CN110532826B - Bar code recognition device and method based on artificial intelligence semantic segmentation - Google Patents
Bar code recognition device and method based on artificial intelligence semantic segmentation Download PDFInfo
- Publication number
- CN110532826B CN110532826B CN201910773585.3A CN201910773585A CN110532826B CN 110532826 B CN110532826 B CN 110532826B CN 201910773585 A CN201910773585 A CN 201910773585A CN 110532826 B CN110532826 B CN 110532826B
- Authority
- CN
- China
- Prior art keywords
- bar code
- image
- information
- decoding
- graph
- 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.)
- Active
Links
Images
Classifications
-
- 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
- G06K7/10544—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation by scanning of the records by radiation in the optical part of the electromagnetic spectrum
- G06K7/10821—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation by scanning of the records by radiation in the optical part of the electromagnetic spectrum further details of bar or optical code scanning devices
-
- 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)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Electromagnetism (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Toxicology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a bar code recognition device and a method based on artificial intelligence semantic segmentation, which perform rough positioning on a bar code image in a complex environment through a convolutional neural network semantic segmentation algorithm to obtain rough bar code selection information, then perform fine positioning on the rough bar code selection information to obtain a decoding image and corresponding decoding additional information, and finally perform decoding operation on the decoding image according to the decoding additional information to obtain bar code electronic information. The invention can identify the bar code under complex conditions and has the characteristics of high efficiency, high speed and high robustness.
Description
Technical Field
The invention relates to the field of bar code identification, in particular to a bar code identification device and method based on artificial intelligence semantic segmentation.
Background
The bar code recognition task mainly comprises three parts of detection, positioning and decoding. The detection task is used for identifying an object with bar code characteristics in the whole image range; the positioning task is used for giving specific coordinates of the bar code on the basis of detection (x 1, y1, x2 and y2 define a rectangular frame); the decoding task is used for analyzing the electronic data information of the positioned bar code.
The front-end physical information acquisition of the bar code identification is based on a laser method and an image sensor method. The laser-based bar code recognition is only effective for a single bar code, and when a plurality of bar codes appear, the technology is difficult to realize simultaneous recognition of the plurality of bar codes, and has great defects in application.
The bar code recognition based on the image sensor utilizes the traditional image recognition technology to complete the tasks of detection, positioning and decoding. For example, the sliding window technology is used for carrying out iterative calculation on different areas of the whole image for multiple times, detection is completed through a traditional manual design bar code characteristic matching method, bar code coordinates are given and positioned through a traditional image recognition technology, and finally decoding work is completed.
The existing barcode recognition based on an image sensor depends on image preprocessing or image enhancement technology to filter interference noise and on the correctness of a manually designed barcode feature template. Therefore, the effect is not good in the case of a complicated environment, a complicated background, and the like, particularly in the case of a long distance, high noise, and the like. Particularly, when multiple bar codes are mapped (one map contains 2 or more bar codes), the number of bar codes increases, which further increases the difficulty and slows down of detection and positioning due to the traversal search characteristic of the conventional technology in detection and positioning.
In a word, the existing barcode identification method has poor performance on the speed and the accuracy of barcode identification under the conditions of multiple barcodes, complex backgrounds, complex environments and the like.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a bar code identification device and method based on artificial intelligence semantic segmentation, which can effectively identify a bar code in a complex environment.
In order to achieve the purpose, the invention adopts the technical scheme that:
a bar code identification device based on artificial intelligence semantic segmentation comprises an image acquisition unit, an image preprocessing unit, a bar code mask coarse positioning unit, a bar code mask fine positioning unit and a decoding unit which are connected in sequence,
the image acquisition unit is used for acquiring an image containing a bar code from the image sensor; the image preprocessing unit is used for processing the image containing the bar code;
the coarse positioning unit of the bar code mask calculates the preprocessed image by utilizing a convolutional neural network semantic segmentation algorithm to obtain coarse selection information of the bar code, wherein the coarse selection information comprises at least one bar code area; the system is also used for filtering the bar code rough selection information to obtain rough selection credible information, wherein the rough selection credible information comprises at least one bar code area;
the bar code mask fine positioning unit is used for finely positioning each bar code region in the roughly selected trusted information to obtain a decoding image and corresponding decoding additional information;
and the decoding unit is used for receiving the decoding image and the corresponding decoding additional information, selecting a corresponding algorithm according to the decoding additional information to perform decoding operation on the decoding image, and acquiring the bar code electronic information.
