CN113468958A - Contact net support number plate identification method - Google Patents
Contact net support number plate identification method Download PDFInfo
- Publication number
- CN113468958A CN113468958A CN202110575633.5A CN202110575633A CN113468958A CN 113468958 A CN113468958 A CN 113468958A CN 202110575633 A CN202110575633 A CN 202110575633A CN 113468958 A CN113468958 A CN 113468958A
- Authority
- CN
- China
- Prior art keywords
- number plate
- picture
- feature
- digital
- contact net
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
Abstract
The invention discloses a contact net support number plate identification method, which comprises the following steps: acquiring a railway photo; positioning a strut position in the railway photograph; positioning the number plate in the strut; dividing characters in the number plate; identifying the number plate after being divided; and outputting a digital identification result. The invention adopts the binary threshold value and the connected domain to limit and position the number plate, finds the position of the number plate in a larger range, has higher accuracy and can automatically position the number plate in a larger range under the background of a more complex environment. The method adopts the combination of the structural feature method, the decision tree and the template matching, improves the accuracy of the template matching, can accurately judge the easily mixed numbers in the template matching, has small operand and simple and convenient method, reduces the complexity of feature extraction, improves the accuracy of the template matching, improves the calculation efficiency of the identification method and improves the automatic identification accuracy of the strut number plate.
Description
Technical Field
The invention belongs to the technical field of railway contact network strut number plate identification, and particularly relates to a contact network strut number plate identification method.
Background
Railway contact networks are usually provided with one contact network support column at intervals of 50 meters, and a number plate is printed or hung on the support column. In the process of railway construction and operation maintenance, the column number plate is generally used for providing position information to facilitate the determination of a target position during the construction, operation and maintenance along the railway. At present, the column number plate mainly depends on manual recording, and the efficiency of manually inputting or recording the column number is low under the condition that 15 kilometers of column number plates are already available in China and more than 300 thousand column number plates are available. Therefore, the method for automatically identifying the number plate of the support column can greatly improve the working efficiency of railway construction and operation maintenance.
Chinese patents CN111008619A, CN108509950A, and CN110728269B disclose methods for identifying contact net number plates, which are complicated to operate, require large sample size, have good effect, and lack of intuition in the process of identifying the positioning number plate and characters, or adopt a neural network, or extract features and deep semantics.
The invention name of Chinese patent CN102663377B is: although the applicability of template matching to positions and picture sizes is enhanced, a good recognition effect cannot be achieved for similar characters.
The invention name of Chinese patent CN101763505B is: the license plate character feature extraction and classification method based on projection symmetry provides a character feature extraction method, but the calculation amount is large and the method is complicated. Other character features obtained by using a neural network have the same defects and high technical requirements.
At present, in a strut number plate recognition algorithm in a contact network safety inspection system, a linear detection method for positioning struts is provided, but the number plate cannot be well positioned in an environment with complex environment and multiple struts.
Disclosure of Invention
The invention aims to provide a contact net post number plate identification method.
The technical scheme of the invention is as follows: a contact net support number plate identification method comprises the following steps:
obtaining railway photos
Taking and storing railway photos;
ii, positioning the post position in the railway photograph
Carrying out pillar positioning processing on the stored railway photos, positioning and intercepting range pictures of pillars;
iii. locating the position of the number plate in the strut
Processing the range picture in the step ii, positioning and intercepting the number plate picture;
iv, segmenting characters in the number plate
Processing the license plate picture in the step iii by using a projection method, and dividing to obtain a digital picture of a single number;
v. identifying the number plate after being divided
Carrying out digital identification on the digital picture in the step iv and outputting a digital result
Vi, outputting the digital recognition result
And carrying out sequential combination on the output digital results to obtain the contact net support number plate.
