CN108961262A - A kind of Bar code positioning method under complex scene - Google Patents
A kind of Bar code positioning method under complex scene Download PDFInfo
- Publication number
- CN108961262A CN108961262A CN201810476195.5A CN201810476195A CN108961262A CN 108961262 A CN108961262 A CN 108961262A CN 201810476195 A CN201810476195 A CN 201810476195A CN 108961262 A CN108961262 A CN 108961262A
- Authority
- CN
- China
- Prior art keywords
- bar code
- gradient
- subregion
- image
- picture
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06K—GRAPHICAL DATA READING; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
- G06K7/00—Methods or arrangements for sensing record carriers, e.g. for reading patterns
- G06K7/10—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation
- G06K7/14—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation using light without selection of wavelength, e.g. sensing reflected white light
- G06K7/1404—Methods for optical code recognition
- G06K7/146—Methods for optical code recognition the method including quality enhancement steps
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- 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
- G06V10/24—Aligning, centring, orientation detection or correction of the image
- G06V10/243—Aligning, centring, orientation detection or correction of the image by compensating for image skew or non-uniform image deformations
Abstract
A kind of Bar code positioning method under complex scene, is divided the image into several sub-regions first, extracts the gradient orientation histogram HOG feature of each subregion, and classified using trained Boosted classifier；Hough transformation is carried out to sorted subregion, to obtain the rotation angle of bar code, and rotation correction is carried out to image；Image after correction is detected by gradient and Hough Line segment detection obtains the accurate bounding box of bar code.The present invention has good detection effect to situations such as each bar code and distortion, uneven illumination, partial occlusion for rotating angle.In addition, the present invention is not necessarily to any prior information and artificial mark when detecting；Meanwhile, it is capable to detect the accurate bounding box of bar code, accurate bar code region is provided for next decoding operate, reduces the region searched for when decoding and cost, improves decoded precision.
Description
Technical field
The invention belongs to technical field of machine vision, are related to Bar Code, are based on machine learning and Hough to be a kind of
Bar code positioning method under the complex scene of transformation.
Background technique
Bar code is a kind of by the width figure mark that a series of equal black and white items are not arranged according to certain coding rule
Know symbol, contain the most information of article, is applied in logistics and retail trade the most extensive.Traditional Bar Code
For photoelectricity Bar Code, dedicated optoelectronic scanning, detection and decoding equipment are needed；With image processing techniques development and
The Bar Code of the appearance of two dimensional code, image-type is come into being, and the extensive use of especially Intelligent mobile equipment more pushes away
The development of image barcode identification technology is moved.
It is past much the bar code identification facility based on image, such as ZXing, Zbar etc., these tools have been had already appeared at present
Toward carrying out the positioning and decoding of bar code using the method for progressive scan, under simple background or artificial alignment bar code area
Accuracy rate and efficiency with higher when domain, and in complicated image scene, such as comprising excessive background, uneven illumination, part overexposure,
Performance when bar code distortion is then barely satisfactory.Therefore, before decoding, complicated background, positioning are filtered out in the picture
Out the exact position of bar code be very it is necessary to.
Current barcode technology can be divided into four classes: the filtering of based on morphological operation, based on image scanning, bottom cap with
And range conversion.It is generally basede on the localization method of morphological operation, due to the use corroded and expanded, is difficult to size and interval
The bar code to differ greatly reaches preferable locating effect using unified parameter；It is simple in background based on the method for image scanning
Accuracy rate with higher under scene, but the locating effect and efficiency under complex scene are barely satisfactory；Individual bottom cap filter
Wave operation and positioning accuracy of the distance transformation method under complex scene is not high.It is tended in conjunction with these types of simple localization method
Preferable locating effect is enough obtained, but detection time will be greatly increased.With the development of deep neural network in recent years, also have
Scholar is applied on Bar code positioning, in terms of experimental data, obtains quite high accuracy rate really, but wants to hardware
Ask very high, and general mobile device is extremely difficult to such hardware requirement.
Bibliography:
[1] applied analysis [J] Chinese market of barcode technology, 2015 (45): 59-61. are pleased in Wang Shouhai, Shen
WANG Shouhai,SHEN Yue.Application analysis of barcode techniques[J]
.Chinese Market,2015(45):59-61.