The filtering rule of the coarse bar code mask positioning unit for acquiring coarse selection credible information is as follows:
(1) calculating the aspect ratio of each bar code area in the roughing information, and deleting the bar code areas with the aspect ratio larger than M;
(2) calculating the bar code angle of each bar code area in the rough selection information, and deleting the items of which the bar code angle is greater than X;
(3) and deleting the illegal code bar code area.
The fine positioning of the coarse selection credible information bar code area by the bar code mask fine positioning unit is as follows:
(1) and (3) extending and expanding a bar code mask: regarding a bar code pixel point of a bar code area, setting T pixels around the pixel point as bar code pixel points by taking the pixel point as a center, wherein the value of the T pixels is the same as that of the center pixel point;
(2) intercepting a bar code area to obtain a bar code image;
(3) performing reverse rotation operation on the bar code graph according to the bar code angle;
(4) acquiring the boundaries (X1 _ b, X2_ b) of the bar codes in the X direction;
(5) and (3) barcode boundary expansion: respectively adding or subtracting preset values B to the barcode boundaries (x 1_ B, x2_ B) on the original basis to obtain new barcode boundaries (x 1_ bk, x2_ bk);
(6) carrying out binarization operation on the bar code graph;
(7) on the basis of the bar code graph, a graph with coordinates of (x 1_ bk, x2_ bk, y1_ k and y2_ k) is intercepted as a decoding graph, and the bar code system and the original coordinates are used as decoding additional information of the graph.
A bar code identification method based on artificial intelligence semantic segmentation comprises the following steps:
step 1, acquiring an image and performing image preprocessing;
step 2, calculating the processed image by using a convolutional neural network semantic segmentation algorithm to obtain bar code roughing information, wherein the bar code roughing information is a bar code mask image containing a bar code area;
step 3, filtering and screening the roughing information to obtain roughing credible information;
step 4, performing independent fine positioning calculation on each bar code area in the roughly selected credible information to obtain a decoding image and corresponding decoding additional information;
and 5, selecting a corresponding algorithm according to the decoding additional information to perform decoding operation on the decoding image, acquiring a plurality of bar code electronic information, and finishing the identification of the bar codes.
The filtering and screening rules of the step 3 are as follows:
(1) calculating the aspect ratio of each bar code area in the roughing information, and deleting the bar code areas with the aspect ratio larger than M;
(2) calculating the bar code angle of each bar code area in the roughing information, and deleting the items of which the bar code angle is larger than X;
(3) and deleting the illegal code bar code area.
In the step 3, when filtering the rough selection information:
when deleting the bar code area with the width and the height larger than M, setting all pixels of the bar code area with the width and the height larger than M as a background category;
when deleting the items of which the bar code angle is larger than X, setting all pixels of the items of which the bar code angle is larger than X as a background category;
and when the illegal coding bar code area is deleted, setting all pixels of the illegal coding bar code area as a background category.
The step 4 is specifically as follows for the fine positioning of each bar code area of the coarse selection information:
(1) and (3) extending the bar code region: regarding a bar code pixel point of a bar code area, setting T pixels around the pixel point as bar code pixel points by taking the pixel point as a center, wherein the value of the T pixels is the same as that of the center pixel point;
(2) intercepting the bar code area to obtain a bar code image;
(3) according to the bar code angle, carrying out reverse rotation operation on the bar code graph, wherein the coordinates of the bar code graph are (x 1_ k, x2_ k, y1_ k and y2_ k);
(4) acquiring the boundaries (X1 _ b, X2_ b) of the bar codes in the X direction;
(5) and (3) barcode boundary expansion: adding or subtracting a certain numerical value B on the original basis (x 1_ B, x2_ B) of the barcode boundary to obtain a new barcode boundary (x 1_ bk, x2_ bk);
(6) carrying out binarization operation on the bar code graph;
(7) on the basis of the bar code graph, a graph with coordinates (x 1_ bk, x2_ bk, y1_ k and y2_ k) is intercepted to be used as a decoding graph, and the bar code system and the original coordinates (x 1, x2, y1 and y 2) are used as decoding additional information of the graph.
