CN106067020A - The system and method for quick obtaining effective image under real-time scene - Google Patents
The system and method for quick obtaining effective image under real-time scene Download PDFInfo
- Publication number
- CN106067020A CN106067020A CN201610390684.XA CN201610390684A CN106067020A CN 106067020 A CN106067020 A CN 106067020A CN 201610390684 A CN201610390684 A CN 201610390684A CN 106067020 A CN106067020 A CN 106067020A
- Authority
- CN
- China
- Prior art keywords
- image
- module
- real
- under
- classification
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/10—Image acquisition
-
- 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
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
The invention discloses the system and method for quick obtaining effective image under a kind of real-time scene, in image acquisition process under this scene, image quality assessment module uses the image quality evaluation mathematical model image to collecting to carry out Fast Evaluation, image classification module uses described image classification mathematical model to judge the image quality evaluation result of input, if it is determined that be effective image, then stop gathering, draw the result successfully obtaining effective image, and image is inputted aftertreatment systems;If it is determined that be invalid image, then continue to gather, and reappraise, until obtaining effective image.The present invention can be on the premise of keeping image acquisition real-time as far as possible, it is ensured that the image finally entering subsequent treatment link is all effective image.
Description
(1) technical field
The present invention relates to field of computer technology, particularly to a kind of digital image collection system and method.
(2) background technology
Gather the scene that digital picture carries out processing, such as, use real-time decoding equipment on bar code decoding, streamline
Product defects detection, make a dash across the red light to take pictures and test the speed and take pictures, it is common that first use image capture device gather image, rear right
The image gathered processes, if processing successfully, then stops;If processing unsuccessfully, then Resurvey image, again process, so
Circulation is until just stopping when processing successfully.And under real-time scene, between image capture device and target, there is relative motion, work as phase
To motion is very fast or picture-taken frequency is relatively low, when the image procossing failure gathered is needed to re-start image acquisition,
Target may have been moved off form, and cause again cannot be to its Resurvey.As can be seen here, under real-time scene, for
The subsequent treatment of the invalid image collected, actually consumes the substantial amounts of real-time process time.
(2) summary of the invention
In order to overcome above-mentioned deficiency, the present invention is " under real-time scene, to assess and to judge spent by an invalid image
Time is much smaller than the time spent by this image of subsequent treatment " it is foundation, it is provided that and under a kind of real-time scene, quick obtaining is effective
The system and method for image.
The technical solution used in the present invention is as follows:
The present invention provides the system of quick obtaining effective image under a kind of real-time scene, and described system includes image acquisition mould
Block, image quality assessment module, image classification module;Described image capture module is connected with image quality assessment module, described
Image quality assessment module is connected with image classification module;Described image classification module is also connected with successive image processing system;
Preferably, described image capture module is responsible for gathering target image, and described image capture module includes optical camera
The image capture devices such as head, infrared camera, scanner, radar;
Preferably, described image capture module obtains image quality evaluation mathematical modulo by gathering great amount of images sample training
Type and image classification mathematical model;
Preferably, described image quality assessment module is based on described image quality evaluation Mathematical Models, described image
Quality assessment module is responsible for being evaluated the quality of the target image of Real-time Collection, and it is right that described image quality assessment module includes
Computer software that picture quality is evaluated and hardware;
Preferably, described image classification module is based on described image classification Mathematical Models, described image classification module
The effectiveness of target image is judged by the result according to image quality assessment module, and described image classification module includes mesh
Logo image effectiveness carries out computer software and the hardware judged;
Preferably, described image classification module and the connected mode of successive image processing system include wired connection, wireless
Connect and by storage medium transmission etc.;
The present invention also provides for a kind of method of quick obtaining effective image under real-time scene, and described method includes following step
Rapid:
Step one, under a certain specific scene, described image capture module obtains by gathering great amount of images sample training
To image quality evaluation mathematical model and image classification mathematical model;Based on described in described image quality evaluation Mathematical Models
Image quality assessment module;Based on image classification module described in described image classification Mathematical Models;
Step 2, in the image acquisition process under this scene, described image quality assessment module uses described figure picture element
Amount is evaluated the mathematical model image to collecting and is carried out Fast Evaluation, then by image quality evaluation result and image input picture
Sort module;
Step 3, described image classification module uses the described image classification mathematical model image quality evaluation knot to input
Fruit judges, if it is determined that be effective image, then stop gathering, draws the result successfully obtaining effective image, and image is defeated
Enter aftertreatment systems;If it is determined that be invalid image, drawing fails obtains the result of effective image, continues to gather, lays equal stress on
Newly perform step 2.
