CN108427969A - A kind of paper sheet defect sorting technique of Multiscale Morphological combination convolutional neural networks - Google Patents
A kind of paper sheet defect sorting technique of Multiscale Morphological combination convolutional neural networks Download PDFInfo
- Publication number
- CN108427969A CN108427969A CN201810260457.4A CN201810260457A CN108427969A CN 108427969 A CN108427969 A CN 108427969A CN 201810260457 A CN201810260457 A CN 201810260457A CN 108427969 A CN108427969 A CN 108427969A
- Authority
- CN
- China
- Prior art keywords
- image
- paper sheet
- defect
- sheet defect
- convolutional neural
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/34—Smoothing or thinning of the pattern; Morphological operations; Skeletonisation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
A kind of paper sheet defect sorting technique of Multiscale Morphological combination convolutional neural networks, morphology operations are carried out to original image with the structural elements of different scale, multiscale morphological gradient image is obtained to the gradient image Weighted Fusion under different scale, in order to enhance paper sheet defect contrast, the information characteristics of prominent defect Gradient Features and Defect Edge, multiscale morphological gradient image is weighted with genetic defects image and is merged, realize defect image enhancing, convolutional neural networks are input to carry out feature extraction and classify, convolutional neural networks are used in paper sheet defect classification, the Accurate classification of paper sheet defect can be fast implemented, it is simple with method, it takes short, the high feature of accuracy of identification.
Description
Technical field
The invention belongs to Image Processing and Pattern Recognition technical field, more particularly to a kind of Multiscale Morphological combination convolution
The paper sheet defect sorting technique of neural network.
Background technology
Feature extraction is the committed step of pattern-recognition, has important application in image analysis and pattern-recognition.Image
The traditional characteristic extracting method of classification is all to pre-define a kind of feature, carries out feature extraction further according to the feature defined and divides
Class.Paper image is easy to be influenced by factors such as illumination, environment in practical application so that defects detection, feature extraction and classification
As the hot spot in paper industry.Currently, a variety of paper sheet defect sorting algorithms have been proposed in scholars.It is right that Yuan Hao et al. passes through
Feature selecting is fixed in paper sheet defect gray level image, proposes to classify support vector machines applied to actual paper sheet defect, but
For paper sheet defect gray level image, expressing gradation is single, and collected defect map is led there may be the variation in brightness
Cause classifying quality undesirable.Therefore, Hu Muyi et al. proposes the gray feature according to different paper defect image, double using dynamic
Threshold segmentation paper sheet defect region, extraction defect characteristic are classified, but Threshold segmentation needs that different paper sheet defects is arranged
Different threshold values causes parameter setting difficult.In order to reduce parameter setting complexity, Zhou Qiang et al. is detected using Hough transform
The method of linear feature classifies to paper sheet defect, and this method recognition effect when defect shape is line style is preferable, but uncomfortable
Paper sheet defect for most of non-linearity shapes is classified, and operation complexity is unfavorable for the canbe used on line in later stage.Based on asking above
Topic, Yang Yannan et al. propose the various features value progress Feature-level fusion to paper sheet defect using fuzzy Fusion device, utilize radial direction
Base neural net classifies to paper sheet defect image, expand paper sheet defect identification range, but extract feature it is single and
It is relatively low so as to cause nicety of grading for shallow-layer feature.For this purpose, Luo Lei et al. proposes the LBP (Locial of extraction paper gray level image
Binary Pattern, local binary pattern) feature, carry out defect recognition.But since LBP methods are to paper surface image line
Manage the Preprocessing Algorithm that clarity is more demanding, needs increase complicated.Therefore, the congruent people of Wu one proposes that one kind is based on
The paper sheet defect recognition methods of Krawtchouk moment invariants and wavelet support vector machines, by calculating paper defect image
Krawtchouk moment invariants construct branch to construct the feature vector of paper sheet defect image according to the feature vector of training sample
Vector machine is held, defect classification is carried out to paper sheet defect image, but calculate step complexity early period to reduce recognition speed.In order to improve
Defect identification precision, Zhou Qiang et al. propositions extract paper sheet defect feature using two-dimensional wavelet transformation denoising, singular value decomposition method,
Carry out defect classification identification.But extraction paper sheet defect feature still has the problems such as feature calculation is complicated.And existing paper sheet defect
The problems such as sorting technique describes son and grader selection dependent on traditional characteristic, feature calculation is complicated.