Further, the positioning process of the supporting pillar in step ii includes the following steps:
firstly, carrying out primary interception on a railway photo to obtain a primary intercepted picture;
then, carrying out vertical straight line detection on the initially captured picture to obtain a vertical straight line for detection;
and finally, carrying out secondary interception on the primary intercepted picture by utilizing the vertical straight line to obtain a range picture.
In the detection process of the vertical straight lines, the vertical straight lines with the straightness of-10 degrees to 10 degrees belong to vertical straight lines for detection.
Further, the processing of the range picture in step iii includes the following steps:
firstly, carrying out binarization processing on a range picture to obtain a binarization image;
then, performing morphological corrosion expansion on the binary image;
then, screening a connected domain;
then, a connected domain of the number plate area is obtained;
and finally, cutting according to the connected domain to obtain the number plate picture.
Further, the connected domain screening conditions are as follows: the area, height and width of the connected domain, the proportion of the connected domain in the minimum box area and the length-width ratio of the minimum box.
Further, the processing of the license plate picture by the projection method in step iv includes the following steps:
firstly, projecting a number plate picture to an x axis;
then, clipping is carried out on the digital area;
then, projecting the cut picture to the y axis;
finally, a digital picture of a single number is obtained.
Further, before performing number identification on the digital picture in step v, a feature vector database of each number from 0 to 9 is first established, and the specific form of establishing the feature vector is as follows:
firstly, dividing a binary image of a single number into three sub-images according to columns;
then, calculating the number of connected domains of each subgraph;
then, the structural features, namely feature 1, feature 2 and feature 3 are respectively extracted
Finally, a feature vector is formed, and the specific form of the feature vector is (feature 1, feature 2, feature 3).
Furthermore, the characteristic 1 is the number of connected domains of 1-15 columns of the image; the characteristic 2 is the number of connected domains of 14-18 rows of the image; and the third characteristic is the number of connected domains of 20-36 columns of the image.
Further, the step v of performing digital identification on the digital picture specifically includes the following steps:
firstly, normalizing the pictures of each digital picture to 36 × 36;
then, splitting the digital picture into three subgraphs;
then, respectively calculating the number of connected domains in the three subgraphs to obtain a feature 1, a feature 2 and a feature 3;
then, combining the feature 1, the feature 2 and the feature 3 to obtain a feature vector;
then, judging according to the decision tree;
and finally, outputting the identification result of the single number.
The invention has the following beneficial effects:
the invention adopts the binary threshold value and the connected domain to limit and position the number plate, finds the position of the number plate in a larger range, has higher accuracy and can automatically position the number plate in a larger range under the background of a more complex environment.
The method adopts the combination of the structural feature method, the decision tree and the template matching, improves the accuracy of the template matching, can accurately judge the easily mixed numbers in the template matching, has small operand and simple and convenient method, reduces the complexity of feature extraction, improves the accuracy of the template matching, improves the calculation efficiency of the identification method and improves the automatic identification accuracy of the strut number plate.
Drawings
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a schematic diagram of a digital picture and a sub-picture according to the present invention;
FIG. 3 is a schematic table of feature vectors in the present invention;
FIG. 4 is a schematic diagram of a decision tree in the present invention;
FIG. 5 is a schematic illustration of a railway picture and intercept scope in an embodiment of the invention;
FIG. 6 is a schematic view of a vertical straight line in an embodiment of the present invention;
FIG. 7 is a schematic diagram of an initially captured picture in an embodiment of the invention;
FIG. 8 is a diagram of a range picture in an embodiment of the invention;
FIG. 9 is a diagram illustrating a binarized image according to an embodiment of the present invention;
FIG. 10 is a schematic view of connected domains of a number plate area in an embodiment of the present invention;
FIG. 11 is a graph showing the results of a first screening of connected domains in an embodiment of the present invention;
FIG. 12 is a graph of the results after lateral expansion of connected domains in an example embodiment of the invention;
FIG. 13 shows the results of a second connected component screening in an embodiment of the present invention;
FIG. 14 is a graph of the results after longitudinal expansion of connected domains in an example embodiment of the invention;
FIG. 15 shows the results of a third screening of connected component domains in an example of the present invention;
FIG. 16 is a schematic illustration of a number plate picture in an embodiment of the invention;
FIG. 17 is a diagram illustrating character segmentation in accordance with an embodiment of the present invention;
fig. 18 is a schematic diagram of character recognition in an embodiment of the present invention.