Preferably, concretely comprising the following steps of described step one:
S1, under a certain specific scene, gathers substantial amounts of image pattern;
S2, is divided into the image pattern gathered in S1 effective image and invalid image two class and demarcates, after effective image is
Continuous result is successful image, invalid image be subsequent treatment result be failed image,
S3, calculates and analyzes the image quality index of all uncalibrated images;
S4, trains the image quality evaluation mathematical model being applicable under this scene, and described mathematical model is for quickly commenting
Estimate the picture quality of the real time imaging gathered under current scene, based on described image quality evaluation mathematical model, set up described
Image quality assessment module;
S5, in conjunction with the result in S3 and S4, trains the image classification mathematical model being applicable under this scene, described mathematics
The effectiveness of the real time imaging that model is gathered under quickly judgement current scene, based on described image classification mathematical model,
Set up described image classification module;
Compared with prior art, the invention has the beneficial effects as follows: under the real-time scene that the present invention provides, quick obtaining is effective
The system and method for image, uses preset image quality evaluation mathematical model and image classification mathematical model, to the figure gathered
As carrying out rapid evaluation and category filter, on the premise of ensureing image acquisition real-time as far as possible, it is ensured that after finally entering
The image of continuous processing system is all effective image.
(3) accompanying drawing explanation
Fig. 1 is the system of quick obtaining effective image under real-time scene
Fig. 2 is the flow chart that under bar code real-time decoding scene, the present invention gathers effective bar code image;
Fig. 3 is the image that on streamline, same bar code picture is captured when diverse location;
(4) specific implementation method
It is prone to bright in order to make the technological means of the present invention, creation characteristic, workflow, using method reach purpose with effect
White understand, below in conjunction with the accompanying drawings and specific embodiment, the specific embodiment of the invention is expanded on further.
System as shown in Figure 1, it is provided that the system of quick obtaining effective image under a kind of real-time scene, described system includes
Image capture module, image quality assessment module, image classification module;Described image capture module and image quality assessment module
Being connected, described image quality assessment module is connected with image classification module;Described image classification module also processes with successive image
System is connected.
Fig. 2 represents that under bar code real-time decoding scene, the present invention gathers the flow chart of effective bar code image, and bar code image is real-time
Under decoding scene, the detailed process of the effective bar code image of quick obtaining of the present invention is as follows:
The image quality evaluation number of bar code image under step one, use image capture module training bar code real-time decoding scene
Learn model and image classification mathematical model;
Step 2, based on described image quality evaluation Mathematical Models image quality assessment module, based on described image
Classification Mathematical Models image classification module;
Step 3, described image quality assessment module use described image quality evaluation mathematical model to calculate Real-time Collection and arrive
The picture quality of bar code image, result of calculation is brought in described classification image classification module;
Step 4, described image classification module use the described image classification mathematical model image quality evaluation knot to input
Fruit judges, if it is determined that be effective image, then stop gathering, draws the result successfully obtaining effective image, and image is defeated
Enter aftertreatment systems;If it is determined that be invalid image, drawing fails obtains the result of effective image, continues to gather, lays equal stress on
Newly perform step 3.
For step one, its detailed step can decompose as follows:
S1, use image capture module gather substantial amounts of bar code image sample as training sample;
S2, being decoded all of training sample and demarcate, decodable bar code image is demarcated as effective bar code image,
Un-decodable bar code image is demarcated as invalid bar code image;
S3, calculate the image quality index of all demarcation samples, set up picture quality based on described image quality index and comment
Valency mathematical model, the image quality evaluation index that this embodiment selects is image definition;
S4, from the angle of image quality evaluation values clearly effectively bar code image and the separability of invalid bar code image classification,
Minima T using the Image Definition value in effective bar code image sample is image effectiveness classification thresholds, base
Image classification mathematical model under described classification thresholds builds current scene.