The problem of depending on Feature Descriptor for traditional images sorting technique, the deep learning energy that Hinton et al. is proposed
Enough Layered Learning characteristics of image effectively avoid traditional images sorting technique dependent on the problem of manual features description, in image
It is widely applied in the various fields such as processing and computer vision.As the representative model of deep learning, convolutional Neural
Network (CNN) can learn characteristics of image automatically, be closed to the shape feature of complicated image, textural characteristics, color characteristic and space
It is that feature carries out further feature information extraction and classifies.It usually can get classifying quality more better than traditional characteristic extracting method.
And paper sheet defect image texture is simple, background is single, and paper sheet defect image such as blackspot, hole, pad trace, fold belong to it is small
And a small amount of defect.Conventional method can only extract single low-level image feature, cause classifying quality undesirable.Therefore, how by depth
Convolutional neural networks are applied realizes that quick and precisely classification is a hot issue in paper sheet defect image.
Invention content
In order to overcome the disadvantages of the above prior art, it combines and rolls up the purpose of the present invention is to provide a kind of Multiscale Morphological
The paper sheet defect sorting technique of product neural network is used in convolutional neural networks in paper sheet defect classification, can fast implement paper
The Accurate classification of defect, have the characteristics that method it is simple, it is time-consuming it is short, accuracy of identification is high.
In order to achieve the above object, the technical solution that the present invention takes is：
A kind of paper sheet defect sorting technique of Multiscale Morphological combination convolutional neural networks, includes the following steps：
Step 1：Prepare paper sheet defect image data, and random division training set and test set；
Step 2：Initialization gives Matlab program running parameters, inputs paper sheet defect training set image；
Step 3：Different scale Morphological Gradient operation is carried out to input picture, obtains the Morphological Gradient under multiple scales
Image；
Step 4：To the gradient image Weighted Fusion under different scale, takes and be worth to final multiscale morphological gradient
Image；
Step 5：Multiscale morphological gradient image with paper sheet defect image merge and realizes image enhancement, is rolled up
Product neural network input picture；
Step 6：Enhanced training image is input to convolutional neural networks training pattern；
Step 7：Test set repeats step 3-5 and obtains grad enhancement image, is input on trained model and carries out defect
Feature extraction is simultaneously classified；
Step 8：Prediction label and physical tags are compared, defect classification accuracy rate is calculated.
The step 3 specific implementation process is：
The Morphological Gradient image G (i) of paper defect image f (x, y) is calculated using Mathematical Morphology Method：
(a) the structural elements se of the different scale size of f is calculatediCorresponding gradient image g (i), formula are as follows：
(b) to g (i) and sei-1Erosion operation is carried out, obtains the Morphological Gradient image G (i) of f, formula is as follows：
G (i)=g (i)！sei-1。
The present invention having the beneficial effect that compared with prior art：
For traditional paper defect classification method step is calculated dependent on Feature Descriptor and grader selection, feature description
The problems such as complicated, paper sheet defect image uneven illumination influences nicety of grading, using depth convolutional neural networks to paper sheet defect
Image carries out feature extraction and classifies, and solves the problems, such as that conventional sorting methods exist, in conjunction with paper sheet defect feature, utilizes more rulers
It spends morphological method and grad enhancement is realized to defect image, then utilize convolutional neural networks (CNN) that image can be learnt automatically
The advantage of feature carries out feature extraction to paper sheet defect image and classifies.The present invention classifies with existing traditional characteristic paper sheet defect
Method is compared, and is not required to carry out defect characteristic extraction and feature description for various defects, can be fast implemented the standard of paper sheet defect
Really classification, while solving the problems, such as to influence nicety of grading due to paper sheet defect image irradiation unevenness, with taking, low, precision is high
Advantage.
Description of the drawings
Fig. 1 (a) is paper sheet defect image " blackspot ", " hole ", " pad trace ", " folding line " in present invention experiment；
Fig. 1 (b) is image enhancement of the present invention using four class paper sheet defect image of pretreatment control methods Canny operators pair
As a result；
Fig. 1 (c) is image enhancement of the present invention using four class paper sheet defect image of pretreatment control methods Sobel operators pair
As a result；
Fig. 1 (d) is that the present invention is increased using the image of four class paper sheet defect image of pretreatment control methods Prewitt operators pair
Strong result；
Fig. 1 (e) is the image enhancement result that the present invention utilizes four class paper sheet defect image of the method for the present invention pair.
