
US10776909B2 (en) | 2017-11-15 | 2020-09-15 | Omron Corporation | Defect inspection apparatus, defect inspection method, and non-transitory computer readable medium |
CN109840900A (zh) * | 2018-12-31 | 2019-06-04 | 常州轻工职业技术学院 | 一种应用于智能制造车间的故障在线检测系统及检测方法 |
CN109840900B (zh) * | 2018-12-31 | 2023-12-19 | 常州工业职业技术学院 | 一种应用于智能制造车间的故障在线检测系统及检测方法 |
CN109859169A (zh) * | 2019-01-03 | 2019-06-07 | 重庆信息通信研究院 | 一种基于卷积神经网络的tft-lcd模组白印检测方法 |
CN109859169B (zh) * | 2019-01-03 | 2022-12-16 | 重庆信息通信研究院 | 一种基于卷积神经网络的tft-lcd模组白印检测方法 |
CN110020691A (zh) * | 2019-04-11 | 2019-07-16 | 重庆信息通信研究院 | 基于卷积神经网络对抗式训练的液晶屏幕缺陷检测方法 |
CN110020691B (zh) * | 2019-04-11 | 2022-10-11 | 重庆信息通信研究院 | 基于卷积神经网络对抗式训练的液晶屏幕缺陷检测方法 |
CN111047550A (zh) * | 2019-11-04 | 2020-04-21 | 睿视智觉(厦门)科技有限公司 | 一种基于机器视觉的产品异常检测方法、装置及设备 |
CN110942457A (zh) * | 2019-11-30 | 2020-03-31 | 天津大学 | 基于数字图像处理技术的太阳能电池板缺陷检测方法 |
CN110942457B (zh) * | 2019-11-30 | 2023-12-08 | 天津大学 | 基于数字图像处理技术的太阳能电池板缺陷检测方法 |
CN111179223B (zh) * | 2019-12-12 | 2023-05-09 | 天津大学 | 基于深度学习的工业自动化缺陷检测方法 |
CN111062919B (zh) * | 2019-12-12 | 2023-11-03 | 苏州韦士肯检测科技有限公司 | 一种轴承套圈外观缺陷检测方法 |
CN111062919A (zh) * | 2019-12-12 | 2020-04-24 | 韦士肯(厦门)智能科技有限公司 | 一种轴承套圈外观缺陷检测方法 |
CN111179223A (zh) * | 2019-12-12 | 2020-05-19 | 天津大学 | 基于深度学习的工业自动化缺陷检测方法 |
CN111553929B (zh) * | 2020-05-12 | 2022-08-09 | 重庆邮电大学 | 基于融合网络的手机屏幕缺陷分割方法、装置及设备 |
CN111553929A (zh) * | 2020-05-12 | 2020-08-18 | 重庆邮电大学 | 基于融合网络的手机屏幕缺陷分割方法、装置及设备 |
CN111833328A (zh) * | 2020-07-14 | 2020-10-27 | 汪俊 | 基于深度学习的飞机发动机叶片表面缺陷检测方法 |
CN111833328B (zh) * | 2020-07-14 | 2023-07-25 | 汪俊 | 基于深度学习的飞机发动机叶片表面缺陷检测方法 |
CN113008998A (zh) * | 2021-03-02 | 2021-06-22 | 南京审计大学 | 一种基于pcnn的隐蔽工程内部缺陷判断方法 |
CN113781585A (zh) * | 2021-08-27 | 2021-12-10 | 华中科技大学 | 一种增材制造零件表面缺陷在线检测方法及系统 |
CN113781585B (zh) * | 2021-08-27 | 2024-02-06 | 华中科技大学 | 一种增材制造零件表面缺陷在线检测方法及系统 |
CN114841937B (zh) * | 2022-04-21 | 2023-12-05 | 燕山大学 | 一种检测汽车轮毂表面缺陷检测方法 |
CN114841937A (zh) * | 2022-04-21 | 2022-08-02 | 燕山大学 | 一种检测汽车轮毂表面缺陷检测方法 |
CN117058151A (zh) * | 2023-10-13 | 2023-11-14 | 山东骏程金属科技有限公司 | 一种基于图像分析的轮毂检测方法及检测系统 |
CN117058151B (zh) * | 2023-10-13 | 2024-01-05 | 山东骏程金属科技有限公司 | 一种基于图像分析的轮毂检测方法及检测系统 |
Also Published As
Publication number | Publication date |
---|---|
US20180253836A1 (en) | 2018-09-06 |
CN105118044B (zh) | 2017-11-07 |
CN105118044A (zh) | 2015-12-02 |
US10803573B2 (en) | 2020-10-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2016201947A1 (zh) | 一种轮形铸造产品缺陷自动检测方法 | |
CN108961217B (zh) | 一种基于正例训练的表面缺陷检测方法 | |
CN103871053B (zh) | 一种基于视觉显著性的布匹瑕疵检测方法 | |
CN113505865B (zh) | 一种基于卷积神经网络的片材表面缺陷图像识别处理方法 | |
CN106780526A (zh) | 一种铁氧体圆片表面裂痕识别方法 | |
CN112907519A (zh) | 一种基于深度学习的金属曲面缺陷分析系统及方法 | |
CN111415339B (zh) | 一种复杂纹理工业产品图像缺陷检测方法 | |
CN112862744A (zh) | 一种基于超声图像的电容内部缺陷智能检测方法 | |
CN107316287A (zh) | 一种矩形铁氧体磁片表面的缺陷识别方法 | |
CN109886931A (zh) | 基于bp神经网络的轮速传感器齿圈表面缺陷检测方法 | |
CN109239073A (zh) | 一种用于汽车车身的表面缺陷检测方法 | |
CN115018846A (zh) | 基于ai智能相机的多目标裂纹缺陷检测方法和装置 | |
CN115018790A (zh) | 一种基于异常检测的工件表面缺陷检测方法 | |
CN111260609A (zh) | 一种基于深度学习的烟支外观缺陷检测方法 | |
CN114565607B (zh) | 基于神经网络的织物缺陷图像分割方法 | |
CN103926255A (zh) | 一种基于小波神经网络的布匹表面瑕疵检测方法 | |
CN113989196B (zh) | 一种基于视觉的耳机硅胶垫圈的外观缺陷检测方法 | |
CN113538342B (zh) | 一种基于卷积神经网络的铝质气雾罐涂层质量检测方法 | |
CN114199127A (zh) | 基于机器视觉的汽车零件尺寸检测系统及方法 | |
CN113884299A (zh) | 基于人工智能的滚塑机械故障检测方法 | |
CN113808104A (zh) | 一种基于分块的金属表面缺陷检测方法及系统 | |
CN117522778A (zh) | 一种空心砖瑕疵检测系统 | |
Li et al. | Detection of small size defects in belt layer of radial tire based on improved faster r-cnn | |
CN117269168A (zh) | 一种新能源汽车精密零配件表面缺陷检测装置及检测方法 | |
CN114202544B (zh) | 一种基于自编码器的复杂工件缺陷检测方法 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 15895512 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 15736518 Country of ref document: US |
|
32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 03.05.2018) |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 15895512 Country of ref document: EP Kind code of ref document: A1 |