For the image quality evaluation mathematical model described in S3, described image quality evaluation mathematical model is to image definition
Concrete calculation procedure as follows:
1, each sub regions center point coordinate is calculated
Bar code image etc. is divided into 12 (3 row 4 arrange) sub regions by this example, then have the center point coordinate of every sub regions
As shown in formula (1):
Wherein, the width of W, H representative image respectively and height, i is subregion place columns, and j is subregion place line number.
2, the definition of each window is calculated
The present embodiment selects to open the window of 16*16 pixel in the center of every sub regions, uses energy gradient function
Calculate the Image Definition value of each window, as shown in formula (2):
Wherein, (x y) represents coordinate points (x, y) grey scale pixel value at place to I.
3, the definition of image is calculated
Shown in the calculating of Image Definition value such as formula (3):
Wherein, αijFor the corresponding subregion window weights determined according to priori.
Classifying mathematical model for the image described in S4, described image classification mathematical model is to the judgement of image effectiveness
Comprise the following steps that, use image definition S (I) described in image effectiveness classification thresholds T with S3 described in S4 to compare;
If T is relatively big, then draw the conclusion that present image is invalid;If T is less, then draw the effective conclusion of present image;
In order to further verify that the present invention proposes method effectiveness on acquisition effective image and real-time, in conjunction with
Concrete experiment is further shown.This example uses the Intel i5 processor of 2.6G dominant frequency, the bat of image capture device
Being 100fps according to speed, accompanying drawing 3 (a)-(c) represents captured during the priority position of same bar code picture motion on streamline successively
Image, and the definition of image improves successively, and Fig. 3 (a) and Fig. 3 (b) is invalid bar code image, and Fig. 3 (c) is effective bar code figure
Picture.Traditional image-pickup method makes to have carried out the bar code image in Fig. 3 the computing of 3 decodings, and decodes Fig. 3 (c)
Success;And the image-pickup method in the present invention has carried out 3 intelligibility evaluation to the bar code image in Fig. 3, and only to Fig. 3
(c) decoding success.Table 1 illustrates by two kinds of methods the time spent by bar code image decoding in Fig. 3, traditional images collection
Method makes successfully decoded time-consuming 1470.022ms altogether;And the invention enables final successfully decoded time-consuming 515.921ms altogether, the least
In tradition acquisition method.Its reason is as shown in table 1: assessment judges that the time spent by an invalid bar code image is much smaller than solving
The time of this image of code.It addition, in the present embodiment the averaged acquisition time of single image be 10ms, intelligibility evaluation is the most time-consuming
It is about 0.5ms, if the image collected all to be carried out intelligibility evaluation, then single image averaged acquisition and intelligibility evaluation every time
Time amounts to 10.5ms.In sum, the real-time of low speed camera collection image is had little to no effect by the present invention, and along with
The raising of camera collection image rate, the impact on the real-time that high-speed camera head gathers image can step up, but from solution
From the point of view of code efficiency, the real-time of its decoding but can bigger lifting.
Table 1
The ultimate principle of the present invention and principal character and advantages of the present invention have more than been shown and described.The technology of the industry
Personnel, it should be appreciated that the present invention is not restricted to the described embodiments, simply illustrating this described in above-described embodiment and description
The principle of invention, without departing from the spirit and scope of the present invention, the present invention also has various changes and modifications, and these become
Change and improvement both falls within scope of the claimed invention.Claimed scope by appending claims and
Equivalent defines.
Claims (16)
1. the system of quick obtaining effective image under a real-time scene, it is characterised in that include image capture module, figure picture element
Amount evaluation module, image classification module;Described image capture module is connected with image quality assessment module, and described picture quality is commented
Valency module is connected with image classification module;Described image classification module is also connected with successive image processing system.
The system of quick obtaining effective image under real-time scene the most according to claim 1, it is characterised in that described image
Acquisition module is responsible for gathering target image, and described image capture module is: optical camera, infrared camera, scanner or thunder
Reach.
The system of quick obtaining effective image under real-time scene the most according to claim 1, it is characterised in that described image
Acquisition module obtains image quality evaluation mathematical model and image classification mathematical model by gathering great amount of images sample training.
The system of quick obtaining effective image under real-time scene the most according to claim 1, it is characterised in that described image
Quality assessment module is responsible for adopting in real time based on described image quality evaluation Mathematical Models, described image quality assessment module
The quality of the target image of collection is evaluated, and described image quality assessment module includes the computer being evaluated picture quality
Software and hardware.