Fig. 2 (a) is that the present invention is special in the visualization of each convolutional layer of convolutional neural networks using paper sheet defect " hole " image
Sign figure；
Fig. 2 (b) is that the present invention carries out paper sheet defect " hole " image of grad enhancement in convolution god using the method for the present invention
Visualization feature figure through each convolutional layer of network.
Specific implementation mode
The present invention is described in further details with reference to the accompanying drawings and examples.
Morphology operations are carried out to original image with the structural elements of different scale, the gradient image under different scale is weighted
Fusion obtains multiscale morphological gradient image.In order to enhance paper sheet defect contrast, prominent defect Gradient Features and defect side
Multiscale morphological gradient image is weighted with genetic defects image and merges by the information characteristics of edge, realizes that defect image increases
By force, it is input to convolutional neural networks to carry out feature extraction and classify, steps are as follows for specific implementation：
Step 1：Prepare paper sheet defect image data, and random division training set and test set；
Step 2：Initialization, the operating parameter of preset sequence, seiFor the corresponding structural elements of i-th of scale, radius is taken herein
The circular configuration element that size is 1,3,5,7, l=0, m=0, k=length (training set)；
Wherein, l indicates that the prediction label number equal with physical tags, m are constant, and k indicates test set image number, defeated
Enter paper sheet defect training set image；
Step 3：The Morphological Gradient image G (i) of paper defect image f (x, y) is calculated using Mathematical Morphology Method：
(a) the structural elements se of the different scale size of f is calculatediCorresponding gradient image g (i), formula are as follows：
(b) to g (i) and sei-1Erosion operation is carried out, obtains the Morphological Gradient image G (i) of f, formula is as follows：
G (i)=g (i)！sei-1；
Step 4：Mean value is taken to the Morphological Gradient image weighting fusion under different scale, obtains Multiscale Morphological finally
Gradient image MG, formula are as follows：
Wherein, n indicates scale number；
Step 5：Calculate grad enhancement image fMG, formula is as follows：
fMG=(f+MG)；
Step 6：By the corresponding grad enhancement image f of training setMGIt is input to convolutional neural networks training pattern；
Step 7：Test set repeats step 3-5, obtains test set grad enhancement image fMG, it is enterprising to be input to trained model
Row defect characteristic extracts and classifies, and obtains the corresponding prediction label prel of training set datak；
Step 8：The prediction label prel of training setkWith physical tags lkComparison；
If (a) prelk=lk, l=l+1, m=m+1；
(b) otherwise, l is constant, and end condition is：M=k；
(c) defect Classification and Identification rate is calculated, formula is as follows：
Embodiment one：
In order to test the present invention to the validity and superiority of paper sheet defect image classification, emulation experiment is in CPU:
Intel (TM) i7-6700U, 3.3GHz, the hardware environment and MATLAB of memory 16GB, NVIDIA Quadro K620 video cards
It is carried out under the software environment of R2017a.
Utilize three kinds of control methods：Traditional method for extracting paper sheet defect HOG features are classified, and classification results (HOG is obtained
+ SVM), extraction LBP features are classified to obtain classification results (LBP+SVM), four class paper of preprocess method Canny operators pair
It is input to the classification results (Canny+CNN) that CNN models obtain, four class paper sheet defect of Sobel operators pair after defect image enhancing
The classification results (Sobel+CNN) that CNN models obtain, four class paper sheet defect figure of Prewitt operators pair are input to after image enhancement
The four class paper sheet defect figure of classification results (Prewitt+CNN) and the method for the present invention pair that CNN models obtain is input to after image intensifying
Feature extraction is carried out after image intensifying and classify as a result, experimental result with reference to table 1.
Table 1 is to carry out classification and time used comparison knot to paper sheet defect image used in the present invention using each method in table
Fruit.
Table 1
As can be seen from Table 1, it is several that the paper sheet defect image classification accuracy obtained using the method for the present invention is above other
Kind conventional method, it is with the obvious advantage；Although the classification time used is not most short, it is significantly lower than most of other tradition side
Method, state of having the advantage, therefore, it is possible to show that the method for the present invention is a kind of good paper sheet defect sorting technique.
