CN104048966B - The detection of a kind of fabric defect based on big law and sorting technique - Google Patents

The detection of a kind of fabric defect based on big law and sorting technique Download PDF

Info

Publication number
CN104048966B
CN104048966B CN201410200849.3A CN201410200849A CN104048966B CN 104048966 B CN104048966 B CN 104048966B CN 201410200849 A CN201410200849 A CN 201410200849A CN 104048966 B CN104048966 B CN 104048966B
Authority
CN
China
Prior art keywords
fault
class
image
carried
detection
Prior art date
Application number
CN201410200849.3A
Other languages
Chinese (zh)
Other versions
CN104048966A (en
Inventor
石纪军
刘华山
陈霞
胡江浩
唐雅琴
Original Assignee
东华大学
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Priority to CN201410095814 priority Critical
Priority to CN201410095814.8 priority
Priority to CN2014100958148 priority
Application filed by 东华大学 filed Critical 东华大学
Priority to CN201410200849.3A priority patent/CN104048966B/en
Publication of CN104048966A publication Critical patent/CN104048966A/en
Application granted granted Critical
Publication of CN104048966B publication Critical patent/CN104048966B/en

Links

Abstract

The invention discloses the detection of a kind of fabric defect based on big law and sorting technique, slave computer carries out fabric defect detection to cloth cover image information, the defect detection data that host computer is uploaded according to slave computer, classifies fault.Wherein, slave computer defect detection includes: be filtered the image information gathered, interpolation, variance sample pretreatment, the segmentation of big law image and binaryzation, and cavity is filled, fritter processes and operates, bianry image channel connection processes, and extracts each defect regions information and preserves.Fault classification is included by host computer: fault is divided into region class and non-area class fault, through class fault and latitude class fault, dark class fault and bright class fault.The identification of fault is given slave computer and is completed by the method, the classification of fault and man-machine interaction are given host computer and are completed, reasonably task divides and makes the classification of the identification to fabric defect more efficient, meet the requirement of online defect detection in real time, it is possible to apply in collisional transfer detection real-time on loom line.

Description

The detection of a kind of fabric defect based on big law and sorting technique

Technical field

The present invention relates to the detection of a kind of fabric defect based on big law and sorting technique, be applied to the quality testing of textile, belong to technical field of image processing.

Background technology

In textile production, grading and sell in each link, the detection of fabric defects and the key link that identification is that fabric quality controls, tool is of great significance.Easily there is missing inspection and flase drop in traditional artificial open hole detection, thus fabric defects detects the heat subject becoming recent domestic research automatically.Domestic scholars has been proposed for some fabric defects automatic testing methods at present.

Application No. 201310119051.1, Publication No. CN103207186A, the Chinese patent of invention entitled " automatic cloth inspecting machine defect detection recognition methods and system thereof ", provide a kind of automatic cloth inspecting machine defect detection recognition methods, first pass through charge-coupled image sensor (CCD) pickup wire array camera and gather cloth picture, then defects identification algorithm is utilized to carry out accurately distinguishing of normal cloth and Fabric Defects Inspection, the picture being determined as fabric defects by machine is initially stored in the hard disk of cloth inspecting machine industrial computer, then the information such as fault picture and defect position are transferred to remote terminal in office via network by this main frame, fault picture makes fault have preferably resolution of eye by algorithm for image enhancement, carry out the judgement of fault classification again according to the fault picture after this process by perching workman.Although this invention improves perching efficiency to a certain extent, but it uses CCD line-scan digital camera to gather picture, on the one hand CCD compares contact-type image sensor (CIS), the optical system of CCD camera needs open scene, so that take bigger space, it is unfavorable for the renovation and utilization to existing equipment, adds device design complexities and the light source impact on defects identification rate cannot be eliminated;Additionally linear array CCD camera especially colored thread array CCD camera is expensive so that the manufacturing cost of equipment rises, and is unfavorable for the popularization and application of patented technology.In fault sorting technique, this invention still depends on manually, bans artificial perching the most completely.