The system of quick obtaining effective image under real-time scene the most according to claim 1, it is characterised in that described image
Sort module is based on described image classification Mathematical Models;Described image classification module is according to the knot of image quality assessment module
The effectiveness of target image is judged by fruit, and described image classification module includes the meter judging target image effectiveness
Calculation machine software and hardware.
The system of quick obtaining effective image under real-time scene the most according to claim 1, it is characterised in that described image
Sort module is included wired connection, wireless connections with the connected mode of successive image processing system or is transmitted by storage medium.
7. the method for quick obtaining effective image under a real-time scene, it is characterised in that comprise the following steps:
Step one, under a certain specific scene, described image capture module obtains figure by gathering great amount of images sample training
As quality evaluation mathematical model and image classification mathematical model;Based on image described in described image quality evaluation Mathematical Models
Quality assessment module;Based on image classification module described in described image classification Mathematical Models;
Step 2, in the image acquisition process under this scene, described image quality assessment module uses described picture quality to comment
The quickly evaluation of the image to collecting of valency mathematical model, then divides image quality evaluation result and image input picture
Generic module;
Step 3, described image classification module uses described image classification mathematical model to enter the image quality evaluation result of input
Row judges, if it is determined that be effective image, then stops gathering, and draws the result successfully obtaining effective image, and by after image input
Continuous processing system;If it is determined that be invalid image, drawing fails obtains the result of effective image, continues to gather, and again holds
Row step 3.
The method of quick obtaining effective image under real-time scene the most according to claim 7, it is characterised in that described step
Rapid one concretely comprise the following steps:
S1, under a certain specific scene, gathers substantial amounts of image pattern;
S2, is divided into the image pattern gathered in S1 effective image and invalid image two class and demarcates, and effective image is follow-up place
Reason result be successful image, invalid image be subsequent treatment result be failed image,
S3, calculates and analyzes the image quality index of all uncalibrated images;
S4, trains the image quality evaluation mathematical model being applicable under this scene, and described mathematical model is worked as rapid evaluation
The picture quality of the real time imaging gathered under front scene, based on described image quality evaluation mathematical model, sets up described image
Quality assessment module;
S5, in conjunction with the result in S3 and S4, trains the image classification mathematical model being applicable under this scene, described mathematical model
For quickly judging the effectiveness of the real time imaging gathered under current scene, based on described image classification mathematical model, set up
Described image classification module.
9. the method for quick obtaining effective image under a real-time scene, it is characterised in that specifically comprise the following steps that
The image quality evaluation mathematical modulo of bar code image under step one, use image capture module training bar code real-time decoding scene
Type and image classification mathematical model;
Step 2, based on described image quality evaluation Mathematical Models image quality assessment module, classify based on described image
Mathematical Models image classification module;
Step 3, described image quality assessment module use described image quality evaluation mathematical model to calculate the bar that Real-time Collection arrives
The picture quality of code image, brings into result of calculation in described classification image classification module;
Step 4, described image classification module use described image classification mathematical model to enter the image quality evaluation result of input
Row judges, if it is determined that be effective image, then stops gathering, and draws the result successfully obtaining effective image, and by after image input
Continuous processing system;If it is determined that be invalid image, drawing fails obtains the result of effective image, continues to gather, and again holds
Row step 3.
The method of quick obtaining effective image under a kind of real-time scene the most according to claim 9, it is characterised in that institute
The step one stated particularly as follows:
S1, use image capture module gather substantial amounts of bar code image sample as training sample;
S2, being decoded all of training sample and demarcate, decodable bar code image is demarcated as effective bar code image, can not
The bar code image of decoding is demarcated as invalid bar code image;
S3, calculate the image quality index of all demarcation samples, set up image quality evaluation number based on described image quality index
Learn model;
S4, from the angle of image quality evaluation values clearly effectively bar code image and the separability of invalid bar code image classification, use
Effectively minima T of the Image Definition value in bar code image sample is image effectiveness classification thresholds, based on institute
State the image classification mathematical model that classification thresholds builds under current scene.
The method of quick obtaining effective image under 11. a kind of real-time scenes according to claim 10, it is characterised in that institute
Stating the image quality evaluation index that image quality evaluation mathematical model used is image definition.