Table 2 is to utilize each method pair in table in the present invention in order to prove the feasibility and practicability of the present invention
Caltech101 image objects identify that data set and KTH-TIPS texture image datasets carry out classification and time used comparison knot
Fruit.
Table 2
Table 2 show each operator to paper sheet defect image carry out grad enhancement after as a result, edge contour contrast is relatively former
Figure is promoted, but it is bigger to Defect Edge directional dependencies.Canny operators and Prewitt operator ratio Sobel operators
Noise removal capability is strong, is easy to smooth out some marginal informations, Sobel operators have humidification to image border, but operator used is
Fixed-size structural element, be not suitable for there are the paper sheet defect image of weak boundary carry out grad enhancement.And the present invention can enhance
Edge contour information and Gradient Features in paper sheet defect image, while retaining the background information and defect target periphery letter of artwork
Breath, takes into account the detection that morphology Mesoscale and microscale structure element is conducive to gradient details, and large-scale structure element is conducive to inhibit to make an uproar
The advantages of sound, effectively prominent defect Gradient Features and marginal information feature, solve paper sheet defect classification and depend on feature description
The problems such as son and grader selection, feature calculation is complicated.
Claims (2)
1. a kind of paper sheet defect sorting technique of Multiscale Morphological combination convolutional neural networks, which is characterized in that including as follows
Step：
Step 1：Prepare paper sheet defect image data, and random division training set and test set；
Step 2：Initialization gives Matlab program running parameters, inputs paper sheet defect training set image；
Step 3：Different scale Morphological Gradient operation is carried out to input picture, obtains the Morphological Gradient figure under multiple scales
Picture；
Step 4：To the gradient image Weighted Fusion under different scale, takes and be worth to final multiscale morphological gradient figure
Picture；
Step 5：Multiscale morphological gradient image with paper sheet defect image merge and realizes image enhancement, obtains convolution god
Through network inputs image；
Step 6：Enhanced training image is input to convolutional neural networks training pattern；
Step 7：Test set repeats step 3-5 and obtains grad enhancement image, is input on trained model and carries out defect characteristic
It extracts and classifies；
Step 8：Prediction label and physical tags are compared, defect classification accuracy rate is calculated.
2. a kind of paper sheet defect sorting technique of Multiscale Morphological combination convolutional neural networks according to claim 1,
It is characterized in that, the step 3 specific implementation process is：
The Morphological Gradient image G (i) of paper defect image f (x, y) is calculated using Mathematical Morphology Method：
(a) the structural elements se of the different scale size of f is calculatediCorresponding gradient image g (i), formula are as follows：
(b) to g (i) and sei-1Erosion operation is carried out, obtains the Morphological Gradient image G (i) of f, formula is as follows：
G (i)=g (i)！sei-1。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810260457.4A CN108427969A (en) | 2018-03-27 | 2018-03-27 | A kind of paper sheet defect sorting technique of Multiscale Morphological combination convolutional neural networks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810260457.4A CN108427969A (en) | 2018-03-27 | 2018-03-27 | A kind of paper sheet defect sorting technique of Multiscale Morphological combination convolutional neural networks |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108427969A true CN108427969A (en) | 2018-08-21 |
Family
ID=63159286
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810260457.4A Pending CN108427969A (en) | 2018-03-27 | 2018-03-27 | A kind of paper sheet defect sorting technique of Multiscale Morphological combination convolutional neural networks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108427969A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109389110A (en) * | 2018-10-11 | 2019-02-26 | 北京奇艺世纪科技有限公司 | A kind of area determination method and device |
