CN113516110A - Gas meter character wheel coordinate extraction method based on image segmentation - Google Patents
Gas meter character wheel coordinate extraction method based on image segmentation Download PDFInfo
- Publication number
- CN113516110A CN113516110A CN202111066958.7A CN202111066958A CN113516110A CN 113516110 A CN113516110 A CN 113516110A CN 202111066958 A CN202111066958 A CN 202111066958A CN 113516110 A CN113516110 A CN 113516110A
- Authority
- CN
- China
- Prior art keywords
- character wheel
- image
- wheel window
- window image
- gas meter
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Character Discrimination (AREA)
- Character Input (AREA)
Abstract
The invention relates to a gas meter character wheel coordinate extraction method based on image segmentation, which comprises the following steps: the method comprises the steps that an image acquisition terminal acquires a character wheel window image of the gas meter, the character wheel window image is learned by using a Faster-Rcnn model, and a characteristic parameter file is generated; compressing the acquired character wheel window image and uploading the compressed character wheel window image to an image recognition platform, restoring the compressed character wheel window image by the image recognition platform, and writing the address of the character wheel window image into a Redis monitoring queue; extracting coordinate parameters of character wheel window images in the Redis monitoring queue based on the generated characteristic parameter file; and the image recognition platform performs linear fitting on the generated coordinate parameters and then sends the coordinate parameters to the image acquisition terminal, and the image acquisition terminal divides the character wheel window image according to the coordinate parameters to obtain characters and sends the characters to the image recognition platform for recognition processing.
Description
Technical Field
The invention relates to the technical field of character wheel coordinate identification, in particular to a gas meter character wheel coordinate extraction method based on image segmentation.
Background
In order to obtain characters on the gas meter, the camera shooting acquisition terminal intercepts a character image from a character wheel window according to the character wheel coordinates and the acquired picture according to the coordinates, and then uploads the intercepted character image to a recognition system for recognition in a wireless transmission mode, so that the current characters of the gas meter are obtained. By only uploading the character image, the data volume of communication can be greatly reduced, the communication transmission time is saved, the communication reliability is improved, and the power consumption of the gas meter is reduced.
However, the currently extracted character wheel coordinates are obtained on site manually, each character is moved to a designated area by using a pattern cutting tool, and then the coordinates of the characters are issued to a camera shooting acquisition terminal, so that the following problems exist:
1, the coordinates to be intercepted by a camera shooting acquisition terminal can be determined to correspond to the characters of the gas meter only by manual on-site adjustment, and the installation and debugging are troublesome;
2, if because of external reasons, for example, wind, artificial reasons lead to the collection terminal of making a video recording or the gas table takes place the displacement, then the collection terminal of making a video recording is no longer accurate according to the character of coordinate intercepting, has very big error even, then needs to arrive the scene again and adjust this moment, consequently maintains very trouble.
Disclosure of Invention
The invention aims to accurately segment characters in a character wheel window image according to character wheel coordinate parameters, and provides a gas meter character wheel coordinate extraction method based on image segmentation.
In order to achieve the above object, the embodiments of the present invention provide the following technical solutions:
the gas meter character wheel coordinate extraction method based on image segmentation comprises the following steps:
step S1: the method comprises the steps that an image acquisition terminal acquires a character wheel window image of the gas meter, the character wheel window image is learned by using a Faster-Rcnn model, and a characteristic parameter file of the character wheel window image is generated;
step S2: compressing a character wheel window image of the gas meter acquired by the image acquisition terminal and then uploading the compressed character wheel window image to an image recognition platform, and writing the address of the character wheel window image into a Redis monitoring queue after the compressed character wheel window image is restored by the image recognition platform;
step S3: extracting coordinate parameters of character wheel window images in the Redis monitoring queue based on a characteristic parameter file generated by fast-Rcnn model learning;
step S4: and the image recognition platform performs linear fitting on the generated coordinate parameters corresponding to the character wheel window image and then sends the coordinate parameters to the image acquisition terminal, and the image acquisition terminal divides the acquired character wheel window image according to the coordinate parameters after linear fitting to obtain characters in the character wheel window image and sends the acquired characters to the image recognition platform for recognition processing.