[2] research [J] Packaging Engineering of the sea Wang Yajing, Dou Zhen Bar Code, 2008,29 (8): 240-241.
WANG Yajing,DOU Zhenhai.Investigation of barcode recognition
technology[J].Packaging Engineering,2008,29(8):240-241.
[3]Zebra Crossing.[Online].Available:http://code.google.com/p/zxing/
[4]Zbar.[Online].Available:http://zbar.sourceforge.net/
[5]Katona M,Nyul L G.A Novel Method for Accurate and Efficient
Barcode Detection with Morphological Operations[C]//Eighth International
Conference on Signal Image Technology and Internet Based Systems.IEEE
Computer Society,2012:307-314.
[6] Wang Xialing, Lv Yue, bar code automatic positioning under literary grain husk complex background and inhomogeneous illumination environment and identify
[J] intelligence system journal, 2010,5 (1): 35-40.
WANG Lingxia,LV Yue,WENG Yin.Automatic location and recognition of
barcodes on objects with complex background and non-uniform lighting[J].CAAI
Transactions on Intelligent Systems,2010,5(1):35-40.
[7]Bodnar P,Nyul L G.Improving Barcode Detection with Combination of
Simple Detectors[M].2012.
[8]Zamberletti A,Gallo I,Albertini S.Robust Angle Invariant 1D
Barcode Detection[C]//Iapr Asian Conference on Pattern Recognition.IEEE
Computer Society,2013:160-164.
[9]Creusot C,Munawar A.Real-Time Barcode Detection in the Wild[C]//
IEEE Winter Conference on Applications of Computer Vision.IEEE Computer
Society,2015:239-245.
Summary of the invention
The problem to be solved in the present invention is: with universal and image processing techniques the development of Intelligent mobile equipment, figure
As the Bar Code of formula is widely used, and current Bar code positioning technology is difficult to adapt to complicated image scene；
And some higher technology calculation amounts of positioning accuracy are very big, are extremely difficult to the requirement of real-time.To sum up, existing method is very
Difficulty accomplishes the compatibility of high real-time and high accuracy.
The technical solution of the present invention is as follows: a kind of Bar code positioning method under complex scene, divides the image into several first
Sub-regions extract the gradient orientation histogram HOG feature of each subregion, are carried out using Boosted classifier to subregion
Classification, is divided into bar code region and non-bar code region；Hough transformation is carried out to sorted subregion, obtains the rotation angle of bar code
Degree, and rotation correction is carried out to image, so that bar code is vertical in the picture；To the image after correction by gradient detection and suddenly
Husband's Line segment detection obtains the accurate bounding box of bar code, completes the positioning to bar code.
Further, the present invention the following steps are included:
1) picture I to be measured is divided into the subregion of several m*n, extract each subregion (i, j) HOG feature c (i,
J), and using Boosted classifier classify to each subregion, judge whether it is bar code region；
2) Hough transformation is carried out to 1) middle gained bar code region and obtains bar code angle, θb:
Picture I to be measured obtains edge picture I after Canny edge detectione, to each for wherein belonging to bar code region
Edge carries out Hough transformation, obtains two-dimentional accumulated matrix AH, in two-dimentional accumulated matrix AHIn each column accumulated value indicate be same
Points on the straight line of one slope, and θ value corresponding to most column of counting is the rotation angle, θ of bar codeb；
3) coarse positioning of bar code is carried out using Hough Line segment detection, and according to the result of gradient detection to the bounding box of bar code
Carry out accurate adjustment.
Extract HOG feature specifically:
In each subregion, every wcell*hcellA pixel forms a cell factory cell, oftenIt is a thin
Born of the same parents' unit forms a block block, carries out window scanning by step-length of stride pixel, counts the gradient of each scanning window
Direction histogram, the group number of histogram is nbins, therefore the dimension of HOG feature c (i, j) are as follows:
Subregion classification is carried out using Boosted classifier method particularly includes:
Input by subregion HOG feature c (i, j) as Boosted classifier, is exported:
Boosted classifier is used to filter out the bar code region in original image I, the part number concentrated using given barcode data
According to Boosted classifier is trained as given training set, for giving the picture I in training sett, calculate several instructions
Practice to (in, out), wherein in is the HOG feature vector of subregion, and the value of out is 1 or 0, when subregion is bar code area
Value is 1 when domain, and is called positive sample, and otherwise value is 0, and is called negative sample.