The method of quick obtaining effective image under 12. a kind of real-time scenes according to claim 11, it is characterised in that institute
State image quality evaluation mathematical model the calculating process of image definition is concretely comprised the following steps:
Step one, bar code image are divided into 3 row 4 and arrange totally 12 sub regions, and calculate each sub regions center point coordinate;
Step 2, at subregion center windowing, the definition of calculation window;
Step 3, calculate the definition of whole image.
The method of quick obtaining effective image under 13. a kind of real-time scenes according to claim 12, it is characterised in that institute
The step calculating each sub regions center point coordinate stated is: bar code image is divided into 3 row 4 to arrange totally 12 sub regions, Mei Gezi district
Shown in the center point coordinate in territory such as formula (1):
Wherein, the width of W, H representative image respectively and height, i is subregion place columns, and j is subregion place line number.
The method of quick obtaining effective image under 14. a kind of real-time scenes according to claim 12, it is characterised in that institute
The step of the definition calculating each window stated is: opens the window of 16*16 pixel in the center of subregion, uses energy
Gradient function calculates the Image Definition value of each window, as shown in formula (2):
Wherein, (x y) represents coordinate points (x, y) grey scale pixel value at place to I.
The method of quick obtaining effective image under 15. a kind of real-time scenes according to claim 12, it is characterised in that: institute
The step of the definition calculating image stated is: shown in the calculating of Image Definition value such as formula (3):
Wherein, αijFor the corresponding subregion window weights determined according to priori.
The method of quick obtaining effective image under 16. a kind of real-time scenes according to claim 12, it is characterised in that: will
Described image effectiveness classification thresholds T and described image definition S (I) compare, if T is relatively big, then draw current figure
As invalid conclusion;If T is less, then draw the effective conclusion of present image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610390684.XA CN106067020A (en) | 2016-06-02 | 2016-06-02 | The system and method for quick obtaining effective image under real-time scene |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610390684.XA CN106067020A (en) | 2016-06-02 | 2016-06-02 | The system and method for quick obtaining effective image under real-time scene |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106067020A true CN106067020A (en) | 2016-11-02 |
Family
ID=57421064
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610390684.XA Pending CN106067020A (en) | 2016-06-02 | 2016-06-02 | The system and method for quick obtaining effective image under real-time scene |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106067020A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107665361A (en) * | 2017-09-30 | 2018-02-06 | 珠海芯桥科技有限公司 | A kind of passenger flow counting method based on recognition of face |
CN108229240A (en) * | 2016-12-09 | 2018-06-29 | 杭州海康威视数字技术股份有限公司 | A kind of method and device of determining picture quality |
CN109978029A (en) * | 2019-03-13 | 2019-07-05 | 北京邮电大学 | A kind of invalid image pattern screening technique based on convolutional neural networks |
WO2023236372A1 (en) * | 2022-06-09 | 2023-12-14 | 苏州大学 | Surface defect detection method based on image recognition |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1971583A (en) * | 2006-11-28 | 2007-05-30 | 北京中星微电子有限公司 | Recognition method and system of two-dimension code |
CN101339603A (en) * | 2008-08-07 | 2009-01-07 | 电子科技大学中山学院 | Method for selecting qualified iris image from video frequency stream |
CN102169275A (en) * | 2010-04-28 | 2011-08-31 | 上海盈方微电子有限公司 | Automatic focusing system of digital camera for non-uniform sampling window planning based on golden section |
CN102404602A (en) * | 2011-09-23 | 2012-04-04 | 浙江工业大学 | Vidicon definition detection method based on definition test card |
CN102902819A (en) * | 2012-10-30 | 2013-01-30 | 浙江宇视科技有限公司 | Intelligent video analysis method and device |
CN103268493A (en) * | 2013-05-29 | 2013-08-28 | 上海电机学院 | Framing method for license plate in RGB format |
CN103475898A (en) * | 2013-09-16 | 2013-12-25 | 北京理工大学 | Non-reference image quality assessment method based on information entropy characters |