The image preprocessing in step 1 includes, but is not limited to, the following: denoising, blurring, sharpening, color-to-black, scaling, mean value reduction, and normalization.
After the scheme is adopted, firstly, the bar code image in the complex environment is roughly positioned through a convolutional neural network semantic segmentation algorithm to obtain bar code rough selection information, then, the rough selection information is finely positioned to obtain a decoding image and corresponding decoding additional information, and finally, the decoding image is decoded according to the decoding additional information to obtain bar code electronic information. Compared with the prior art, the semantic segmentation algorithm based on artificial intelligence can more accurately identify the characteristics of the bar code, so that the approximate position of the bar code is positioned under complex conditions, and the method has high efficiency; and further screening and positioning are carried out according to the approximate position of the bar code, so as to obtain accurate bar code position information, and the method has high robustness and high accuracy. In a word, the invention can identify the bar code under complex conditions and has the characteristics of high efficiency, high speed and high robustness.
Drawings
FIG. 1 is a schematic block diagram of a bar code identification device according to the present invention;
FIG. 2 is a flow chart of a barcode identification method of the present invention;
FIG. 3 is a rough selection information image obtained by rough positioning according to the present invention;
FIG. 4 is a decoding diagram obtained by fine positioning according to the present invention.
Detailed Description
As shown in figure 1, the invention discloses a bar code recognition device based on artificial intelligence semantic segmentation, which comprises an image acquisition unit, an image preprocessing unit, a bar code mask coarse positioning unit, a bar code mask fine positioning unit and a decoding unit, wherein the output end of the image acquisition unit is connected with the input end of the image preprocessing unit, the output end of the image preprocessing unit is connected with the input end of the bar code mask coarse positioning unit, the output end of the bar code mask coarse positioning unit is connected with the input end of the bar code mask fine positioning unit, the output end of the bar code mask fine positioning unit is connected with the input end of the decoding unit, and the output end of the decoding unit outputs the electronic information of a bar code.
The system comprises an image acquisition unit, a processing unit and a display unit, wherein the image acquisition unit is used for acquiring an image containing a bar code from an image sensor; and the image preprocessing unit is used for carrying out necessary image processing on the image containing the bar code so as to improve the identification accuracy or facilitate subsequent calculation. Image processing includes, but is not limited to: denoising, blurring, sharpening, color-to-black, scaling, mean-value reduction, normalization, etc.
The coarse positioning unit of the code mask of bar, utilize the neural network semantic segmentation algorithm of convolution to calculate the picture after preconditioning, obtain the coarse information of the bar code, the coarse information of the bar code includes at least a bar code area; the system is also used for filtering the bar code rough selection information to obtain rough selection credible information, wherein the rough selection credible information comprises at least one bar code area. The filtering rule for obtaining the rough selection credible information is as follows:
(1) and calculating the aspect ratio of each barcode region, and deleting the barcode regions with the aspect ratio larger than M, wherein the specific value of M needs to be set according to a scene, and the typical value is 10.
(2) And calculating the bar code angle of each bar code area, deleting the items of which the bar code angle is larger than X, wherein the specific value needs to be set according to a scene, and the typical value is 45 degrees.
(3) And deleting the bar code area of the illegal code system, wherein the specific set of the illegal code system is determined by application and is generally a code system which cannot appear under the current scene.
And the bar code mask fine positioning unit is used for finely positioning each bar code region in the roughly selected trusted information to obtain a decoding image and corresponding decoding additional information. The fine positioning specifically comprises: (1) and (3) extending and expanding a bar code mask: and regarding the bar code pixel points of the bar code area, setting T pixels around the pixel point as the bar code pixel points by taking the pixel point as a center, wherein the numerical value is the same as that of the center pixel point.
(2) And intercepting the bar code area to obtain a bar code image.
(3) And performing reverse rotation operation on the bar code graph according to the bar code angle, wherein the coordinates of the bar code graph are (x 1_ k, x2_ k, y1_ k and y2_ k).
(4) And acquiring the boundary of the bar code in the X direction. This can be done using a variety of image processing techniques, such as by projection calculations on the X-axis, in conjunction with appropriate threshold parameters, to identify barcode boundaries (X1 _ b, X2_ b).