In the scheme, the acquired character wheel window image is compressed and then sent to the image recognition terminal only when the character wheel window image is used for the first time or the position of the equipment moves, so that the coordinate parameter is calculated and acquired, the character wheel window image is subjected to character segmentation by the image acquisition terminal according to the coordinate parameter, the position of the field debugging equipment does not need to be reproduced by personnel, the coordinate parameter can be automatically updated and calculated, and the character extraction accuracy is improved.
The step S1 specifically includes the following steps:
step S1-1: acquiring a character wheel window image of the gas meter, labeling the position of a target body in the character wheel window image by using a labelImg image label, and forming a data file in an xml format after labeling; the number of the target bodies and the position of each target body in the character wheel window image are recorded in the data file;
step S1-2: inputting the acquired character wheel window image and the marked data file in the xml format into a Faster-Rcnn model, and performing feature extraction training to extract the features of the character wheel window image and form a feature parameter file.
The Faster-Rcnn model comprises a convolutional layer, a candidate region recommender, and a pooling layer;
the convolution layer adopts a VGG-16 model to extract the characteristics of the character wheel window image, firstly, the original character wheel window image with the size of P multiplied by Q is zoomed and cut into the size of M multiplied by N, and a characteristic vector with the size of (M/16) multiplied by (N/16) is formed after 13 convolution layers; the convolution kernel of each convolution layer is of a scale of 3 × 3;
after the candidate area recommender calculates the feature vectors output by the convolutional layers by adopting an RPN algorithm, the feature vectors are divided into two paths, one path is used for judging whether the character wheel window image is a foreground or a background, the other path is used for reconstructing the character wheel window image into a one-dimensional feature vector, and then a logistic regression algorithm is used for judging whether the character wheel window image is the foreground or the background; after the two paths of calculation are finished, selecting a foreground candidate frame, and calculating by an RPN algorithm to obtain the position of the candidate frame;
the pooling layer acquires a characteristic subgraph according to the position of the candidate frame, and forms the characteristics of the character wheel window image after 4 pooling layers to generate a characteristic parameter file; the convolution kernel for each of the pooling layers is of a scale of 2 x 2.
The step S2 specifically includes the following steps:
step S2-1: the image acquisition terminal compresses the acquired character wheel window image, uploads the compressed character wheel window image to the image identification platform through an interface provided by the Spring boot server, and uploads the compressed character wheel window image in a binary stream mode when transmitting the compressed character wheel window image;
step S2-2: the image recognition platform analyzes the uploaded binary stream to obtain a gray matrix, and the compressed character wheel window image is restored according to the gray matrix;
step S2-3: and the image identification platform stores the gray matrix as a character wheel window image and writes the address of the character wheel window image into a Redis monitoring queue.
The step S3 specifically includes the following steps:
step S3-1: calling a characteristic parameter file generated by fast-Rcnn model learning in the step S1 by the image recognition platform, and monitoring an address in a Redis monitoring queue by using the characteristic parameter file;
step S3-2: and extracting the coordinate parameters of the character wheel window images in the Redis monitoring queue by using the characteristic parameter file.
In the scheme, the compressed character wheel window image has low resolution and is blurred, so that the image data transmission amount can be reduced, but the compressed character wheel window image can only be used for determining coordinates even after being restored by the image recognition platform and cannot be directly subjected to character recognition, so that the image recognition platform sends the coordinates back to the image acquisition terminal, the image acquisition terminal is enabled to segment characters, and finally the characters are uploaded to the image recognition platform.
The step S4 specifically includes the following steps:
the generated coordinate parameters corresponding to the character wheel window image comprise { (x1, y1), (x2, y2), (. xn, yn) }, wherein n is the number of characters in the character wheel window image, x is the abscissa of the character, and y is the ordinate of the character;
and a function y = f (x; b) is provided, and an optimal estimated value of the parameter b is searched through the coordinate parameters x and y, so that an optimal theoretical curve y = f (x; b) is obtained, and the coordinate parameters after linear fitting are obtained.