Application No. 201010222997.7, Publication No. CN101957326A, the Chinese patent of invention entitled " a kind of multispectral monitoring method of textile surface quality and device ", the CIS array being made up of contact-type image sensor, multispectral LED break-make control, image data acquiring, electric machine controller, master controller, touch screen or keyboard, display, ink jet type indicator, communication interface and printer form.Master controller is by the motor rotation in motor controller controls cloth inspecting machine, textile cloth inspecting machine power set drive under through CIS array, motor often takes a step forward, just carry out the image data acquiring under the most various spectral illumination, view data is identified by master controller, the information positions such as fault, different fibre, classification, grade are preserved, and is sent to Surveillance center by communication interface.This invention is detecting system under fabric defects line, needs cloth to draw winding apparatus, compares on-line checking and adds detection operation, and cost of labor increases.This invention device is designed without utilizing existing loom, and uses multispectral detection, and from the point of view of the user for single cloth detection demand, this invention device is more complicated, and relatively costly, efficiency of algorithm is low.

Summary of the invention

The technical problem to be solved in the present invention be to provide a kind of in real time, effectively, may replace the fabric defect detection of artificial naked eyes perching and know method for distinguishing.

In order to solve above-mentioned technical problem, the technical scheme is that offer a kind of fabric defect based on big law detection and sorting technique.It is characterized in that: detection and the classification of fault are carried out by slave computer and host computer two parts: slave computer gathers cloth cover image information by imageing sensor, then the detection of fabric defect is carried out, the defect detection data that host computer is uploaded according to slave computer, classify to fault;Wherein, slave computer defect detection step is as follows: the cloth cover image information gathered is filtered by (1), the sampling pretreatment of interpolation, variance;(2) segmentation of big law image and binaryzation are carried out;(3) carry out cavity filling, fritter processes operation;(4) bianry image channel connection process is carried out;(5) extract each defect regions information and preserve;The step that fault is classified by host computer is as follows: fault is divided into region class and non-area class fault according to fault ratio of semi-minor axis length by (1);(2) according to fault spindle tilt, fault is divided into through class fault and latitude class fault;(3) according to the overall gray value of fault, fault is divided into dark class fault and bright class fault.

Preferably, described fabric defect detection method specifically comprises the following steps that

Step 1: read in digital picture initial data and be stored in variable I;

Step 2: I is carried out a mean filter and processes;

Step 3: I is carried out bilinear interpolation process;

Step 4: I is carried out a variance sampling processing;

Step 5: I is carried out bilinear interpolation process;

Step 6: I is carried out Otsu big law image dividing processing and obtains image segmentation threshold T;

Step 7: I is carried out binary conversion treatment by the big law of Otsu;

Step 8: I is carried out cavity filling and processes;

Step 9: I is carried out fritter and processes operation;

Step 10: I is carried out bianry image eight channel connection and processes;

Step 11: extract the image attributes of I and be stored in many structure variables stats;

Step 12: extract fault number Num1 in variable stats;

Preferably, described fabric defect sorting technique specifically comprises the following steps that

Step a: set the Y3Y2Y1Y0 four figures variable as fault result of determination, it is judged that whether Num1 less than or equal to 0, if Num1 is less than or equal to 0, then Y3=0 judge this textile image as indefectible image, otherwise Y3=1 jumps to step b;

Step b: fault and region are had the Elliptical Ratio of identical standard second-order moment around mean relatively, judge that region has the oval long axis length of identical standard second-order moment around mean and region and has the ratio of oval minor axis length of identical standard second-order moment around mean whether less than fault ratio of semi-minor axis length and more than the inverse of fault ratio of semi-minor axis length, if, then Y2=1 judges that this image fault as region class fault and jumps to step d, and otherwise Y2=0 judges that this image fault as non-area class fault and jumps to step c;

Step c: judge that whether the fault main shaft absolute value with the angle Q of x-axis is more than spindle inclination An, if the absolute value of Q is more than spindle inclination An, then Y1=1 judge this image fault as through class fault and jump to step d, otherwise Y1=0 judges that this image fault as latitude class fault and jumps to step d;

Step d: judge whether fault overall gray value P is more than fault average gray H, if P value is more than H-number, then Y0=1 judges that this image fault as bright class fault and jumps to step e, and otherwise Y0=0 judges that this image fault as dark class fault and jumps to step e;

Step e: table 1 is classification results State-output table, wherein X represents that this position is without judgement;