(5) The barcode boundary expands. According to the requirements of different code systems, the barcode boundary needs to be added or subtracted with a certain value B on the original basis (x 1_ B, x2_ B), specifically, x1_ bk = x1_ B-B, x2_ bk = x2_ B + B, so as to obtain a new barcode boundary (x 1_ bk, x2_ bk).
(6) And (4) binarization, namely performing binarization operation on the bar code image according to proper threshold value information.
(7) Finally, on the basis of the bar CODE graph, a graph with coordinates (x 1_ bk, x2_ bk, y1_ k and y2_ k) is intercepted as a decoding graph, and a bar CODE system (such as EAN13, QR, CODE 128 and the like) and original coordinates (x 1, x2, y1 and y 2) are used as decoding additional information of the graph.
And the decoding unit is used for receiving the decoding image and the corresponding decoding additional information, selecting a corresponding algorithm according to the decoding additional information to perform decoding operation on the decoding image, and acquiring the bar code electronic information.
Based on the same inventive concept, the invention also discloses a bar code identification method based on artificial intelligence semantic segmentation, and as shown in figure 2, the identification method comprises the following steps:
step 1, obtaining an image and carrying out image preprocessing.
When acquiring an image, an image containing a barcode is acquired from an image sensor through a medium (such as USB or ethernet). Image pre-processing, in turn, involves the necessary image processing tasks to improve recognition accuracy or facilitate subsequent computations, including but not limited to the following: denoising, blurring, sharpening, color-to-black, scaling, mean-value reduction, normalization, etc.
And 2, calculating the processed image by using a convolutional neural network semantic segmentation algorithm to obtain bar CODE rough selection information, wherein the bar CODE rough selection information is a bar CODE mask image containing a bar CODE region, each pixel point of the bar CODE mask image is represented by an integer value, the typical bit width of the integer value is 8 bits, the value range is 0-255, and the meaning of the integer value represents the bar CODE system category (such as background, EAN13, QR, CODE 128 and the like).
The convolutional neural network semantic segmentation algorithm may employ any neural network model, such as ICNet.
And 3, filtering and screening the roughing information to obtain roughing credible information (a new bar code mask image).
The filtering and brushing rules are as follows:
(1) calculating the aspect ratio of each barcode region, and deleting barcode regions with the aspect ratio larger than M (setting all pixels of the region as a background category, such as 0), wherein the specific value of M needs to be set according to the scene, and the typical value is 10.
(2) Calculating the bar code angle of each bar code area, deleting the items of which the bar code angle is larger than X (setting all pixels of the area as background categories, such as 0), setting the specific value according to the scene, and setting the typical value to be 45 degrees. The calculation of the bar code angle can adopt a traditional algorithm, for example, the slope of the line of the frame is calculated, and then the angle is obtained.
(3) The illegal code bar code region is deleted (all pixels in the region are set to a background category, such as 0), and the specific set of illegal codes is determined by the application, and is usually a code system which cannot appear in the current scene.
And 4, performing independent fine positioning calculation on each bar code area in the roughly selected credible information to obtain a decoding image and corresponding decoding additional information.
The fine positioning calculation for each barcode region is specifically as follows:
(1) and (3) extending the bar code region: and regarding the bar code pixel points of the bar code area, setting T pixels around the pixel point as the bar code pixel points by taking the pixel point as a center, wherein the numerical value is the same as that of the center pixel point. For example, the coordinates of a barcode pixel in the barcode region are (x, y), the numerical value is N, and when the extension is extended, the pixels in the (x-T, x + T, y-T, y + T) region are all set to be N.
(2) And intercepting the bar code area to obtain a bar code image.
(3) And (4) performing reverse rotation operation on the bar code image according to the bar code angle (obtained by calculation in the step (3)), so that the inclination angle of the bar code is about 0, the calculation of subsequent bar code boundary extension and the interception of a bar code area are facilitated (the speed is higher), and the analysis of a bar code decoding algorithm is also facilitated. For example, when the bar code angle is 30 degrees, a rotation operation of-30 degrees is performed on the bar code image. The coordinate system of the barcode diagram at this time is (x 1_ k, x2_ k, y1_ k, y2_ k).
(4) And acquiring the boundary of the bar code in the X direction. This can be done using a variety of image processing techniques, such as by projection calculations on the X-axis, in conjunction with appropriate threshold parameters, to identify barcode boundaries (X1 _ b, X2_ b).