Compared with the prior art, the invention has the beneficial effects that:
according to the compressed image uploaded by the image acquisition terminal, the image recognition platform automatically generates the coordinate parameters of the characters, and the image acquisition terminal uploads the cut characters to the image recognition platform according to the character coordinate parameters issued by the image recognition platform, so that the data volume of communication is greatly reduced, the communication transmission time is saved, the communication reliability is improved, and the power consumption of the acquisition terminal is reduced. When the position deviation of the image acquisition terminal or the gas meter is found, the command can be remotely issued to acquire the character coordinate parameters again, and the workload of manual on-site maintenance is reduced.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a schematic structural diagram of the Faster-Rcnn model of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present invention without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
Example (b):
the invention is realized by the following technical scheme, as shown in figure 1, the gas meter character wheel coordinate extraction method based on image segmentation comprises the following steps:
step S1: the method comprises the steps that an image acquisition terminal acquires a character wheel window image of the gas meter, the character wheel window image is learned by using a Faster-Rcnn model, and a characteristic parameter file of the character wheel window image is generated.
Acquiring a character wheel window image of the gas meter, wherein the size of the character wheel window image acquired originally is 120 x 160, labeling the position of a target body in the character wheel window image by using label img image labeling, and forming an xml-format data file after labeling; the number of the target bodies and the position of each target body in the character wheel window image are recorded in the data file, for example, if 8 character wheels are arranged on the gas meter, 8 target bodies can be recorded, and the positions of the target bodies are also recorded.
Inputting the acquired character wheel window image and the marked data file in the xml format into a Faster-Rcnn model, and performing feature extraction training to extract the features of the character wheel window image and form a feature parameter file.
Referring to FIG. 2, the Faster-Rcnn model includes 13 convolutional layers, 4 pooling layers, convolutional layer output feature map, and candidate region at the output of the candidate region recommender.
The convolution layer adopts a VGG-16 model to extract the characteristics of the character wheel window image, firstly, the original character wheel window image with the size of P multiplied by Q is zoomed and cut into the size of M multiplied by N, and a characteristic vector with the size of (M/16) multiplied by (N/16) is formed after 13 convolution layers; the convolution kernel of each of the convolutional layers is of a scale of 3 x 3.
After the candidate area recommender calculates the feature vectors output by the convolutional layers by adopting an RPN algorithm, the feature vectors are divided into two paths, one path is used for judging whether the character wheel window image is a foreground or a background, the other path is used for reconstructing the character wheel window image into a one-dimensional feature vector, and then a logistic regression algorithm is used for judging whether the character wheel window image is the foreground or the background; and after the two paths of calculation are finished, selecting a foreground candidate frame, and calculating by an RPN algorithm to obtain the position of the candidate frame, so that the interested characteristic subgraph can be obtained.
The pooling layer acquires a characteristic subgraph according to the position of the candidate frame, and forms the characteristics of the character wheel window image after 4 pooling layers to generate a characteristic parameter file; the convolution kernel for each of the pooling layers is of a scale of 2 x 2.
The Faster-Rcnn model is used as a black box to realize an end-to-end learning process, parameters of the VGG-16 model are integrated to be used as image feature lifting, an RPN algorithm is integrated to plan a candidate region, main learning parameters comprise a logistic regression parameter for judging whether a foreground or a background, and a regression parameter for accurately determining the position of a candidate frame.
Step S2: and compressing the character wheel window image of the gas meter acquired by the image acquisition terminal and then uploading the compressed character wheel window image to an image recognition platform, and writing the address of the character wheel window image into a Redis monitoring queue after the compressed character wheel window image is restored by the image recognition platform.
The size of the original character wheel window image collected by the image collecting terminal is 120 x 160, in order to reduce the transmission quantity of image data, the image can be compressed to be 60 x 80, the compressed character wheel window image is uploaded to an image recognition platform through an interface provided by a Spring boot server, and the compressed character wheel window image is uploaded in a binary stream mode when being transmitted.
And the image identification platform analyzes the uploaded binary stream to obtain a gray matrix, restores the compressed character wheel window image according to the gray matrix, and restores the image to the size of 120 × 160. Meanwhile, the image recognition platform stores the gray matrix as a character wheel window image, and writes the address of the character wheel window image into a Redis monitoring queue.
Step S3: and extracting the coordinate parameters of the character wheel window images in the Redis monitoring queue based on a characteristic parameter file generated by fast-Rcnn model learning.
And calling the characteristic parameter file generated by fast-Rcnn model learning in the step S1 by the image identification platform, monitoring the address in the Redis monitoring queue by using the characteristic parameter file, wherein the address of a plurality of character wheel window images possibly exist in the Redis monitoring queue at the same time, so that the characteristic parameter file is required to be monitored in real time, and extracting the coordinate parameter of the character wheel window images in the Redis monitoring queue by using the characteristic parameter file.