Table 1 classification results State-output table

Y3Y2Y1Y0 Output 0XXX Without fault 1000 Dark latitude class fault 1001 Bright pick class fault 1010 Latent menstruation class fault 1011 Bright through class fault 11X0 Region class hidden flaws point 11X1 The bright fault of region class

Result of determination Y3Y2Y1Y0 is contrasted table 1, finally judge and export seven kinds of fault types: without fault, bright through class fault, latent menstruation class fault, bright pick class fault, dark latitude class fault, bright area class fault, dark areas class fault, fault counting number value of simultaneously having classified adds 1, if fault of having classified number is less than or equal to the fault number existed, jumping to step b and continue to classify next fault, otherwise classification terminates.

Preferably, in described fabric defect detection method step 2, mean filter processing template scope is: 8*8 window is to 10*10 window.

The method that the present invention provides overcomes the deficiencies in the prior art, human eye detection is replaced by machine vision, host computer and slave computer are carried out reasonable task division, there is feature quick, correct, efficient, that real-time is good, flase drop, loss can be greatly reduced, while improving productivity ratio, can effectively reduce waste and the loss of human and material resources, financial resources and the energy that waster causes, reduce the volume of equipment simultaneously, reduce manufacturing cost.

Accompanying drawing explanation

The fabric defect based on big law that Fig. 1 provides for the present invention detects and sorting technique general flow chart;

Fig. 2 is slave computer exemplary plot;

Fig. 3 is cloth cover image acquisition and Image semantic classification and fault extracts flow chart;

Fig. 4 is fabric defect sorting technique flow chart.

Detailed description of the invention

For making the present invention become apparent, hereby with a preferred embodiment, and accompanying drawing is coordinated to be described in detail below.

The fabric defect based on big law that Fig. 1 provides for the present invention detects and sorting technique general flow chart, the identification of cloth cover image information collecting and fault is given slave computer and is completed by described fabric defect based on big law detection and sorting technique, the classification of fault and man-machine interaction are given host computer and are completed, reasonably task divides and makes the classification of the identification to fabric defect more efficient, meets the requirement of online defect detection in real time.

In conjunction with Fig. 2, slave computer gathers cloth cover image information by CIS, digital signal processor (DSP) image information is carried out fabric defect identification, and recognition result is transferred to host computer, host computer be responsible for classification and the man-machine interaction of fault.Wherein CIS, DSP are all connected with field programmable gate array (FPGA) module and A/D module, and FPGA module is responsible for the sequencing contro of CIS image acquisition, and A/D module then carries out analog digital conversion to the analogue signal of CIS output.

DSP carries out defects identification based on big law algorithm to cloth cover image information, and recognition result is transferred to host computer, and host computer is responsible for classifying fault and the display of classification results.

Fig. 3 is that in the present embodiment, cloth cover image acquisition and Image semantic classification extract flow chart with fault, specifically comprises the following steps that

Step 1: gather cloth cover image information;

Step 2: read in original image data and be stored in variable I;

Step 3: I is carried out a mean filter and processes, make image blurringization;Mean filter processing template scope is: 8*8 window is to 10*10 window;

Step 4: I is carried out bilinear interpolation process, after Fuzzy processing image, image size can occur necessarily to change, and uses bilinear interpolation to recover original image size;

Step 5: I carries out a variance sampling processing, strengthens cloth image fault information;

Step 6: I carries out bilinear interpolation process, after strengthening fault information, image size changes, then uses a bilinear interpolation to recover original image size;

Step 7: I carries out Otsu big law image dividing processing, obtains image binaryzation segmentation threshold T by contrast cloth image fault information;

Step 8: I is carried out binary conversion treatment by the big law of Otsu;

Step 9: I carries out cavity filling and processes, eliminates the internal cavity existed of fault;

Step 10: I is carried out fritter and processes operation, eliminate the fault less than predetermined area;

Step 11: I is carried out bianry image eight channel connection and processes, facilitate the attributes extraction of image fault;

Step 12: extract the image attributes of I and be stored in many structure variables stats;

Step 13: extract fault number Num1 in variable stats.