Detailed Description
The present invention is described in detail below with reference to the accompanying drawings and examples:
as shown in fig. 1 to 18, a contact net post number plate identification method includes the following steps:
obtaining railway photos
Taking and storing railway photos;
ii, positioning the post position in the railway photograph
Carrying out pillar positioning processing on the stored railway photos, positioning and intercepting range pictures of pillars;
iii. locating the position of the number plate in the strut
Processing the range picture in the step ii, positioning and intercepting the number plate picture;
iv, segmenting characters in the number plate
Processing the license plate picture in the step iii by using a projection method, and dividing to obtain a digital picture of a single number;
v. identifying the number plate after being divided
Carrying out digital identification on the digital picture in the step iv and outputting a digital result
Vi, outputting the digital recognition result
And carrying out sequential combination on the output digital results to obtain the contact net support number plate.
The positioning treatment of the support column in the step ii comprises the following steps:
firstly, carrying out primary interception on a railway photo to obtain a primary intercepted picture;
then, carrying out vertical straight line detection on the initially captured picture to obtain a vertical straight line for detection;
and finally, carrying out secondary interception on the primary intercepted picture by utilizing the vertical straight line to obtain a range picture.
In the detection process of the vertical straight lines, the vertical straight lines with the straightness of-10 degrees to 10 degrees belong to vertical straight lines for detection.
And (c) processing the range picture in the step iii, wherein the processing comprises the following steps:
firstly, carrying out binarization processing on a range picture to obtain a binarization image;
then, performing morphological corrosion expansion on the binary image;
then, screening a connected domain;
then, a connected domain of the number plate area is obtained;
and finally, cutting according to the connected domain to obtain the number plate picture.
And (3) screening the connected domain under the following screening conditions: the area, height and width of the connected domain, the proportion of the connected domain in the minimum box area and the length-width ratio of the minimum box.
And (iv) processing the license plate pictures by the projection method, wherein the processing method comprises the following steps:
firstly, projecting a number plate picture to an x axis;
then, clipping is carried out on the digital area;
then, projecting the cut picture to the y axis;
finally, a digital picture of a single number is obtained.
Before the digital picture in step v is subjected to digital identification, firstly, a feature vector database of each number of 0-9 is established, and the specific form of establishing the feature vector is as follows:
firstly, dividing a binary image of a single number into three sub-images according to columns;
then, calculating the number of connected domains of each subgraph;
then, the structural features, namely feature 1, feature 2 and feature 3 are respectively extracted
Finally, a feature vector is formed, and the specific form of the feature vector is (feature 1, feature 2, feature 3).
The characteristic 1 is the number of connected domains of 1-15 columns of the image; the characteristic 2 is the number of connected domains of 14-18 rows of the image; and the third characteristic is the number of connected domains of 20-36 columns of the image.
And (f) performing digital identification on the digital picture in the step v, specifically comprising the following steps:
firstly, normalizing the pictures of each digital picture to 36 × 36;
then, splitting the digital picture into three subgraphs;
then, respectively calculating the number of connected domains in the three subgraphs to obtain a feature 1, a feature 2 and a feature 3;
then, combining the feature 1, the feature 2 and the feature 3 to obtain a feature vector;
then, judging according to the decision tree;
and finally, outputting the identification result of the single number.
In the decision tree judging process, three layers, a second layer and a first layer are respectively used as decision tree branch judging conditions to establish a decision tree related to three characteristics, and leaf nodes of the decision tree are single numbers or number sets. And if the leaf nodes are the sets, further using template matching to continue judging. If the leaf node is a number, the judgment result is directly output. Wherein, the leaf nodes are aggregated, the elements are not easy to mix when the templates are matched, and the accuracy is high.