(5) The barcode boundary expands. According to the requirements of different code systems, the barcode boundary needs to be added or subtracted with a certain value B on the original basis (x 1_ B, x2_ B), specifically, x1_ bk = x1_ B-B, x2_ bk = x2_ B + B, so as to obtain a new barcode boundary (x 1_ bk, x2_ bk).
(6) And (4) binaryzation, namely performing binaryzation operation on the bar code graph according to proper threshold information.
(7) On the basis of the bar CODE graph, a graph with coordinates (x 1_ bk, x2_ bk, y1_ k and y2_ k) is intercepted to be used as a decoding graph, and a bar CODE system (such as EAN13, QR, CODE 128 and the like) and original coordinates (x 1, x2, y1 and y 2) are used as decoding additional information of the graph.
Step 5, inputting the plurality of decoding images obtained in the step 4 and the corresponding decoding additional information into a bar code decoding module; the bar code decoding module selects a corresponding algorithm to decode the decoding image according to the decoding additional information, acquires a plurality of bar code electronic information and completes the identification of the bar code.
The key point of the invention is that firstly, the invention carries out coarse positioning on the bar code image in the complex environment by the semantic segmentation algorithm of the convolutional neural network to obtain the coarse selection information of the bar code, then carries out fine positioning on the coarse selection information to obtain a decoding image and corresponding decoding additional information, and finally carries out decoding operation on the decoding image according to the decoding additional information to obtain the electronic information of the bar code.
The above description is only exemplary of the present invention and is not intended to limit the technical scope of the present invention, so that any minor modifications, equivalent changes and modifications made to the above exemplary embodiments according to the technical spirit of the present invention are within the technical scope of the present invention.
Claims (4)
1. The utility model provides a bar code recognition device based on artificial intelligence semantic segmentation which characterized in that: comprises an image acquisition unit, an image preprocessing unit, a bar code mask coarse positioning unit, a bar code mask fine positioning unit and a decoding unit which are connected in sequence,
the image acquisition unit is used for acquiring an image containing a bar code from the image sensor; the image preprocessing unit is used for processing the image containing the bar code;
the coarse positioning unit of the bar code mask calculates the preprocessed image by utilizing a convolutional neural network semantic segmentation algorithm to obtain coarse selection information of the bar code, wherein the coarse selection information of the bar code is a bar code mask image containing a bar code area, each pixel point of the bar code mask image is represented by an integer value, and the integer value represents the category of the bar code system; the system is also used for filtering the bar code rough selection information to obtain rough selection credible information, wherein the rough selection credible information comprises at least one bar code area;
the filtering rule of the coarse bar code mask positioning unit for acquiring coarse selection credible information is as follows:
(1) calculating the aspect ratio of each bar code area in the roughing information, and deleting the bar code areas with the aspect ratio larger than M;
(2) calculating the bar code angle of each bar code area in the roughing information, and deleting the items of which the bar code angle is larger than X;
(3) deleting the illegal code bar code area;
the bar code mask fine positioning unit is used for finely positioning each bar code region in the roughly selected credible information to obtain a decoding image and corresponding decoding additional information;
the fine positioning of the coarse selection credible information bar code area by the bar code mask fine positioning unit is as follows:
(1) and (3) extending and expanding a bar code mask: regarding a bar code pixel point of a bar code area, setting T pixels around the pixel point as bar code pixel points by taking the pixel point as a center, wherein the value of the bar code pixel points is the same as that of the center pixel point;
(2) intercepting the expanded bar code area to obtain a bar code image;
(3) according to the bar code angle, carrying out reverse rotation operation on the bar code image to a horizontal state;
(4) taking the boundaries (X1 _ b, X2_ b) in the X direction for the bar code graph in a horizontal state;
(5) and (3) barcode boundary expansion: respectively adding and subtracting preset values B on the original basis (x 1_ B, x2_ B) of the barcode boundary to obtain a new barcode boundary (x 1_ bk, x2_ bk);
(6) carrying out binarization operation on the bar code graph;
(7) on the basis of the bar code graph, a graph with coordinates (x 1_ bk, x2_ bk, y1_ k and y2_ k) is intercepted to be used as a decoding graph, and the bar code system and the original coordinates obtained by a bar code mask coarse positioning unit are used as decoding additional information of the graph;
and the decoding unit is used for receiving the decoding image and the corresponding decoding additional information, and selecting a corresponding algorithm to perform decoding operation on the decoding image according to the decoding additional information to acquire the bar code electronic information.