The compressed character wheel window image has low resolution and fuzzy image, so that the image data transmission amount can be reduced, but the compressed character wheel window image can only be used for determining coordinates even after being restored by the image recognition platform and cannot be directly used for character recognition, so that the image recognition platform can send the coordinates back to the image acquisition terminal, the image acquisition terminal is enabled to segment characters, and finally the characters are uploaded to the image recognition platform.
Step S4: and the image recognition platform performs linear fitting on the generated coordinate parameters corresponding to the character wheel window image and then sends the coordinate parameters to the image acquisition terminal, and the image acquisition terminal divides the acquired character wheel window image according to the coordinate parameters after linear fitting to obtain characters in the character wheel window image and sends the acquired characters to the image recognition platform for recognition processing.
Because characters on the gas meter are basically on a straight line, but the coordinate parameters of each character may not be on a straight line any more, for example, when the characters in fig. 1 are viewed on two-dimensional coordinates, the abscissa x of each character is not equal, and the ordinate y is not equal, it is necessary to perform linear fitting on the coordinate parameters of the characters.
Let the generated coordinate parameters corresponding to the character wheel window image include { (x1, y1), (x2, y2), (. xn, yn) }, where n is the number of characters in the character wheel window image, the upper left corner of the character wheel window image is the origin, x is the abscissa of the character, and y is the ordinate of the character.
And a function y = f (x; b) is provided, an optimal estimated value of the parameter b is sought through the coordinate parameters x and y, so that an optimal theoretical curve y = f (x; b) is obtained, and the coordinate parameters after linear fitting are obtained from the optimal theoretical curve y = f (x; b), so that the character coordinates are on the same vertical coordinate.
And then the image acquisition terminal divides each character according to the received character wheel coordinate parameters and finally sends the character to an image recognition platform for recognition processing.
Step S5: the image acquisition terminals perform character segmentation on the newly acquired character wheel window images according to the coordinate parameters sent by the image identification platform; and repeating the steps S1-S4 until the position of the gas meter or the image acquisition terminal moves or the gas meter or the image acquisition terminal needs to be debugged again, so as to obtain new coordinate parameters.
The image recognition platform can be a program carried on the gas enterprise terminal, so that the gas enterprise terminal has the function of the image recognition platform, and personnel can check characters of the gas meter through the gas enterprise terminal. The terminal of the gas enterprise is the existing terminal of the gas enterprise, the scheme protects the method for extracting the character wheel coordinate, the final purpose is to obtain characters on the character wheel, the image recognition platform is also used for realizing the method, the specific structure and the like of the image recognition platform are not required to be limited, and the specific structure and the like of the image recognition platform are only used for providing one of the realization forms.
The method has the advantages that when the image acquisition terminal corresponding to any gas meter is used for the first time, the character wheel window image needs to be acquired, compressed and uploaded to the image recognition platform, the coordinate parameter corresponding to the character wheel window image is obtained through calculation according to the characteristic parameter file generated through fast-Rcnn model learning, and the image acquisition terminal divides characters in the character wheel window image according to the coordinate parameter. After the image acquisition terminal acquires the coordinate parameters when used for the first time, when the characters of the gas meter are acquired each time, character segmentation can be directly carried out on the newly acquired character wheel window image according to the coordinate parameters, and the newly acquired character wheel window image does not need to be compressed and then uploaded to an image recognition platform. Until the gas meter or the image acquisition terminal is moved by mistake, the position of the gas meter or the image acquisition terminal is changed, and then the image acquisition terminal can acquire the character wheel window image again in a remote instruction transmission mode and upload the character wheel window image to an image identification platform after compressing to obtain new coordinate parameters, so that the image acquisition terminal can perform character segmentation on the character wheel window image acquired later according to the newly obtained coordinate parameters.