In step ii, when the railway photo is initially captured, one third of the left side is captured, and the upper and lower portions 1/2 to 3/4 are used as possible regular views of the number plate.
In step iii, when the picture is binarized, a threshold value 15 is set, and a black or nearly black area is left.
The first connected domain screening comprises the following steps:
a. no pretreatment is required.
b. Screening conditions are as follows: and removing the connected domain with too long or too short length, too wide or too narrow width, too large or too small area and too small or too large proportion of the minimum external rectangle in the binary image.
And a second connected domain screening, which comprises the following steps:
a. pretreatment: and transversely expanding the binary image.
b. Screening conditions are as follows: and removing the connected domain with too long or too short length, too wide or too narrow width and too large or too small area in the edge graph, wherein the connected domain accounts for the connected domain with too small or too large proportion of the minimum external rectangle.
And (3) carrying out third connected domain screening, which comprises the following steps:
a. pretreatment: and longitudinally expanding the binary image.
b. Screening conditions are as follows: and removing the connected domain with too long or too short length, too wide or too narrow width and too large or too small area in the edge graph, wherein the connected domain accounts for the connected domain with too small or too large proportion of the minimum external rectangle.
The first direct screening removes part of the connected domain, especially near the number of the number plate, and provides conditions for the expansion of the two subsequent steps, so that the subsequent expansion does not link the number and non-number regions together.
And in the second connected domain screening, the numbers are longitudinally arranged, so that the numbers cannot be connected together due to transverse expansion, the characteristics of the connected domains regularly change, and the connected domains are deleted without being regular.
In the third screening of the connected domains, because the number plates are longitudinally arranged, the longitudinal expansion can change the number from a single connected domain into a whole, the connected domains are regularly changed, and the connected domains which do not accord with the rules are deleted. The regular change means that the areas of the separated connected domains are increased suddenly due to connection, and the proportion of the length, the width and the area of the minimum circumscribed rectangle is changed regularly.
The establishment mode of the decision tree is as follows:
and taking a set consisting of 10 numbers of 0-9 as the root of the decision tree.
The feature three is used as a decision condition of the first layer, and the set of roots is divided into 2 subsets, one is {0,1,3,4,7,8,9} and one is {2,5,6 }.
And using the characteristic two as a decision condition of the second layer, and dividing {0,1,3,4,7,8,9} into {0,3,4,7} and {3,8,9} and {1,4 }.
Using feature one as the decision condition of the third layer, divide {0,3,4,7} into {0,4,7} and {3,7}, divide {3,8,9} into {8} and {3,9}, and divide {2,5,6} into {2,5} and {6 }.
Example one
Obtaining railway photos
Taking and storing railway photos;
ii, positioning the post position in the railway photograph
Firstly, carrying out primary interception on a railway photo to obtain a primary intercepted picture, as shown in figure 5;
then, performing vertical straight line detection on the initially captured picture to obtain a vertical straight line for detection, as shown in fig. 6;
and finally, carrying out secondary interception on the primary intercepted picture by using the vertical straight line to obtain a range picture, such as fig. 7 and 8.
In the detection process of the vertical straight lines, the vertical straight lines with the straightness of 10 degrees left and right belong to vertical straight lines for detection.
Iii. locating the position of the number plate in the strut
Firstly, performing binarization processing on a range picture to obtain a binarization picture, wherein the binarization picture is a picture binarized by taking 15 as a threshold value as shown in fig. 9;
then, performing morphological corrosion expansion on the binary image;
then, screening a connected domain;
then, obtaining a connected domain of the number plate area, namely the number plate area connected domain after morphological corrosion expansion and connected domain screening as shown in fig. 10;
and finally, cutting according to the connected domain to obtain a number plate picture, namely, the number plate picture cut according to the connected domain as shown in fig. 11.