Fig. 4 is the method flow diagram that the fabric defects image collected carries out in the present embodiment fabric defects classification, and its specific embodiments step is as follows:

Step a: judge Num1 whether less than or equal to 0, if Num1 is less than or equal to 0, then Y3=0 judge this textile image as indefectible image, otherwise Y3=1 jumps to step b;

Step b: judge that fault and region have the oval long axis length of identical standard second-order moment around mean and region and have the ratio of oval minor axis length of identical standard second-order moment around mean whether less than fault ratio of semi-minor axis length and more than the inverse of fault ratio of semi-minor axis length, if in the range of, Y2=1 judges that this image fault as region class fault and jumps to step d, and otherwise Y2=0 judges that this image fault as non-area class fault and jumps to step c;

Step c: judge that whether the fault main shaft absolute value with the angle Q of x-axis is more than spindle inclination An, if | Q | value is more than spindle inclination An, then Y1=1 judge this image fault as through class fault and jump to step d, otherwise Y1=0 judges that this image fault as latitude class fault and jumps to step d;

Step d: judge fault overall gray value P whether more than fault average gray H, if P value is more than H-number, then Y0=1 judges that this image fault as bright class fault and jumps to step e, otherwise Y0=0 judge this image fault as dark class fault, jump to step e.

Step e: table 1 is classification results State-output table, wherein X represents that this position is without judgement;

Table 1 classification results State-output table

Result of determination Y3Y2Y1Y0 is contrasted table 1, finally judge and export seven kinds of fault types: without fault, bright through class fault, latent menstruation class fault, bright pick class fault, dark latitude class fault, bright area class fault, dark areas class fault, fault counting number value of simultaneously having classified adds 1, if fault of having classified number is less than or equal to the fault number existed, jumping to step (b) and continue to classify next fault, otherwise classification terminates.

The method that the present invention provides overcomes in existing automatic cloth inspecting machine defect detection identification system that CCD line-scan digital camera volume is big, cost is high, system and device is complicated, the deficiency that efficiency of algorithm is low, human eye detection is replaced by machine vision, there is feature quick, correct, efficient, flase drop, loss can be greatly reduced, while improving productivity ratio, can effectively reduce waste and the loss of human and material resources, financial resources and the energy that waster causes;Use CIS imageing sensor to gather view data, eliminate the requirement to light source of the CCD line-scan digital camera, reduce the volume of equipment simultaneously and reduce manufacturing cost, the popularization and application of product the most of the present invention;The present invention is detection probe to be arranged on loom, take full advantage of the motor drawing device of loom, inspection one is knitted in realization, on the one hand cloth examination device cost is saved, on the other hand the perching time is saved, decrease the cost of labor of weaving mill, have broad application prospects and huge market economy benefit.

Claims (3)