In the scheme, when the character wheel window image acquisition device is used for the first time or the position of the device is moved, the acquired character wheel window image is compressed and then sent to the image recognition terminal to calculate and acquire the coordinate parameter, the character wheel window image is subjected to character segmentation by the image acquisition terminal according to the coordinate parameter, the position of the field debugging device does not need to be reproduced by personnel, the coordinate parameter can be automatically updated and calculated, and the accuracy of character extraction is improved.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (6)
1. The gas meter character wheel coordinate extraction method based on image segmentation is characterized by comprising the following steps: the method comprises the following steps:
step S1: the method comprises the steps that an image acquisition terminal acquires a character wheel window image of the gas meter, the character wheel window image is learned by using a Faster-Rcnn model, and a characteristic parameter file of the character wheel window image is generated;
step S2: compressing a character wheel window image of the gas meter acquired by the image acquisition terminal and then uploading the compressed character wheel window image to an image recognition platform, and writing the address of the character wheel window image into a Redis monitoring queue after the compressed character wheel window image is restored by the image recognition platform;
step S3: extracting coordinate parameters of character wheel window images in the Redis monitoring queue based on a characteristic parameter file generated by fast-Rcnn model learning;
step S4: and the image recognition platform performs linear fitting on the generated coordinate parameters corresponding to the character wheel window image and then sends the coordinate parameters to the image acquisition terminal, and the image acquisition terminal divides the acquired character wheel window image according to the coordinate parameters after linear fitting to obtain characters in the character wheel window image and sends the acquired characters to the image recognition platform for recognition processing.
2. The gas meter character wheel coordinate extraction method based on image segmentation as claimed in claim 1, wherein: the step S1 specifically includes the following steps:
step S1-1: acquiring a character wheel window image of the gas meter, labeling the position of a target body in the character wheel window image by using a labelImg image label, and forming a data file in an xml format after labeling; the number of the target bodies and the position of each target body in the character wheel window image are recorded in the data file;
step S1-2: inputting the acquired character wheel window image and the marked data file in the xml format into a Faster-Rcnn model, and performing feature extraction training to extract the features of the character wheel window image and form a feature parameter file.
3. The gas meter character wheel coordinate extraction method based on image segmentation as claimed in claim 2, wherein: the Faster-Rcnn model comprises a convolutional layer, a candidate region recommender, and a pooling layer;
the convolution layer adopts a VGG-16 model to extract the characteristics of the character wheel window image, firstly, the original character wheel window image with the size of P multiplied by Q is zoomed and cut into the size of M multiplied by N, and a characteristic vector with the size of (M/16) multiplied by (N/16) is formed after 13 convolution layers; the convolution kernel of each convolution layer is of a scale of 3 × 3;
after the candidate area recommender calculates the feature vectors output by the convolutional layers by adopting an RPN algorithm, the feature vectors are divided into two paths, one path is used for judging whether the character wheel window image is a foreground or a background, the other path is used for reconstructing the character wheel window image into a one-dimensional feature vector, and then a logistic regression algorithm is used for judging whether the character wheel window image is the foreground or the background; after the two paths of calculation are finished, selecting a foreground candidate frame, and calculating by an RPN algorithm to obtain the position of the candidate frame;
the pooling layer acquires a characteristic subgraph according to the position of the candidate frame, and forms the characteristics of the character wheel window image after 4 pooling layers to generate a characteristic parameter file; the convolution kernel for each of the pooling layers is of a scale of 2 x 2.
4. The gas meter character wheel coordinate extraction method based on image segmentation as claimed in claim 1, wherein: the step S2 specifically includes the following steps:
step S2-1: the image acquisition terminal compresses the acquired character wheel window image, uploads the compressed character wheel window image to the image identification platform through an interface provided by the Spring boot server, and uploads the compressed character wheel window image in a binary stream mode when transmitting the compressed character wheel window image;
step S2-2: the image recognition platform analyzes the uploaded binary stream to obtain a gray matrix, and the compressed character wheel window image is restored according to the gray matrix;
step S2-3: and the image identification platform stores the gray matrix as a character wheel window image and writes the address of the character wheel window image into a Redis monitoring queue.
5. The gas meter character wheel coordinate extraction method based on image segmentation as claimed in claim 1, wherein: the step S3 specifically includes the following steps:
step S3-1: calling a characteristic parameter file generated by fast-Rcnn model learning in the step S1 by the image recognition platform, and monitoring an address in a Redis monitoring queue by using the characteristic parameter file;
step S3-2: and extracting the coordinate parameters of the character wheel window images in the Redis monitoring queue by using the characteristic parameter file.