Iv, segmenting characters in the number plate
Firstly, projecting a number plate picture to an x axis;
then, clipping is carried out on the digital area;
then, projecting the cut picture to the y axis;
finally, a digital picture of a single number is obtained, and the process is shown in fig. 12;
v. identifying the number plate after being divided
Firstly, normalizing the pictures of each digital picture to 36 × 36;
then, splitting the digital picture into three subgraphs;
then, respectively calculating the number of connected domains in the three subgraphs to obtain a feature 1, a feature 2 and a feature 3;
then, combining the feature 1, the feature 2 and the feature 3 to obtain a feature vector;
then, judging according to the decision tree;
finally, the recognition result of the single number is output, and the process is shown in fig. 13;
vi, outputting the digital recognition result
And carrying out sequential combination on the output digital results to obtain the contact net support number plate.
The character recognition process is described by taking 6 in fig. 2 as an example, and the following is specifically described:
as shown in fig. 2, the digital picture is divided to obtain three subgraphs, feature vectors (1, 3, 2) are obtained according to the number of connected domains of the subgraphs, and the feature vectors are input into the decision tree of fig. 4 to obtain a recognition result 6. Correspondingly, if the digital set is obtained, template matching is carried out on the elements in the set to obtain an identification result.
For the scene graph of the embodiment, the first two times of screening are enough, and the binary graph is hardly changed in the subsequent screening process. And the characteristics of the number plate area and the background area are fully utilized by adopting the three-time screening, so that the method can adapt to more interference and more accurately position the number plate.
The invention adopts the binary threshold value and the connected domain to limit and position the number plate, finds the position of the number plate in a larger range, has higher accuracy and can automatically position the number plate in a larger range under the background of a more complex environment.
The method adopts the combination of the structural feature method, the decision tree and the template matching, improves the accuracy of the template matching, can accurately judge the easily mixed numbers in the template matching, has small operand and simple and convenient method, reduces the complexity of feature extraction, improves the accuracy of the template matching, improves the calculation efficiency of the identification method and improves the automatic identification accuracy of the strut number plate.
Claims (9)
1. A contact net pillar number plate identification method is characterized in that: the method comprises the following steps:
obtaining a railroad photograph
Taking and storing railway photos;
(ii) locating the post position in the railway photograph
Carrying out pillar positioning processing on the stored railway photos, positioning and intercepting range pictures of pillars;
(iii) locating the position of the number plate in the post
Processing the range picture in the step (ii), positioning and intercepting the number plate picture;
(iv) segmenting characters in the number plate
Processing the number plate picture in the step (iii) by using a projection method, and dividing to obtain a digital picture of a single number;
(v) identifying the divided number plate
Performing digital identification on the digital picture in the step (iv) and outputting a digital result
(vi) outputting the result of the number recognition
And carrying out sequential combination on the output digital results to obtain the contact net support number plate.
2. The contact net post number plate identification method of claim 1, wherein: the pillar positioning process in step (ii) includes the steps of:
firstly, carrying out primary interception on a railway photo to obtain a primary intercepted picture;
then, carrying out vertical straight line detection on the initially captured picture to obtain a vertical straight line for detection;
and finally, carrying out secondary interception on the primary intercepted picture by utilizing the vertical straight line to obtain a range picture.
3. The contact net post number plate identification method of claim 2, wherein: in the detection process of the vertical straight lines, the vertical straight lines with the straightness of-10 degrees to 10 degrees belong to vertical straight lines for detection.
4. The contact net post number plate identification method of claim 1, wherein: the processing of the range picture in the step (iii) comprises the following steps:
firstly, carrying out binarization processing on a range picture to obtain a binarization image;
then, performing morphological corrosion expansion on the binary image;
then, screening a connected domain;
then, a connected domain of the number plate area is obtained;
and finally, cutting according to the connected domain to obtain the number plate picture.