1. fabric defect based on a big law detection and sorting technique, it is characterized in that: detection and the classification of fault are carried out by slave computer and host computer two parts, slave computer gathers cloth cover image information by imageing sensor, then the detection of fabric defect is carried out, the defect detection data that host computer is uploaded according to slave computer, classify to fault;Wherein, slave computer defect detection step is as follows: the cloth cover image information gathered is filtered by (1), the sampling pretreatment of interpolation, variance;(2) segmentation of big law image and binaryzation are carried out;(3) carry out cavity filling, fritter processes operation;(4) bianry image channel connection process is carried out;(5) extract each defect regions information and preserve;The step that fault is classified by host computer is as follows: fault is divided into region class and non-area class fault according to fault ratio of semi-minor axis length by (1);(2) according to fault spindle tilt, fault is divided into through class fault and latitude class fault;(3) according to the overall gray value of fault, fault is divided into dark class fault and bright class fault;
Described fabric defect sorting technique specifically comprises the following steps that
Step a: set the Y3Y2Y1Y0 four figures variable as fault result of determination, Num1 is fault number, it is judged that whether Num1 less than or equal to 0, if Num1 is less than or equal to 0, then Y3=0 judge this textile image as indefectible image, otherwise Y3=1 jumps to step b;
Step b: fault and region are had the Elliptical Ratio of identical standard second-order moment around mean relatively, judge that region has the oval long axis length of identical standard second-order moment around mean and region and has the ratio of oval minor axis length of identical standard second-order moment around mean whether less than fault ratio of semi-minor axis length and more than the inverse of fault ratio of semi-minor axis length, if, then Y2=1 judges that this image fault as region class fault and jumps to step d, and otherwise Y2=0 judges that this image fault as non-area class fault and jumps to step c;
Step c: judge that whether the fault main shaft absolute value with the angle Q of x-axis is more than spindle inclination An, if the absolute value of Q is more than spindle inclination An, then Y1=1 judge this image fault as through class fault and jump to step d, otherwise Y1=0 judges that this image fault as latitude class fault and jumps to step d;
Step d: judge whether fault overall gray value P is more than fault average gray H, if P value is more than H-number, then Y0=1 judges that this image fault as bright class fault and jumps to step e, and otherwise Y0=0 judges that this image fault as dark class fault and jumps to step e;
Step e: table 1 is classification results State-output table, wherein X represents that this position is without judgement;
Table 1 classification results State-output table
Result of determination Y3Y2Y1Y0 is contrasted table 1, finally judge and export seven kinds of fault types: without fault, bright through class fault, latent menstruation class fault, bright pick class fault, dark latitude class fault, bright area class fault, dark areas class fault, fault counting number value of simultaneously having classified adds 1, if fault of having classified number is less than or equal to the fault number existed, jumping to step b and continue to classify next fault, otherwise classification terminates.
The detection of a kind of fabric defect based on big law and sorting technique, it is characterised in that: described fabric defect detection method specifically comprises the following steps that
Step 1: read in digital picture initial data and be stored in variable I;
Step 2: I is carried out a mean filter and processes;
Step 3: I is carried out bilinear interpolation process;
Step 4: I is carried out a variance sampling processing;
Step 5: I is carried out bilinear interpolation process;
Step 6: I is carried out Otsu big law image dividing processing and obtains image segmentation threshold T;
Step 7: I is carried out binary conversion treatment by the big law of Otsu;
Step 8: I is carried out cavity filling and processes;
Step 9: I is carried out fritter and processes operation;
Step 10: I is carried out bianry image eight channel connection and processes;
Step 11: extract the image attributes of I and be stored in many structure variables stats;
Step 12: extract fault number Num1 in variable stats.
The detection of a kind of fabric defect based on big law and sorting technique, it is characterised in that: in described fabric defect detection method step 2, mean filter processing template scope is: 8*8 window is to 10*10 window.
CN201410200849.3A 2014-03-14 2014-05-13 The detection of a kind of fabric defect based on big law and sorting technique CN104048966B (en)

Priority Applications (4)

Application Number Priority Date Filing Date Title
CN201410095814 2014-03-14
CN201410095814.8 2014-03-14
CN2014100958148 2014-03-14
CN201410200849.3A CN104048966B (en) 2014-03-14 2014-05-13 The detection of a kind of fabric defect based on big law and sorting technique

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410200849.3A CN104048966B (en) 2014-03-14 2014-05-13 The detection of a kind of fabric defect based on big law and sorting technique

Publications (2)

Publication Number Publication Date
CN104048966A CN104048966A (en) 2014-09-17
CN104048966B true CN104048966B (en) 2016-08-03

Family

ID=51502084

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410200849.3A CN104048966B (en) 2014-03-14 2014-05-13 The detection of a kind of fabric defect based on big law and sorting technique

Country Status (1)

Country Link
CN (1) CN104048966B (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105699386B (en) * 2016-02-29 2019-02-26 东华大学 A kind of automatic cloth inspection labeling method using contact-type image sensor
CN105738376B (en) * 2016-02-29 2018-07-17 东华大学 A kind of automatic cloth inspecting machine using contact-type image sensor
CN105784712B (en) * 2016-02-29 2018-07-17 东华大学 A kind of automatic cloth inspection method using contact-type image sensor
CN107169961A (en) * 2017-05-15 2017-09-15 中烟追溯(北京)科技有限公司 A kind of cigarette sorting detecting system and method based on CIS IMAQs
CN107220649A (en) * 2017-05-27 2017-09-29 江苏理工学院 A kind of plain color cloth defects detection and sorting technique
CN107248154A (en) * 2017-05-27 2017-10-13 江苏理工学院 A kind of cloth aberration real-time on-line detecting method