Rotation correction is carried out to image, and calculates gradient the specific method is as follows:
Picture I to be measured is rotated into θbAngle, so that bar code is vertical in the picture, at this point, bar code region vertical direction gradient
Maximum, horizontal direction gradient is minimum, and the vertical direction gradient G of bar code is calculated separately using Sobel operatoryWith horizontal direction gradient
Gx, final gradient G=Gy-Gx, thus obtain gradient map G.
It is pinpoint to bar code progress that the specific method is as follows:
Image after correction detects to obtain gradient map G by gradient, carries out opening operation to gradient map G, removes isolated
Noise spot obtains figure Fg, meanwhile, in gradient map G, line segment is detected using Hough Line Segment Detection Algorithm, only retains all inclinations
Angle is no more than 5 ° of line segment, i.e., substantially horizontal line segment；These line segments are drawn in completely black picture, obtain line segment spy
Sign figure Fl, defined feature picture F=Fg+Fl, carry out the projection of horizontal direction and vertical direction respectively to feature image F, pass through throwing
The size of shadow determines the rectangular area where bar code to get to bounding box, after obtaining bar code region, by its rotation-θb, θbFor item
Region of the rotation angle of code to get bar code in original image.
The present invention propose it is a kind of localization method is detected based on the one-dimension code of machine learning and Hough transformation, first by image point
Several sub-regions are cut into, extract gradient orientation histogram (HOG) feature of each subregion, and using trained
Boosted classifier is classified；Hough transformation is carried out to sorted subregion, to obtain the rotation angle of bar code, and it is right
Image carries out rotation correction；Image after correction is detected by gradient and Hough Line segment detection obtains bar code and accurately surrounds
Box.The present invention has good detection to situations such as each bar code and distortion, uneven illumination, partial occlusion for rotating angle
Effect.
Present invention firstly provides the tilt angles that bar code is detected with Boosted classifier and Hough transformation, can be compatible with
Situations such as bar code and distortion, uneven illumination, partial occlusion of each rotation angle, and believe when detecting without any priori
Breath and artificial mark；Meanwhile the present invention is combined using gradient detection algorithm and Hough line detection algorithm and is able to detect shaping
The accurate bounding box of code provides accurate bar code region for next decoding operate, reduce the region searched for when decoding with
It spends, improves decoded precision.
Detailed description of the invention
Fig. 1 is the flow chart of the method for the present invention.
Fig. 2 is the intermediate effect figure of the method for the present invention.
Fig. 3 is verification and measurement ratio and detection time comparison diagram of the method for the present invention when different sub-district field widths are high.
Fig. 4 is Bar code positioning of embodiment of the present invention result figure.
Fig. 5 is the testing result comparison of the method for the present invention and other art methods.
Specific embodiment
The present invention provides a kind of new Bar code positioning methods, can quickly and accurately position under complicated image scene
To bar code region.The invention mainly comprises two parts of bar code angle calculation and Bar code positioning.
As shown in Figure 1, the present invention divides the image into several subregions first, then root in bar code angle calculation part
According to gradient orientation histogram (HOG) feature of subregion, filtered out using Boosted classifier may include bar code region,
And the rotation angle of bar code is calculated by Hough transformation；In Bar code positioning part, pass through bar code zone level direction after rotation
Obtain to highlight the gradient image in bar code region with the gradient disparities of vertical direction, horizontal direction in the image that will test
Line segment is combined with gradient image, obtains the accurate bounding box of bar code by the method for projection.Specific embodiment is as follows:
1, bar code angle calculation:
By width, equal series of parallel black and white item does not form bar code, with obviously feature, the present invention adopt
It is described with HOG feature, the classification of feature is carried out using Boosted classifier.The realization of Boosted classifier referring to
[10]Viola P,Jones M.Rapid Object Detection using a Boosted Cascade of Simple
Features[C]//Computer Vision and Pattern Recognition,2001.CVPR
2001.Proceedings of the 2001IEEE Computer Society Conference on.IEEE,2003:I-
511-I-518vol.1.