5. The contact net post number plate identification method of claim 4, wherein: and (3) screening the connected domain under the following screening conditions: the area, height and width of the connected domain, the proportion of the connected domain in the minimum box area and the length-width ratio of the minimum box.
6. The contact net post number plate identification method of claim 1, wherein: the license plate picture processed by the projection method in the step (iv) comprises the following steps:
firstly, projecting a number plate picture to an x axis;
then, clipping is carried out on the digital area;
then, projecting the cut picture to the y axis;
finally, a digital picture of a single number is obtained.
7. The contact net post number plate identification method of claim 1, wherein: before performing number identification on the digital picture in the step (v), firstly establishing a feature vector database of each number from 0 to 9, wherein the specific form of establishing the feature vector is as follows:
firstly, dividing a binary image of a single number into three sub-images according to columns;
then, calculating the number of connected domains of each subgraph;
then, the structural features, namely feature 1, feature 2 and feature 3 are respectively extracted
Finally, a feature vector is formed, and the specific form of the feature vector is (feature 1, feature 2, feature 3).
8. The contact net post number plate identification method of claim 7, wherein: the characteristic 1 is the number of connected domains of 1-15 columns of the image; the characteristic 2 is the number of connected domains of 14-18 rows of the image; and the third characteristic is the number of connected domains of 20-36 columns of the image.
9. The contact net post number plate identification method of claim 1, wherein: and (v) performing digital identification on the digital picture in the step (v), specifically comprising the following steps:
firstly, normalizing the pictures of each digital picture to 36 × 36;
then, splitting the digital picture into three subgraphs;
then, respectively calculating the number of connected domains in the three subgraphs to obtain a feature 1, a feature 2 and a feature 3;
then, combining the feature 1, the feature 2 and the feature 3 to obtain a feature vector;
then, judging according to the decision tree;
and finally, outputting the identification result of the single number.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110575633.5A CN113468958B (en) | 2021-05-26 | 2021-05-26 | Contact net support number plate identification method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110575633.5A CN113468958B (en) | 2021-05-26 | 2021-05-26 | Contact net support number plate identification method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113468958A true CN113468958A (en) | 2021-10-01 |
CN113468958B CN113468958B (en) | 2022-11-18 |
Family
ID=77871431
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110575633.5A Active CN113468958B (en) | 2021-05-26 | 2021-05-26 | Contact net support number plate identification method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113468958B (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5375176A (en) * | 1993-04-19 | 1994-12-20 | Xerox Corporation | Method and apparatus for automatic character type classification of European script documents |
CN102902974A (en) * | 2012-08-23 | 2013-01-30 | 西南交通大学 | Image based method for identifying railway overhead-contact system bolt support identifying information |
CN104615974A (en) * | 2015-01-15 | 2015-05-13 | 成都交大光芒科技股份有限公司 | Continuous supporting pole number plate image recognizing method based on tracking algorithm |
CN107563380A (en) * | 2017-09-08 | 2018-01-09 | 上海理工大学 | A kind of vehicle license plate detection recognition method being combined based on MSER and SWT |
CN108710878A (en) * | 2018-04-18 | 2018-10-26 | 武汉工程大学 | A kind of railway contact line pillar number plate character segmentation method and system |
WO2019043446A1 (en) * | 2017-09-04 | 2019-03-07 | Nng Software Developing And Commercial Llc | A method and apparatus for collecting and using sensor data from a vehicle |
CN111046872A (en) * | 2019-12-12 | 2020-04-21 | 深圳市杰恩世智能科技有限公司 | Optical character recognition method |
CN111310628A (en) * | 2020-02-10 | 2020-06-19 | 武汉科技大学 | Paper currency forming mode inspection and identification method based on paper currency printing pattern characteristics |