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5815198A (en) * 1996-05-31 1998-09-29 Vachtsevanos; George J. Method and apparatus for analyzing an image to detect and identify defects
CN101866427A (en) * 2010-07-06 2010-10-20 西安电子科技大学 Method for detecting and classifying fabric defects
CN102175692A (en) * 2011-03-17 2011-09-07 嘉兴学院 System and method for detecting defects of fabric gray cloth quickly
CN102879401A (en) * 2012-09-07 2013-01-16 西安工程大学 Method for automatically detecting and classifying textile flaws based on pattern recognition and image processing

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5815198A (en) * 1996-05-31 1998-09-29 Vachtsevanos; George J. Method and apparatus for analyzing an image to detect and identify defects
CN101866427A (en) * 2010-07-06 2010-10-20 西安电子科技大学 Method for detecting and classifying fabric defects
CN102175692A (en) * 2011-03-17 2011-09-07 嘉兴学院 System and method for detecting defects of fabric gray cloth quickly
CN102879401A (en) * 2012-09-07 2013-01-16 西安工程大学 Method for automatically detecting and classifying textile flaws based on pattern recognition and image processing

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
基于Blob 算法的织物疵点检测算法的研究;石美红等;《现代电子技术》;20101231(第24期);29-32 *
基于下采样的改进的织物图像预处理方法;郭攀峰;《工业控制计算机》;20110131;第24卷(第1期);80-83 *
工业流水线织物疵点检测及分类算法研究;王永灿;《中国优秀硕士学位论文全文数据库 信息科技辑》;20090615(第6期);正文第8,14,43-44页 *
织物疵点在线检测理论与应用的研究;李建福;《中国优秀硕士学位论文全文数据库 信息科技辑》;20090315(第3期);正文第34-35,36-37,40-41页 *
织物纹理的表征和自动识别的研究;姚芳;《中国优秀硕士学位论文全文数据库 信息科技辑》;20100815(第8期);正文第40,48页 *

Also Published As

Publication number Publication date
CN104048966A (en) 2014-09-17

Similar Documents

Publication Publication Date Title
CN107392896B (en) A kind of Wood Defects Testing method and system based on deep learning
CN103913468B (en) Many defects of vision checkout equipment and the method for large-scale LCD glass substrate on production line
CN101996405B (en) Method and device for rapidly detecting and classifying defects of glass image
CN102879401B (en) Method for automatically detecting and classifying textile flaws based on pattern recognition and image processing
CN103090804B (en) Automatic detection system and detection method of finished product magnet ring image
CN104331521B (en) Transformer anomaly identification method based on image procossing
CN104360501B (en) A kind of LCD screen defective vision detection method and device
CN104077577A (en) Trademark detection method based on convolutional neural network
CN102253048B (en) Machine vision detection method and system for detection of various products
CN105044122B (en) A kind of copper piece surface defect visible detection method based on semi-supervised learning model
CN107123114A (en) A kind of cloth defect inspection method and device based on machine learning
Ghazvini et al. Defect detection of tiles using 2D-wavelet transform and statistical features
CN102621077B (en) Corn seed purity nondestructive detection method based on hyper-spectral reflection image collecting system
CN101572803B (en) Customizable automatic tracking system based on video monitoring
CN103234976A (en) Warp knitting machine cloth flaw on-line visual inspection method based on Gabor transformation
CN104992449A (en) Information identification and surface defect on-line detection method based on machine visual sense
CN201434842Y (en) Automatic chip appearance defect detection system
CN104764744B (en) Visual inspection device and method for inspecting freshness of poultry eggs
CN202002894U (en) Quick online paper flaw detecting system based on machine vision
CN104361314B (en) Based on infrared and transformer localization method and device of visual image fusion
CN102297867B (en) Detection system for assembly quality of wiring harness
CN202362253U (en) On-line visual detection system for coating quality of battery pole piece
Tong et al. A new image-based method for concrete bridge bottom crack detection
CN106875373B (en) Mobile phone screen MURA defect detection method based on convolutional neural network pruning algorithm
CN102974551A (en) Machine vision-based method for detecting and sorting polycrystalline silicon solar energy

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20160803

Termination date: 20190513