2. A bar code identification method based on artificial intelligence semantic segmentation is characterized by comprising the following steps: the identification method comprises the following steps:
step 1, acquiring an image and carrying out image preprocessing;
step 2, calculating the processed image by using a convolutional neural network semantic segmentation algorithm to obtain bar code roughing information, wherein the bar code roughing information is a bar code masking image containing a bar code area, each pixel point of the bar code masking image is represented by an integer value, and the integer value represents the bar code system category;
step 3, filtering and screening the roughing information to obtain roughing credible information;
the filtering and screening rules of the step 3 are as follows:
(1) calculating the aspect ratio of each barcode region in the rough selection information, and deleting the barcode regions with the aspect ratio larger than M;
(2) calculating the bar code angle of each bar code area in the roughing information, and deleting the items of which the bar code angle is larger than X;
(3) deleting the illegal code bar code area;
step 4, performing independent fine positioning calculation on each bar code area in the roughly selected credible information to obtain a decoding image and corresponding decoding additional information;
the step 4 of fine positioning of each bar code area of the rough selection information is as follows:
(1) and (3) extending the bar code region: regarding a bar code pixel point of a bar code area, setting T pixels around the pixel point as bar code pixel points by taking the pixel point as a center, wherein the value of the T pixels is the same as that of the center pixel point;
(2) intercepting the expanded bar code area to obtain a bar code image;
(3) according to the bar code angle, reversely rotating the bar code graph to a horizontal state, wherein the coordinates of the bar code graph are (x 1_ k, x2_ k, y1_ k and y2_ k);
(4) taking the boundaries (X1 _ b, X2_ b) in the X direction for the bar code graph in a horizontal state;
(5) and (3) barcode boundary expansion: adding or subtracting a certain numerical value B on the original basis (x 1_ B, x2_ B) of the barcode boundary to obtain a new barcode boundary (x 1_ bk, x2_ bk);
(6) carrying out binarization operation on the bar code graph;
(7) on the basis of the bar code graph, a graph with coordinates (x 1_ bk, x2_ bk, y1_ k and y2_ k) is intercepted to be used as a decoding graph, and the bar code system category information and the original coordinates (x 1, x2, y1 and y 2) obtained in the step 2 are used as decoding additional information of the graph;
and 5, selecting a corresponding algorithm according to the decoding additional information to perform decoding operation on the decoding image, acquiring a plurality of bar code electronic information, and finishing the identification of the bar codes.
3. The barcode recognition method based on artificial intelligence semantic segmentation as claimed in claim 2, wherein: in the step 3, when filtering the rough selection information:
when deleting the bar code area with the width and the height larger than M, setting all pixels of the bar code area with the width and the height larger than M as a background category;
when deleting the barcode area with the barcode angle larger than X, setting all pixels of the barcode area with the barcode angle larger than X as a background category;
and when the illegal coding bar code area is deleted, setting all pixels of the illegal coding bar code area as a background category.