-
2021
- 2021-05-26 CN CN202110575633.5A patent/CN113468958B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5375176A (en) * | 1993-04-19 | 1994-12-20 | Xerox Corporation | Method and apparatus for automatic character type classification of European script documents |
CN102902974A (en) * | 2012-08-23 | 2013-01-30 | 西南交通大学 | Image based method for identifying railway overhead-contact system bolt support identifying information |
CN104615974A (en) * | 2015-01-15 | 2015-05-13 | 成都交大光芒科技股份有限公司 | Continuous supporting pole number plate image recognizing method based on tracking algorithm |
WO2019043446A1 (en) * | 2017-09-04 | 2019-03-07 | Nng Software Developing And Commercial Llc | A method and apparatus for collecting and using sensor data from a vehicle |
CN107563380A (en) * | 2017-09-08 | 2018-01-09 | 上海理工大学 | A kind of vehicle license plate detection recognition method being combined based on MSER and SWT |
CN108710878A (en) * | 2018-04-18 | 2018-10-26 | 武汉工程大学 | A kind of railway contact line pillar number plate character segmentation method and system |
CN111046872A (en) * | 2019-12-12 | 2020-04-21 | 深圳市杰恩世智能科技有限公司 | Optical character recognition method |
CN111310628A (en) * | 2020-02-10 | 2020-06-19 | 武汉科技大学 | Paper currency forming mode inspection and identification method based on paper currency printing pattern characteristics |
Non-Patent Citations (5)
Title |
---|
DADONG ZHAO ET AL.: "Research on recognition algorithm of free handwritten numerals based on combined structural features", 《IEEE INTERNATIONAL CONFERENCE ON ELECTRO-INFORMATION TECHNOLOGY , EIT 2013》 * |
杜鑫 等: "纸币数字的多特征综合识别算法", 《计算机工程与应用》 * |
赵坚勇: "《应用电视技术》", 31 December 2013 * |
陈军胜: "组合结构特征的自由手写体数字识别算法研究", 《计算机工程与应用》 * |
陈明华: "印刷体数字识别算法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Also Published As
Publication number | Publication date |
---|---|
CN113468958B (en) | 2022-11-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111709420B (en) | Text detection method, electronic device and computer readable medium | |
CN111626146B (en) | Merging cell table segmentation recognition method based on template matching | |
CN106407883A (en) | Complex table and method for identifying handwritten numbers in complex table | |
CN107563379B (en) | Method for positioning text in natural scene image | |
CN107967475A (en) | A kind of method for recognizing verification code based on window sliding and convolutional neural networks | |
CN103034848B (en) | A kind of recognition methods of form types | |
CN111401353B (en) | Method, device and equipment for identifying mathematical formula | |
JPH0728940A (en) | Image segmentation for document processing and classification of image element | |
CN101515325A (en) | Character extracting method in digital video based on character segmentation and color cluster | |
CN104050471A (en) | Natural scene character detection method and system | |
CN105608454A (en) | Text structure part detection neural network based text detection method and system | |
CN109977723A (en) | Big bill picture character recognition methods | |
CN110276279B (en) | Method for detecting arbitrary-shape scene text based on image segmentation | |
CN107818321A (en) | A kind of watermark date recognition method for vehicle annual test | |
CN108765349A (en) | A kind of image repair method and system with watermark | |
CN113537227B (en) | Structured text recognition method and system | |
CN112883926B (en) | Identification method and device for form medical images | |
CN111860487B (en) | Inscription marking detection and recognition system based on deep neural network | |
CN104598907A (en) | Stroke width figure based method for extracting Chinese character data from image | |
CN107886065A (en) | A kind of Serial No. recognition methods of mixing script | |
CN109766892A (en) | A kind of ray detection image tagged information character dividing method based on edge detection | |
CN112102250A (en) | Method for establishing and detecting pathological image detection model with training data as missing label | |
CN101615255B (en) | Video text multi-frame interfusion method | |
CN113468958B (en) | Contact net support number plate identification method | |
CN107145888A (en) | Video caption real time translating method |
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 |