6. The gas meter character wheel coordinate extraction method based on image segmentation as claimed in claim 1, wherein: the step S4 specifically includes the following steps:
the generated coordinate parameters corresponding to the character wheel window image comprise { (x1, y1), (x2, y2), (. xn, yn) }, wherein n is the number of characters in the character wheel window image, x is the abscissa of the character, and y is the ordinate of the character;
and a function y = f (x; b) is provided, and an optimal estimated value of the parameter b is searched through the coordinate parameters x and y, so that an optimal theoretical curve y = f (x; b) is obtained, and the coordinate parameters after linear fitting are obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111066958.7A CN113516110B (en) | 2021-09-13 | 2021-09-13 | Gas meter character wheel coordinate extraction method based on image segmentation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111066958.7A CN113516110B (en) | 2021-09-13 | 2021-09-13 | Gas meter character wheel coordinate extraction method based on image segmentation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113516110A true CN113516110A (en) | 2021-10-19 |
| CN113516110B CN113516110B (en) | 2021-12-21 |
Family
ID=78063285
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111066958.7A Active CN113516110B (en) | 2021-09-13 | 2021-09-13 | Gas meter character wheel coordinate extraction method based on image segmentation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113516110B (en) |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101581590A (en) * | 2008-05-14 | 2009-11-18 | 李刚 | Camera direct-reading meter with low power consumption and remote transmission device |
| CN203396435U (en) * | 2013-09-04 | 2014-01-15 | 成都千嘉科技有限公司 | Fuel gas metering device based on image recognition |
| US20140046922A1 (en) * | 2012-08-08 | 2014-02-13 | Microsoft Corporation | Search user interface using outward physical expressions |
| CN104880230A (en) * | 2015-06-18 | 2015-09-02 | 成都千嘉科技有限公司 | Magnetic inducting type direct-reading system of liquid hermetic meter |
| CN107356201A (en) * | 2017-07-07 | 2017-11-17 | 杭州山科智能科技股份有限公司 | One kind shooting water meter character wheel coordinate acquiring method and device |
| CN108052943A (en) * | 2017-12-29 | 2018-05-18 | 杭州占峰科技有限公司 | A kind of instrument character wheel recognition methods and equipment |
| CN109284762A (en) * | 2018-09-26 | 2019-01-29 | 旺微科技（上海）有限公司 | A kind of detection method and detection system of multiword wheel instrument digital position |
| CN110135430A (en) * | 2019-04-28 | 2019-08-16 | 广东技术师范大学 | A kind of aluminium mold ID automatic recognition system based on deep neural network |
| CN111079763A (en) * | 2019-12-05 | 2020-04-28 | 北京嘉楠捷思信息技术有限公司 | Training sample generation, model training, character recognition method and device thereof |
| CN111210474A (en) * | 2020-02-26 | 2020-05-29 | 上海麦图信息科技有限公司 | Method for acquiring real-time ground position of airplane in airport |
| CN111814917A (en) * | 2020-08-28 | 2020-10-23 | 成都千嘉科技有限公司 | Character wheel image digital identification method with fuzzy state |
| CN112001383A (en) * | 2020-08-10 | 2020-11-27 | 长沙奇巧匠人软件有限公司 | Water meter code intelligent identification method based on convolutional neural network technology |
| CN112040198A (en) * | 2020-09-16 | 2020-12-04 | 南通天成现代农业科技有限公司 | Intelligent water meter reading identification system and method based on image processing |
-
2021
- 2021-09-13 CN CN202111066958.7A patent/CN113516110B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101581590A (en) * | 2008-05-14 | 2009-11-18 | 李刚 | Camera direct-reading meter with low power consumption and remote transmission device |
| US20140046922A1 (en) * | 2012-08-08 | 2014-02-13 | Microsoft Corporation | Search user interface using outward physical expressions |
| CN203396435U (en) * | 2013-09-04 | 2014-01-15 | 成都千嘉科技有限公司 | Fuel gas metering device based on image recognition |
| CN104880230A (en) * | 2015-06-18 | 2015-09-02 | 成都千嘉科技有限公司 | Magnetic inducting type direct-reading system of liquid hermetic meter |
| CN107356201A (en) * | 2017-07-07 | 2017-11-17 | 杭州山科智能科技股份有限公司 | One kind shooting water meter character wheel coordinate acquiring method and device |
| CN108052943A (en) * | 2017-12-29 | 2018-05-18 | 杭州占峰科技有限公司 | A kind of instrument character wheel recognition methods and equipment |