Since bar code can regard series of parallel line segment as, each line segment tilt angle having the same, the present invention
It is detected using Hough transformation.Hough transformation is a kind of method of very effective detection discontinuous point boundary shape, is passed through
Image coordinate space is transformed into parameter space to carry out the fitting of straight line or curve.Straight line in image can indicate
For
ρ=xcos θ+ysin θ (3)
Wherein, ρ >=0 represent straight line to origin vertical range, θ ∈ [0,2 π) indicate straight line and x-axis angle.Image is sat
Mark any point (x in spacei,yi) pole coordinate parameter space (ρ, θ) can be used, i.e., one in hough space is sinusoidal bent
Line expression, the conllinear two o'clock (x in image coordinate spacei,yi) and (xj,yi) it is mapped to two in pole coordinate parameter space
Curve will intersect at a point (ρ0,θ0).It is specific when calculating, parameter space is considered as discrete, establishes a two-dimentional accumulated matrix A
(ρ, θ), to the point (x in each image coordinatei,yi) formula (3) are substituted into using each discrete value of θ respectively, obtain one group
(ρi,θi), and by matrix element A (ρi,θi) plus 1, it calculates after completing, (ρ corresponding to the peak value of A (ρ, θ)0,θ0) it is original
The linear equation parameter for collinearly counting most in figure.Accordingly, bar code angle calculation method particularly includes:
1) division of subregion:
Picture I to be measured is divided into the subregion of several m*n, extract each subregion (i, j) HOG feature c (i,
J), in each subregion, every wcell*hcellA pixel forms a cell factory (cell), oftenA cell
Unit forms a block (block), carries out window scanning by step-length of stride pixel, counts the gradient of each scanning window
Direction histogram, the group number of histogram is nbins, therefore the dimension of HOG feature c (i, j) are as follows:
2) classified using Boosted classifier to each subregion:
Input by c (i, j) as Boosted classifier, is exported:
By formula (2) it is found that Boosted classifier is to use given yardage for filtering out the bar code region in original image I
It is trained according to the partial data of concentration as given training set.A picture I in given training sett, then can calculate
Several training are to (in, out), wherein in is the HOG feature vector of subregion, and the value of out is 1 or 0, when subregion is
Value is 1 when bar code region, and is called positive sample, and otherwise value is 0, and is called negative sample.In order to guarantee positive negative sample
Equilibrium, picture I trained for onet, retain all bar code regions as positive sample, randomly select several background area conducts
Negative sample.
3) bar code angle is calculated using Hough transformation, specifically:
Picture I to be measured obtains edge picture I after Canny edge detectione, to wherein ctThe region of (i, j)=1, i.e. item
Each edge in code region carries out Hough transformation, obtains two-dimentional accumulated matrix AH.In two-dimentional accumulated matrix AHIn it is each
What the accumulated value of column indicated is the points on the straight line of same slope, and θ value corresponding to most column of counting is to mapping
The tilt angle theta in bar code opposed vertical direction in piece figureb。
2, Bar code positioning:
According to the tilt angle theta of bar codeb, the accurate positioning of bar code, tool are carried out using gradient detection and Hough Line segment detection
Body is as follows:
1) rotation correction is carried out to image, and calculates gradient:
Picture I to be measured is rotated into θbAngle, so that bar code is vertical in the picture, i.e. the code line level of bar code.At this point, item
Code region vertical direction gradient is maximum, and horizontal direction gradient is minimum.The vertical direction ladder of bar code is calculated separately using Sobel operator
Spend GyWith horizontal direction gradient Gx, final gradient G=Gy-Gx, thus obtain gradient map G.
2) gradient map and Hough Line segment detection is combined to be accurately positioned bar code:
Opening operation is carried out to gradient map G, removes isolated noise spot, obtains figure Fg.Meanwhile in gradient map G, using suddenly
Husband's Line Segment Detection Algorithm detects line segment, only retains the line segment that all tilt angles are no more than 5 °, i.e., substantially horizontal line segment.?
These line segments are drawn in completely black picture, obtain line segment feature figure Fl.Defined feature picture F=Fg+Fl.To feature image F points
Not carry out horizontal direction and vertical direction projection, the rectangular area where bar code can be determined by the size of projection.It obtains
Behind bar code region, by its rotation-θbTo get region of the bar code in original image.