| CN109447080A (en) * | 2018-11-12 | 2019-03-08 | 北京奇艺世纪科技有限公司 | A kind of character identifying method and device |
| CN110473178A (en) * | 2019-07-30 | 2019-11-19 | 上海深视信息科技有限公司 | A kind of open defect detection method and system based on multiple light courcess fusion |
| CN110648322A (en) * | 2019-09-25 | 2020-01-03 | 杭州智团信息技术有限公司 | Method and system for detecting abnormal cervical cells |
| CN110738609A (en) * | 2019-09-11 | 2020-01-31 | 北京大学 | method and device for removing image moire |
| CN111444866A (en) * | 2020-03-31 | 2020-07-24 | 武汉科技大学 | Paper making cause inspection and identification method based on deep learning |
| CN112557406A (en) * | 2021-02-19 | 2021-03-26 | 浙江大胜达包装股份有限公司 | Intelligent inspection method and system for paper product production quality |
| CN114120453A (en) * | 2021-11-29 | 2022-03-01 | 北京百度网讯科技有限公司 | Living body detection method and device, electronic equipment and storage medium |
| CN118014989A (en) * | 2024-04-02 | 2024-05-10 | 东莞市时实电子有限公司 | Intelligent detection method for surface defects of power adapter |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101900673A (en) * | 2010-06-02 | 2010-12-01 | 北京林业大学 | Method of on-line non-destructive testing paper performance |
| CN102095731A (en) * | 2010-12-02 | 2011-06-15 | 山东轻工业学院 | System and method for recognizing different defect types in paper defect visual detection |
| CN102542288A (en) * | 2011-11-28 | 2012-07-04 | 北京航空航天大学 | Construction and merging classification method for high spectrum data multi-characteristic space |
| CN103530621A (en) * | 2013-11-04 | 2014-01-22 | 中国矿业大学(北京) | Coal and rock image identification method based on back propagation (BP) neural network |
| US20160187199A1 (en) * | 2014-08-26 | 2016-06-30 | Digimarc Corporation | Sensor-synchronized spectrally-structured-light imaging |
| CN106355579A (en) * | 2016-09-14 | 2017-01-25 | 南京理工大学 | Defect detecting method of cigarette carton surface wrinkles |
| CN107239775A (en) * | 2017-07-28 | 2017-10-10 | 湖南大学 | Terrain classification method and device |
-
2018
- 2018-03-27 CN CN201810260457.4A patent/CN108427969A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101900673A (en) * | 2010-06-02 | 2010-12-01 | 北京林业大学 | Method of on-line non-destructive testing paper performance |
| CN102095731A (en) * | 2010-12-02 | 2011-06-15 | 山东轻工业学院 | System and method for recognizing different defect types in paper defect visual detection |
| CN102542288A (en) * | 2011-11-28 | 2012-07-04 | 北京航空航天大学 | Construction and merging classification method for high spectrum data multi-characteristic space |
| CN103530621A (en) * | 2013-11-04 | 2014-01-22 | 中国矿业大学(北京) | Coal and rock image identification method based on back propagation (BP) neural network |
| US20160187199A1 (en) * | 2014-08-26 | 2016-06-30 | Digimarc Corporation | Sensor-synchronized spectrally-structured-light imaging |
| CN106355579A (en) * | 2016-09-14 | 2017-01-25 | 南京理工大学 | Defect detecting method of cigarette carton surface wrinkles |
| CN107239775A (en) * | 2017-07-28 | 2017-10-10 | 湖南大学 | Terrain classification method and device |
Non-Patent Citations (5)
| Title |
|---|
| DEMIN WANG: "A multiscale gradient algorithm for image segmentation using watershelds", 《PATTERN RECOGNITION》 * |
| JESLY JAMES等: "Automated Acute Myelogenous Lukemia Detection in Blood Microscopic Image", 《INTERNATIONAL JOURNAL OF SCIENCE AND RESEARCH (IJSR)》 * |
| 梁英波 等: "基于多尺度数学形态学梯度的图像检测", 《周口师范学院学报》 * |
| 罗磊 等: "基于图像分析的纸张表面缺陷检测系统的研究与实现", 《西安理工大学学报》 * |
| 金艳 等: "基于神经网络和形态学的钢表面缺陷识别", 《机床与液压》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109389110B (en) * | 2018-10-11 | 2021-03-19 | 北京奇艺世纪科技有限公司 | Region determination method and device |
| CN109389110A (en) * | 2018-10-11 | 2019-02-26 | 北京奇艺世纪科技有限公司 | A kind of area determination method and device |
| CN109447080A (en) * | 2018-11-12 | 2019-03-08 | 北京奇艺世纪科技有限公司 | A kind of character identifying method and device |
| CN109447080B (en) * | 2018-11-12 | 2020-04-17 | 北京奇艺世纪科技有限公司 | Character recognition method and device |
| CN110473178A (en) * | 2019-07-30 | 2019-11-19 | 上海深视信息科技有限公司 | A kind of open defect detection method and system based on multiple light courcess fusion |