| CN109284762A (en) * | 2018-09-26 | 2019-01-29 | 旺微科技（上海）有限公司 | A kind of detection method and detection system of multiword wheel instrument digital position |
| CN110135430A (en) * | 2019-04-28 | 2019-08-16 | 广东技术师范大学 | A kind of aluminium mold ID automatic recognition system based on deep neural network |
| CN111079763A (en) * | 2019-12-05 | 2020-04-28 | 北京嘉楠捷思信息技术有限公司 | Training sample generation, model training, character recognition method and device thereof |
| CN111210474A (en) * | 2020-02-26 | 2020-05-29 | 上海麦图信息科技有限公司 | Method for acquiring real-time ground position of airplane in airport |
| CN112001383A (en) * | 2020-08-10 | 2020-11-27 | 长沙奇巧匠人软件有限公司 | Water meter code intelligent identification method based on convolutional neural network technology |
| CN111814917A (en) * | 2020-08-28 | 2020-10-23 | 成都千嘉科技有限公司 | Character wheel image digital identification method with fuzzy state |
| CN112040198A (en) * | 2020-09-16 | 2020-12-04 | 南通天成现代农业科技有限公司 | Intelligent water meter reading identification system and method based on image processing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113516110B (en) | 2021-12-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7099509B2 (en) | Computer vision system for digitization of industrial equipment gauges and alarms | |
| JP4633841B2 (en) | Estimating 3D road layout from video sequences by tracking pedestrians | |
| CN106875437B (en) | RGBD three-dimensional reconstruction-oriented key frame extraction method | |
| CN113052876B (en) | Video relay tracking method and system based on deep learning | |
| CN115331172A (en) | Workshop dangerous behavior recognition alarm method and system based on monitoring video | |
| CN112528974B (en) | Distance measuring method and device, electronic equipment and readable storage medium | |
| CN114155610B (en) | Panel assembly key action identification method based on upper half body posture estimation | |
| CN112464766A (en) | Farmland automatic identification method and system | |
| CN116095347B (en) | Construction engineering safety construction method and system based on video analysis | |
| CN116343103B (en) | Natural resource supervision method based on three-dimensional GIS scene and video fusion | |
| CN111770450B (en) | Workshop production monitoring server, mobile terminal and application | |
| CN112052736A (en) | Cloud computing platform-based field tea tender shoot detection method | |
| CN110827269A (en) | Crop growth change condition detection method, device, equipment and medium | |
| CN112785557A (en) | Belt material flow detection method and device and belt material flow detection system | |
| CN114241012B (en) | High-altitude parabolic determination method and device | |
| CN111402250A (en) | Machine vision defect detection method and platform based on edge calculation | |
| CN117593702B (en) | Remote monitoring method, device, equipment and storage medium | |
| CN113516110B (en) | Gas meter character wheel coordinate extraction method based on image segmentation | |
| CN106530330A (en) | Low-rank sparse-based video target tracking method | |
| CN113569849A (en) | Car fills electric pile interface detection intelligent interaction system based on computer vision | |
| CN111625892A (en) | Construction engineering management system and method based on panoramic technology and BIM model | |
| CN116844172A (en) | Digital old water meter identification method based on deep learning | |
| CN115131407A (en) | Robot target tracking method, device and equipment for digital simulation environment | |
| CN115379197A (en) | Method and device for detecting lens surface of camera product on production line | |
| CN113705442A (en) | Outdoor large-board advertising picture monitoring and identifying system and 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 | ||
| CP01 | Change in the name or title of a patent holder | ||
| CP01 | Change in the name or title of a patent holder |
Address after: No. 536, Section 1, airport 1st Road, Southwest Airport, Shuangliu District, Chengdu, Sichuan 610211 Patentee after: Chengdu Qianjia Technology Co.,Ltd. Address before: No. 536, Section 1, airport 1st Road, Southwest Airport, Shuangliu District, Chengdu, Sichuan 610211 Patentee before: Chengdu Qianjia Technology Co.,Ltd. |