Fig. 2 is the implementation result figure of each step of the present invention, and bar code image credit to be detected is in two open barcode datas
Collection: the ArTe-Lab rotation barcode data collection and Munster University's barcode data collection of the offers such as Zamberletti.In Fig. 2 respectively
Are as follows: (a) sub-zone dividing schematic diagram, (b) Canny detect edge graph Ie, (c) Boosted detection of classifier go out bar code region
The two-dimentional accumulated matrix A that schematic diagram, (d) Hough transformation obtainHAnd its projection, (e) gradient map G, (f) pass through opening operation gradient
Scheme Fg, (g) feature image F is in projection both vertically as well as horizontally, (h) bar code bounding box effect picture.
In the present invention, the subregion size of division has a significant impact to the precision and detection time of subsequent detection, sub-district
Domain is smaller, reduces to the time that each subregion is classified, however the quantity of subregion also increases, and correspondingly also increases detection
Time.Fig. 3 show choose different sub-district field width it is high when, when present invention verification and measurement ratio on both data sets and detection
Between.Comprehensively consider detection accuracy and detection efficiency, it is proposed that selecting sub-district field width m is 96, high n is 32.
Fig. 4 is the several embodiment result figures positioned using the method for the present invention, and the besieged box positioning of bar code marks in figure,
It can be seen that comprising complicated images scenes such as excessive background, uneven illumination, part overexposure, partial occlusion, bar code distortions,
The method of the present invention can realize the accurate positionin to bar code.
The accuracy rate calculation that the present invention takes are as follows:
Wherein, R indicates the bar code region detected, and G indicates true bar code region, and ε (t) indicates unit-step function,
Therefore:
Recall rate can be expressed as the ratio that the picture number of Detection accuracy A (R, G) higher than 0.5 accounts for total test chart the piece number
Fig. 5 is that one kind of the propositions such as the present invention and Zamberletti is based on the Bar code positioning method of multi-layer perception (MLP) (MLP)
Recall rate comparison diagram, MLP method is referring to [11] Zamberletti A, Gallo I, Albertini S.Robust Angle
Invariant 1D Barcode Detection[C]//Iapr Asian Conference on Pattern
Recognition.IEEE Computer Society,2013:160-164..The present invention is relative to other bar codes as seen from Figure 5
Localization method has higher recall rate.The test statistics of two datasets are shown to be selected as when the threshold value of accuracy rate A (R, G)
When 0.5, recall rate of the invention is up to 0.93, much higher than the 0.64 of MLP method.Meanwhile the present invention is averaged a spoke code picture
Detection time be 92 milliseconds, be fully able to meet the testing requirements of real-time.
Claims (6)
1. a kind of Bar code positioning method under complex scene is extracted it is characterized in that dividing the image into several sub-regions first
The gradient orientation histogram HOG feature of each subregion, classifies to subregion using Boosted classifier, is divided into bar code
Region and non-bar code region；Hough transformation is carried out to sorted subregion, obtains the rotation angle of bar code, and carry out to image
Rotation correction, so that bar code is vertical in the picture；Image after correction is detected by gradient and Hough Line segment detection obtains
The accurate bounding box of bar code completes the positioning to bar code.
2. the Bar code positioning method under a kind of complex scene according to claim 1, it is characterized in that the following steps are included:
1) picture I to be measured is divided into the subregion of several m*n, extracts the HOG feature c (i, j) of each subregion (i, j),
And classified using Boosted classifier to each subregion, judge whether it is bar code region；
2) Hough transformation is carried out to 1) middle gained bar code region and obtains bar code angle, θb:
Picture I to be measured obtains edge picture I after Canny edge detectione, to each edge for wherein belonging to bar code region
Hough transformation is carried out, two-dimentional accumulated matrix A is obtainedH, in two-dimentional accumulated matrix AHIn each column accumulated value indicate be it is same tiltedly
Points on the straight line of rate, and θ value corresponding to most column of counting is the rotation angle, θ of bar codeb；
3) rotation correction is carried out to image, i.e. rotation θbAngle, and gradient detection and Hough line segment are passed through to the image after correction
Detection obtains the accurate bounding box of bar code, completes the positioning to bar code.