| CN110738609B (en) * | 2019-09-11 | 2022-05-06 | 北京大学 | Method and device for removing image moire |
| CN110738609A (en) * | 2019-09-11 | 2020-01-31 | 北京大学 | method and device for removing image moire |
| CN110648322A (en) * | 2019-09-25 | 2020-01-03 | 杭州智团信息技术有限公司 | Method and system for detecting abnormal cervical cells |
| CN110648322B (en) * | 2019-09-25 | 2023-08-15 | 杭州智团信息技术有限公司 | Cervical abnormal cell detection method and system |
| CN111444866B (en) * | 2020-03-31 | 2023-05-30 | 武汉科技大学 | Paper making cause inspection and identification method based on deep learning |
| CN111444866A (en) * | 2020-03-31 | 2020-07-24 | 武汉科技大学 | Paper making cause inspection and identification method based on deep learning |
| CN112557406B (en) * | 2021-02-19 | 2021-06-29 | 浙江大胜达包装股份有限公司 | Intelligent inspection method and system for paper product production quality |
| CN112557406A (en) * | 2021-02-19 | 2021-03-26 | 浙江大胜达包装股份有限公司 | Intelligent inspection method and system for paper product production quality |
| CN114120453A (en) * | 2021-11-29 | 2022-03-01 | 北京百度网讯科技有限公司 | Living body detection method and device, electronic equipment and storage medium |
| CN118014989A (en) * | 2024-04-02 | 2024-05-10 | 东莞市时实电子有限公司 | Intelligent detection method for surface defects of power adapter |
| CN118014989B (en) * | 2024-04-02 | 2024-06-28 | 东莞市时实电子有限公司 | Intelligent detection method for surface defects of power adapter |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108427969A (en) | A kind of paper sheet defect sorting technique of Multiscale Morphological combination convolutional neural networks | |
| Arulmurugan et al. | Region-based seed point cell segmentation and detection for biomedical image analysis | |
| CN107610087B (en) | Tongue coating automatic segmentation method based on deep learning | |
| CN113592845A (en) | Defect detection method and device for battery coating and storage medium | |
| Plissiti et al. | Combining shape, texture and intensity features for cell nuclei extraction in Pap smear images | |
| CN103996018B (en) | Face identification method based on 4DLBP | |
| CN106780498A (en) | Based on point depth convolutional network epithelium and matrix organization's automatic division method pixel-by-pixel | |
| CN106778687A (en) | Method for viewing points detecting based on local evaluation and global optimization | |
| Xiao et al. | Salient object detection based on eye tracking data | |
| Yu et al. | Aggregating deep convolutional features for melanoma recognition in dermoscopy images | |
| CN107330883A (en) | A kind of medical image lesion region positioning and sorting technique | |
| CN106023151A (en) | Traditional Chinese medicine tongue manifestation object detection method in open environment | |
| Ouyang et al. | The research of the strawberry disease identification based on image processing and pattern recognition | |
| Liang et al. | Genetic programming for evolving figure-ground segmentors from multiple features | |
| Li et al. | Research on a product quality monitoring method based on multi scale PP-YOLO | |
| CN107133562B (en) | Gesture recognition method based on extreme learning machine | |
| US20220319208A1 (en) | Method and apparatus for obtaining feature of duct tissue based on computer vision, and intelligent microscope | |
| WO2024016812A1 (en) | Microscopic image processing method and apparatus, computer device, and storage medium | |
| CN105718947B (en) | Lung Cancer Images sophisticated category method based on LBP and wavelet moment fusion feature | |
| Liu et al. | Extracting lungs from CT images via deep convolutional neural network based segmentation and two-pass contour refinement | |
| Urdal et al. | Prognostic prediction of histopathological images by local binary patterns and RUSBoost | |
| CN111783885A (en) | Millimeter wave image quality classification model construction method based on local enhancement | |
| Pham et al. | CNN-based character recognition for license plate recognition system | |
| Zhang et al. | Artifact detection in endoscopic video with deep convolutional neural networks | |
| CN103366183B (en) | Nonparametric automatic detection method of focal niduses |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180821 |
|
| RJ01 | Rejection of invention patent application after publication |