CN105160339A (en) * | 2015-08-06 | 2015-12-16 | 四川大学 | Two-dimension code printing quality online assessment method |
-
2016
- 2016-06-02 CN CN201610390684.XA patent/CN106067020A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1971583A (en) * | 2006-11-28 | 2007-05-30 | 北京中星微电子有限公司 | Recognition method and system of two-dimension code |
CN101339603A (en) * | 2008-08-07 | 2009-01-07 | 电子科技大学中山学院 | Method for selecting qualified iris image from video frequency stream |
CN102169275A (en) * | 2010-04-28 | 2011-08-31 | 上海盈方微电子有限公司 | Automatic focusing system of digital camera for non-uniform sampling window planning based on golden section |
CN102404602A (en) * | 2011-09-23 | 2012-04-04 | 浙江工业大学 | Vidicon definition detection method based on definition test card |
CN102902819A (en) * | 2012-10-30 | 2013-01-30 | 浙江宇视科技有限公司 | Intelligent video analysis method and device |
CN103268493A (en) * | 2013-05-29 | 2013-08-28 | 上海电机学院 | Framing method for license plate in RGB format |
CN103475898A (en) * | 2013-09-16 | 2013-12-25 | 北京理工大学 | Non-reference image quality assessment method based on information entropy characters |
CN105160339A (en) * | 2015-08-06 | 2015-12-16 | 四川大学 | Two-dimension code printing quality online assessment method |
Non-Patent Citations (2)
Title |
---|
吴利明 等: "一种新的图像清晰度评价方法", 《仪器仪表用户》 * |
潘晴 等: "一种高性价比的动态手势控制系统设计", 《计算机技术与发展》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108229240A (en) * | 2016-12-09 | 2018-06-29 | 杭州海康威视数字技术股份有限公司 | A kind of method and device of determining picture quality |
CN107665361A (en) * | 2017-09-30 | 2018-02-06 | 珠海芯桥科技有限公司 | A kind of passenger flow counting method based on recognition of face |
CN109978029A (en) * | 2019-03-13 | 2019-07-05 | 北京邮电大学 | A kind of invalid image pattern screening technique based on convolutional neural networks |
CN109978029B (en) * | 2019-03-13 | 2021-02-09 | 北京邮电大学 | Invalid image sample screening method based on convolutional neural network |
WO2023236372A1 (en) * | 2022-06-09 | 2023-12-14 | 苏州大学 | Surface defect detection method based on image recognition |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106960195B (en) | Crowd counting method and device based on deep learning | |
CN108686978B (en) | ARM-based fruit category and color sorting method and system | |
CN106127780B (en) | A kind of curved surface defect automatic testing method and its device | |
CN109284674A (en) | A kind of method and device of determining lane line | |
CN106067020A (en) | The system and method for quick obtaining effective image under real-time scene | |
CN105651776A (en) | Device and method for automatically grading beef carcass meat yield based on computer vision | |
CN110399831B (en) | Inspection method and device | |
CN109506628A (en) | Object distance measuring method under a kind of truck environment based on deep learning | |
CN106991668B (en) | Evaluation method for pictures shot by skynet camera | |
CN110889328B (en) | Method, device, electronic equipment and storage medium for detecting road traffic condition | |
CN101271514A (en) | Image detection method and device for fast object detection and objective output | |
KR102199094B1 (en) | Method and Apparatus for Learning Region of Interest for Detecting Object of Interest | |
CN108537787B (en) | Quality judgment method for face image | |
CN104483320A (en) | Digitized defect detection device and detection method of industrial denitration catalyst | |
CN113538375A (en) | PCB defect detection method based on YOLOv5 | |
CN103096117B (en) | Video noise detection method and device | |
CN110189375A (en) | A kind of images steganalysis method based on monocular vision measurement | |
CN113838013A (en) | Blade crack real-time detection method and device in aero-engine operation and maintenance based on YOLOv5 | |
CN112528861A (en) | Foreign matter detection method and device applied to track bed in railway tunnel | |
CN116259002A (en) | Human body dangerous behavior analysis method based on video | |
CN106570440A (en) | People counting method and people counting device based on image analysis | |
CN117726991B (en) | High-altitude hanging basket safety belt detection method and terminal | |
CN110728269B (en) | High-speed rail contact net support pole number plate identification method based on C2 detection data | |
CN114187543A (en) | Safety belt detection method and system in high-altitude power operation scene | |
CN103699876A (en) | Method and device for identifying vehicle number based on linear array CCD (Charge Coupled Device) images |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161102 |