3. the Bar code positioning method under a kind of complex scene according to claim 1 or 2, it is characterized in that extracting HOG feature
Specifically:
In each subregion, every wcell*hcellA pixel forms a cell factory cell, oftenA cell list
Member one block block of composition, carries out window scanning by step-length of stride pixel, counts the gradient direction of each scanning window
Histogram, the group number of histogram is nbins, therefore the dimension of HOG feature c (i, j) are as follows:
4. the Bar code positioning method under a kind of complex scene according to claim 1 or 2, it is characterized in that utilizing Boosted
Classifier carries out subregion classification method particularly includes:
Input by subregion HOG feature c (i, j) as Boosted classifier, is exported:
Boosted classifier is used to filter out the bar code region in original image I, is made using the partial data that given barcode data is concentrated
Boosted classifier is trained for given training set, for giving the picture I in training sett, calculate several training pair
(in, out), wherein in is the HOG feature vector of subregion, and the value of out is 1 or 0, when subregion is bar code region
Value is 1, and is called positive sample, and otherwise value is 0, and is called negative sample.
5. the Bar code positioning method under a kind of complex scene according to claim 1 or 2, it is characterized in that being revolved to image
It transfers to another school just, and calculates gradient the specific method is as follows:
Picture I to be measured is rotated into θbAngle, so that bar code is vertical in the picture, at this point, bar code region vertical direction gradient is maximum,
Horizontal direction gradient is minimum, and the vertical direction gradient G of bar code is calculated separately using Sobel operatoryWith horizontal direction gradient Gx, most
Whole gradient G=Gy-Gx, thus obtain gradient map G.
6. the Bar code positioning method under a kind of complex scene according to claim 1 or 2, it is characterized in that determining bar code
The specific method is as follows for position:
Image after correction detects to obtain gradient map G by gradient, carries out opening operation to gradient map G, removes isolated interference
Point obtains figure Fg, meanwhile, in gradient map G, line segment is detected using Hough Line Segment Detection Algorithm, only retains all tilt angles
Line segment no more than 5 °, i.e., substantially horizontal line segment；These line segments are drawn in completely black picture, obtain line segment feature figure
Fl, defined feature picture F=Fg+Fl, carry out the projection of horizontal direction and vertical direction respectively to feature image F, pass through projection
Size determines the rectangular area where bar code to get to bounding box, after obtaining bar code region, by its rotation-θb, θbFor bar code
Rotate region of the angle to get bar code in original image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810476195.5A CN108961262B (en) | 2018-05-17 | 2018-05-17 | Bar code positioning method in complex scene |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810476195.5A CN108961262B (en) | 2018-05-17 | 2018-05-17 | Bar code positioning method in complex scene |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108961262A true CN108961262A (en) | 2018-12-07 |
| CN108961262B CN108961262B (en) | 2023-04-18 |
Family
ID=64499854
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810476195.5A Active CN108961262B (en) | 2018-05-17 | 2018-05-17 | Bar code positioning method in complex scene |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108961262B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110210409A (en) * | 2019-06-04 | 2019-09-06 | 南昌市微轲联信息技术有限公司 | Form frame-line detection method and system in table document |
| CN111523342A (en) * | 2020-04-26 | 2020-08-11 | 成都艾视特信息技术有限公司 | Two-dimensional code detection and correction method in complex scene |
| CN113033234A (en) * | 2019-12-25 | 2021-06-25 | 伟诠电子股份有限公司 | Bar code detection method and system |
| CN113688649A (en) * | 2021-08-16 | 2021-11-23 | 江苏博赛孚医疗科技有限公司 | Quick QR code positioning method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2006031167A (en) * | 2004-07-13 | 2006-02-02 | Glory Ltd | Barcode recognition device, barcode recognition method and barcode recognition program |