4. The bar code identification method based on artificial intelligence semantic segmentation as claimed in claim 2, wherein: the image preprocessing in step 1 includes, but is not limited to, the following: denoising, blurring, sharpening, color-to-black, scaling, mean value reduction, and normalization.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910773585.3A CN110532826B (en) | 2019-08-21 | 2019-08-21 | Bar code recognition device and method based on artificial intelligence semantic segmentation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910773585.3A CN110532826B (en) | 2019-08-21 | 2019-08-21 | Bar code recognition device and method based on artificial intelligence semantic segmentation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110532826A CN110532826A (en) | 2019-12-03 |
| CN110532826B true CN110532826B (en) | 2022-09-30 |
Family
ID=68663949
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910773585.3A Active CN110532826B (en) | 2019-08-21 | 2019-08-21 | Bar code recognition device and method based on artificial intelligence semantic segmentation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110532826B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112183148B (en) * | 2020-09-18 | 2022-11-18 | 内江师范学院 | Batch bar code positioning method and identification system |
| CN116167394A (en) * | 2023-02-21 | 2023-05-26 | 深圳牛图科技有限公司 | Bar code recognition method and system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010165072A (en) * | 2009-01-13 | 2010-07-29 | Ricoh Co Ltd | Image processing apparatus, image processing system, image processing method, control program, and recording medium |
| CN106056020A (en) * | 2016-06-01 | 2016-10-26 | 东莞理工学院 | Bar code detection and recognition method and system based on machine vision system |
| CN108920992A (en) * | 2018-08-08 | 2018-11-30 | 长沙理工大学 | A kind of positioning and recognition methods of the medical label bar code based on deep learning |
| CN109409161A (en) * | 2018-10-22 | 2019-03-01 | 腾讯科技（深圳）有限公司 | Graphic code recognition methods, device, terminal and storage medium |
| CN109934218A (en) * | 2019-01-22 | 2019-06-25 | 阿里巴巴集团控股有限公司 | A kind of recognition methods and device for logistics single image |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109712165B (en) * | 2018-12-29 | 2022-12-09 | 安徽大学 | Similar foreground image set segmentation method based on convolutional neural network |
-
2019
- 2019-08-21 CN CN201910773585.3A patent/CN110532826B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010165072A (en) * | 2009-01-13 | 2010-07-29 | Ricoh Co Ltd | Image processing apparatus, image processing system, image processing method, control program, and recording medium |
| CN106056020A (en) * | 2016-06-01 | 2016-10-26 | 东莞理工学院 | Bar code detection and recognition method and system based on machine vision system |
| CN108920992A (en) * | 2018-08-08 | 2018-11-30 | 长沙理工大学 | A kind of positioning and recognition methods of the medical label bar code based on deep learning |
| CN109409161A (en) * | 2018-10-22 | 2019-03-01 | 腾讯科技（深圳）有限公司 | Graphic code recognition methods, device, terminal and storage medium |
| CN109934218A (en) * | 2019-01-22 | 2019-06-25 | 阿里巴巴集团控股有限公司 | A kind of recognition methods and device for logistics single image |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110532826A (en) | 2019-12-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110532825B (en) | Bar code identification device and method based on artificial intelligence target detection | |
| CN113591967B (en) | Image processing method, device, equipment and computer storage medium | |
| CN102708552B (en) | Rapid two-dimensional barcode image motion deblurring method | |
| CN111738055B (en) | Multi-category text detection system and bill form detection method based on same | |
| CN112651259B (en) | Two-dimensional code positioning method and mobile robot positioning method based on two-dimensional code | |
| CN110070548B (en) | Deep learning training sample optimization method | |
| CN110532826B (en) | Bar code recognition device and method based on artificial intelligence semantic segmentation | |
| CN111524117A (en) | Tunnel surface defect detection method based on characteristic pyramid network | |
| CN110570442A (en) | Contour detection method under complex background, terminal device and storage medium | |
| CN104899589B (en) | It is a kind of that the pretreated method of two-dimensional bar code is realized using threshold binarization algorithm | |
| Bellon et al. | Edge detection to guide range image segmentation by clustering techniques | |
| Jain et al. | A systematic literature review on qr code detection and pre-processing | |
| CN113609984A (en) | Pointer instrument reading identification method and device and electronic equipment | |
| CN113989604A (en) | Tire DOT information identification method based on end-to-end deep learning | |
| Zhou et al. | Fast and resource-efficient hardware implementation of modified line segment detector | |
| CN112487537A (en) | Building surface multistage optimization extraction method based on full convolution neural network | |
| CN111260564A (en) | Image processing method and device and computer storage medium | |
| Bodnár et al. | A novel method for barcode localization in image domain | |
| CN111292346B (en) | Method for detecting contour of casting box body in noise environment | |
| CN113591507A (en) | Robust two-dimensional code DataMatrix positioning method and system | |
| CN112084874B (en) | Object detection method and device and terminal equipment | |
| CN115063826A (en) | Mobile terminal driver license identification method and system based on deep learning | |
| CN118261874B (en) | Fishing rod fishing performance analysis method based on image segmentation large model and multi-element high-order regression fitting | |
| CN111652012B (en) | Curved surface QR code positioning method based on SSD network model | |
| CN109271986B (en) | Digital identification method based on Second-Confirm |
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 |