| CN104268498A (en) * | 2014-09-29 | 2015-01-07 | 杭州华为数字技术有限公司 | Two-dimension code recognition method and terminal |
-
2018
- 2018-05-17 CN CN201810476195.5A patent/CN108961262B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2006031167A (en) * | 2004-07-13 | 2006-02-02 | Glory Ltd | Barcode recognition device, barcode recognition method and barcode recognition program |
| CN104268498A (en) * | 2014-09-29 | 2015-01-07 | 杭州华为数字技术有限公司 | Two-dimension code recognition method and terminal |
Non-Patent Citations (7)
| Title |
|---|
| PEDRO F. FELZENSZWALB ET AL.: ""Cascade object detection with deformable part models"", 《2010 IEEE COMPUTER SOCIETY CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION》 * |
| PÉTER BODNÁR ET AL.: ""Improving Barcode Detection with Combination of Simple Detectors"", 《2012 EIGHTH INTERNATIONAL CONFERENCE ON SIGNAL IMAGE TECHNOLOGY AND INTERNET BASED SYSTEMS》 * |
| TZU-HAN CHOU ET AL.: ""QR Code Detection Using Convolutional Neural Networks"", 《2015 INTERNATIONAL CONFERENCE ON ADVANCED ROBOTICS AND INTELLIGENT SYSTEMS (ARIS)》 * |
| 于俊伟等: "基于梯度方向评估的条码定位方法研究", 《电脑编程技巧与维护》 * |
| 庞惠文等: "基于直线检测的条码图像几何校正研究", 《包装学报》 * |
| 景亮: ""复杂环境下自动目标识别检测技术研究"", 《中国优秀硕士学位论文全文数据库（硕士） 信息科技辑》 * |
| 陈艾: ""基于机器学习的目标图像定位算法研究"", 《中国优秀硕士学位论文全文数据库（硕士） 信息科技辑》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110210409A (en) * | 2019-06-04 | 2019-09-06 | 南昌市微轲联信息技术有限公司 | Form frame-line detection method and system in table document |
| CN113033234A (en) * | 2019-12-25 | 2021-06-25 | 伟诠电子股份有限公司 | Bar code detection method and system |
| CN113033234B (en) * | 2019-12-25 | 2024-02-13 | 伟诠电子股份有限公司 | Bar code detection method and system |
| CN111523342A (en) * | 2020-04-26 | 2020-08-11 | 成都艾视特信息技术有限公司 | Two-dimensional code detection and correction method in complex scene |
| CN113688649A (en) * | 2021-08-16 | 2021-11-23 | 江苏博赛孚医疗科技有限公司 | Quick QR code positioning method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108961262B (en) | 2023-04-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Moghadam et al. | Fast vanishing-point detection in unstructured environments | |
| Yang et al. | Robust superpixel tracking | |
| CN102693409B (en) | Method for quickly identifying two-dimension code system type in images | |
| Zamberletti et al. | Robust angle invariant 1d barcode detection | |
| CN108961262A (en) | A kind of Bar code positioning method under complex scene | |
| US20210233250A1 (en) | System and method for finding and classifying lines in an image with a vision system | |
| KR101697161B1 (en) | Device and method for tracking pedestrian in thermal image using an online random fern learning | |
| CN102236784A (en) | Screen area detection method and system | |
| US20140301608A1 (en) | Chemical structure recognition tool | |
| CN104123529A (en) | Human hand detection method and system thereof | |
| CN106874901B (en) | Driving license identification method and device | |
| CN109376740A (en) | A kind of water gauge reading detection method based on video | |
| CN105404868B (en) | The rapid detection method of text in a kind of complex background based on interaction platform | |
| CN109947273B (en) | Point reading positioning method and device | |
| CN104751093A (en) | Method and device for acquiring image identification code displayed by host equipment | |
| CN103632137A (en) | Human iris image segmentation method | |
| CA3136674A1 (en) | Methods and systems for crack detection using a fully convolutional network | |
| Haker et al. | Geometric invariants for facial feature tracking with 3D TOF cameras | |
| Bodnár et al. | A novel method for barcode localization in image domain | |
| Kumar | An efficient text extraction algorithm in complex images | |
| TWI543117B (en) | Method for recognizing and locating object | |
| Merino et al. | A framework towards realtime detection and tracking of text | |
| Gao et al. | Vehicle detection based on color and edge information | |
| CN109191489B (en) | Method and system for detecting and tracking aircraft landing marks | |
| Lokkondra et al. | DEFUSE: deep fused end-to-end video text detection and recognition |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |