CN112347805A - Multi-target two-dimensional code detection and identification method, system, device and storage medium - Google Patents
Multi-target two-dimensional code detection and identification method, system, device and storage medium Download PDFInfo
- Publication number
- CN112347805A CN112347805A CN202011335084.6A CN202011335084A CN112347805A CN 112347805 A CN112347805 A CN 112347805A CN 202011335084 A CN202011335084 A CN 202011335084A CN 112347805 A CN112347805 A CN 112347805A
- Authority
- CN
- China
- Prior art keywords
- image
- dimensional code
- target
- training
- dimensional
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
- G06K7/14—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation using light without selection of wavelength, e.g. sensing reflected white light
- G06K7/1404—Methods for optical code recognition
- G06K7/1439—Methods for optical code recognition including a method step for retrieval of the optical code
- G06K7/1443—Methods for optical code recognition including a method step for retrieval of the optical code locating of the code in an image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06K—GRAPHICAL DATA READING; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
- G06K7/00—Methods or arrangements for sensing record carriers, e.g. for reading patterns
- G06K7/10—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation
- G06K7/14—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation using light without selection of wavelength, e.g. sensing reflected white light
- G06K7/1404—Methods for optical code recognition
- G06K7/1408—Methods for optical code recognition the method being specifically adapted for the type of code
- G06K7/1417—2D bar codes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06K—GRAPHICAL DATA READING; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
- G06K7/00—Methods or arrangements for sensing record carriers, e.g. for reading patterns
- G06K7/10—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation
- G06K7/14—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation using light without selection of wavelength, e.g. sensing reflected white light
- G06K7/1404—Methods for optical code recognition
- G06K7/146—Methods for optical code recognition the method including quality enhancement steps
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Toxicology (AREA)
- Electromagnetism (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method, a system, a device and a storage medium for detecting and identifying multi-target two-dimensional codes, wherein the method comprises the following steps: acquiring a real-time multi-two-dimension code scene image; performing image enhancement processing on the multi-two-dimension code scene image to obtain a processed enhanced image; positioning a plurality of two-dimensional codes in the enhanced image through a trained multi-two-dimensional code positioning model according to the enhanced image to obtain all two-dimensional code areas; and carrying out two-dimension code identification on the two-dimension code area to obtain all two-dimension code information. The method is based on deep learning, multi-target detection, multi-environment image enhancement and two-dimensional code recognition technology, is faster and simpler compared with R-CNN, is higher in accuracy compared with YOLO, effectively solves the problem that the traditional two-dimensional code recognition can not realize real-time recognition even if a plurality of two-dimensional codes can be positioned, realizes real-time positioning and recognition of the plurality of two-dimensional codes, and greatly improves the detection efficiency.
Description
Technical Field
The invention relates to the technical field of two-dimensional code identification, in particular to a multi-target two-dimensional code detection and identification method, system, device and storage medium.
Background
Currently popular two-dimensional code location and identification techniques include ZBar, ZXing, and OpenCV. The ZBAR is a popular bar code/two-dimensional code scanning tool for desktop computers, supports camera and picture scanning, supports multiple platforms including Windows, Linux and iPhone mobile phone platforms, and can only realize the identification of a single two-dimensional code; net is a bar code development library based on the Microsoft environment, which can identify a plurality of QR codes in a picture by acquiring the coordinates of the QR codes in the picture, but can only analyze one of the QR codes; OpenCV is a cross-platform computer vision library issued based on BSD (open source), realizes many general algorithms in image processing and computer vision, can be used for discovering and positioning multiple two-dimensional codes, but is limited to the identification of static pictures and cannot realize real-time identification during the moving process of a camera.
The existing two-dimensional code identification technology is basically limited to single identification, in an actual two-dimensional code identification scene, more than one object to be identified is often needed, and even though the traditional two-dimensional code identification technology can identify a plurality of two-dimensional codes, the traditional two-dimensional code identification technology cannot correspond to corresponding information, such as ZXing. In 2016, the GAO and the like combine the zxing. net and a basic image processing technology to provide a multi-target QR code identification method, and coordinates of each two-dimensional code in a picture are acquired to identify one by one, but the method has low efficiency. The recent popular computer vision open source library OpenCV can simultaneously position and identify a plurality of two-dimensional codes, but is only limited to identification on static pictures, has low identification speed and cannot realize real-time identification in the moving process of a camera.
Disclosure of Invention
In order to solve the above technical problems, an object of the present invention is to provide a method, a system, an apparatus, and a storage medium for detecting and identifying a multi-target two-dimensional code.
The technical scheme adopted by the invention is as follows:
in a first aspect, an embodiment of the present invention provides a multi-target two-dimensional code detection and identification method, including the following steps:
acquiring a real-time multi-two-dimension code scene image;
carrying out image enhancement processing on the scene image to obtain a processed enhanced image;
positioning a plurality of two-dimensional codes in the enhanced image through a trained multi-two-dimensional code positioning model according to the enhanced image to obtain all two-dimensional code areas;
and carrying out two-dimension code identification on the two-dimension code area to obtain all two-dimension code information.
Further, the method comprises a multi-two-dimension code positioning model establishing step, wherein the multi-two-dimension code positioning model establishing step specifically comprises the following steps:
acquiring a training image containing a two-dimensional code, and labeling the training image to obtain a training data set;
and training the established multi-two-dimension code positioning model according to a preset loss function and a training data set to obtain the trained multi-two-dimension code positioning model.
Further, the step of acquiring the training image containing the two-dimensional code at least includes one of the following steps:
acquiring an image containing a plurality of two-dimensional codes through image acquisition equipment to obtain a training image;
the method comprises the steps of obtaining a two-dimensional code image and a background image through a web crawler, and randomly synthesizing the two-dimensional code image and the background image to obtain a training image.
Further, the image enhancement processing is performed on the multi-two-dimensional code scene image to obtain a processed enhanced image, and the step specifically includes:
and processing the scene image by adopting a high-dynamic-range image processing algorithm of rapid bilateral filtering to obtain a processed enhanced image.
In a second aspect, an embodiment of the present invention provides a multi-target two-dimensional code detection and identification system, including:
the acquisition unit is used for acquiring a real-time multi-two-dimension code scene image;
the enhancement unit is used for carrying out image enhancement processing on the multi-two-dimensional code scene image to obtain a processed enhanced image;
the positioning unit is used for positioning a plurality of two-dimensional codes in the enhanced image through a trained multi-two-dimensional code positioning model according to the enhanced image to obtain all two-dimensional code areas;
and the identification unit is used for carrying out two-dimensional code identification on the two-dimensional code area to obtain all two-dimensional code information.
Further, the method comprises a multi-two-dimension code positioning model establishing unit, wherein the multi-two-dimension code positioning model establishing unit specifically comprises:
the data set acquisition unit is used for acquiring a training image containing the two-dimensional code and labeling the training image to obtain a training data set;
and the training unit is used for training the established multi-two-dimensional code positioning model according to a preset loss function and a training data set to obtain the trained multi-two-dimensional code positioning model.
Further, the obtaining unit at least includes one of the following:
the first acquisition unit is used for acquiring images containing a plurality of two-dimensional codes through image acquisition equipment to obtain training images;
and the second acquisition unit is used for acquiring the two-dimensional code image and the background image through a web crawler and randomly synthesizing the two-dimensional code image and the background image to obtain a training image.
Further, the enhancement unit is specifically configured to:
and processing the multi-two-dimension code scene image by adopting a high-dynamic-range image processing algorithm of rapid bilateral filtering to obtain a processed enhanced image.
In a third aspect, an embodiment of the present invention provides a multi-target two-dimensional code detection and identification device, including:
at least one processor;
at least one memory for storing at least one program;
when the at least one program is executed by the at least one processor, the at least one processor is enabled to realize the multi-target two-dimensional code detection and identification method.
In a fourth aspect, an embodiment of the present invention provides a computer storage medium, which includes a computer program, and when the computer program runs on a computer, the method for detecting and identifying a multi-target two-dimensional code is executed.
The invention has the beneficial effects that:
the multi-target two-dimensional code detection and identification method, the system, the device and the storage medium are based on deep learning, multi-target detection, multi-environment image enhancement and two-dimensional code identification technologies, are faster and simpler compared with R-CNN, and have higher accuracy compared with YOLO, effectively solve the problem that the traditional two-dimensional code identification cannot realize real-time identification even if a plurality of two-dimensional codes can be positioned, realize real-time positioning and identification of the plurality of two-dimensional codes, and greatly improve the detection efficiency.
Drawings
FIG. 1 is a flow chart of steps of a multi-target two-dimensional code detection and identification method of the present invention;
FIG. 2 is a block diagram of a multi-target two-dimensional code detection and identification system according to the present invention;
FIG. 3 is a flow chart of the ZBar algorithm of the multi-target two-dimensional code detection and identification system of the present invention;
FIG. 4a is an input image of a multi-target two-dimensional code detection and recognition system according to the present invention;
FIG. 4b is an 8 × 8 feature diagram of a multi-target two-dimensional code detection and recognition system according to the present invention;
fig. 4c is a 4 × 4 feature diagram of the multi-target two-dimensional code detection and identification system of the present invention.
Detailed Description
The following further describes embodiments of the present invention with reference to the accompanying drawings:
referring to fig. 1, an embodiment of the present invention provides a multi-target two-dimensional code detection and identification method, including the following steps:
s1, acquiring a real-time multi-two-dimensional code scene image;
s2, performing image enhancement processing on the scene image to obtain a processed enhanced image;
s3, positioning the two-dimension codes in the enhanced image through the trained multi-two-dimension code positioning model according to the enhanced image to obtain all two-dimension code areas;
and S4, performing two-dimension code identification on the two-dimension code area to obtain all two-dimension code information.
In the embodiment, ZBar is used to identify the two-dimensional code, and referring to fig. 3, the ZBar technology principle is to apply a digital image processing technology to an image containing the two-dimensional code, filter the image through a series of filters to remove noise, sharpen and enhance contrast of the two-dimensional code image, and then perform edge detection and shape analysis to determine a symbol position, a symbol direction, and the like. And finally, extracting data from the original two-dimensional code image, and inputting the data into a database for storage.
Further, the method comprises a multi-two-dimension code positioning model establishing step, wherein the multi-two-dimension code positioning model establishing step specifically comprises the following steps:
acquiring a training image containing a two-dimensional code, and labeling the training image to obtain a training data set;
and training the established multi-two-dimension code positioning model according to a preset loss function and a training data set to obtain the trained multi-two-dimension code positioning model.
The SSD used in the embodiment is a convolutional neural network for target detection, and combines the regression idea of YOLO and the anchor mechanism of Faster R-CNN. The regression idea can simplify the calculation complexity of the neural network, improve the real-time performance of the algorithm, and the anchors mechanism can extract the features with different aspect ratio sizes. In addition, the SSD adopts a multi-scale target feature extraction method aiming at the characteristic that features of different scales express different features, and the design is favorable for improving the robustness of detecting the targets of different scales.
The SSD architecture is based on a feed-forward convolutional neural network, which generates a series of fixed-size bounding boxes (bounding boxes) containing objects and scores (score) representing the objects to be detected, which is the probability size of the object in a certain class, and then generates many prediction results (prediction), and then uses non-maximum suppression (NMS) to screen out the final results.
The very beginning of the SSD model is a standard architecture for image classification, where VGG-16 may be used. After this, additional auxiliary structures are added. After the infrastructure network structure, additional convolutional layers are added, the size of these convolutional layers is decreased layer by layer, each layer has a feature map with different specification, that is, a feature extraction method for the layer image, the image is divided into grids, for example, fig. 4b is a feature map of 8 × 8, and fig. 4c is a feature map of 4 × 4, where each grid is called a feature map cell. This structure allows the prediction to be performed at multiple scales.
Default boxes are a series of boxes of extracted features that the SSD model artificially sets. There are a series of default boxes on each feature map cell, each red box is a default box, there are 4 default boxes in the example, and there are several extra convolutional layers in the network with different sizes of feature maps, and the position of each default box is fixed relative to its corresponding feature map cell. In each feature map cell, we predict the offset value (offsets) between the bounding box and the default box, and score (calculated for each class probability) of the object contained in each bounding box.
Specifically, for each of the k boxes at a location, we need to compute c classes, score for each class, and 4 offsets for this box relative to its default box. Thus, on each feature map cell in the feature maps, (c +4) × k filters are required. For a feature map of m × n size, output results are generated (c +4) × k × m × n.
In the embodiment, pictures containing a plurality of two-dimensional codes are collected through videos or photographs, then manual labeling is carried out by using labelImg software, a box is drawn for each two-dimensional code in each picture, the type of the box is input, and the box is clicked for storage, so that the software can automatically generate an xml file conforming to a passacal voc picture data set format.
In the embodiment, the two-dimension code picture and various background pictures on the network are crawled by a web crawler, the position coordinates of the two-dimension code are randomly specified, the two-dimension code picture is synthesized into the background pictures, and then the xml file conforming to the format of the past voc picture data set is generated by programming.
In the embodiment, operations such as translation, rotation, mirroring, random clipping, contrast adjustment and the like are performed on the picture data set, so that the data set is enlarged, and the model is more robust.
During training, it is determined which default box corresponds to the ground transistor box, and fig. 4a includes four boxes of the two-dimensional code, that is, the box where the object to be measured is actually located. For each group channel box, we select a suitable default box from a series of default boxes with different positions, sizes and aspect ratios, and the criterion is to select the default box with the largest cross-over ratio (jaccard overlap), in other words, select the default box with the highest degree of overlap with the group channel box.
The objective of the SSD model training is measured by a loss function, which is divided into two parts, one part is localization loss and the other part is confidence loss, and the total loss function is obtained by weighted summation of the two parts, wherein the localization loss represents the deviation of the model predicted default box and the actual two-dimensional code position ground truth box, and the confidence loss represents the difference between the model predicted confidence (confidence) and the actual score (1 if two-dimensional code, or 0 if not), and is calculated by a softmax function, and the formula is as follows:
wherein N is the number of default boxes matched with the target; α is a weight, generally set to 1;
Lconfis confidence loss, which is the softmax loss of confidence for each predictor, as follows:
here, the
Whether the ith default box is matched with the jth group tree box or not is shown, if so, the matching is 1, otherwise, the matching is 0;
Llocis localization loss, which is the Smooth L1 loss between a predict box (the box where the two-dimensional code predicted by the final model is located) and a ground truth box (g), specifically, regression prediction is performed on the center (cx, cy), width (w) and height (h) of the default box:
hard negative mining: after the matching is completed, it is obvious that most default boxes are not matched with the group route boxes and are called negative samples (negative), and only a few default boxes on the match become positive samples (positive). If the ratio of the negative sample and the positive sample is not adjusted, the training is not favorable. To balance the number of positive and negative samples, we pick out a small fraction of the larger confidence loss box and discard the rest to maintain the ratio of negative to positive samples at around 3: 1.
Further, the step of acquiring the training image containing the two-dimensional code at least includes one of the following steps:
acquiring an image containing a plurality of two-dimensional codes through image acquisition equipment to obtain a training image;
the method comprises the steps of obtaining a two-dimensional code image and a background image through a web crawler, and randomly synthesizing the two-dimensional code image and the background image to obtain a training image.
Further, the image enhancement processing is performed on the multi-two-dimensional code scene image to obtain a processed enhanced image, and the step specifically includes:
and processing the multi-two-dimension code scene image by adopting a high-dynamic-range image processing algorithm of rapid bilateral filtering to obtain a processed enhanced image.
Among them, in the environment of actual industrial production, the lighting condition is an important factor affecting the image quality, and includes two cases: the image is too bright or too dark. The two situations may occur on the same image at the same time, so an algorithm capable of simultaneously processing the over-bright and over-dark situations is needed to ensure the image quality, so as to facilitate the subsequent multi-two-dimensional code image positioning and identification.
This problem is solved in the present embodiment using a fast bilateral filtering based high dynamic range image processing algorithm. The dynamic range refers to a range of brightness values that can be expressed by image pixels, for example, the dynamic range of an 8-bit image is 256 bits, and the brightness change in the real world often exceeds this range, and if the illumination is too weak, the pixel value is lower than the dynamic range during imaging, the pixel value is uniformly expressed as black, a dark region is formed on the image, which results in loss of detail information, and if the illumination is too strong, the pixel value exceeds the dynamic range, which results in a bright region, which also results in loss of detail. There is a need for image representation formats that use High Dynamic Range (HDR) to provide higher dynamic range and richer image detail.
Referring to fig. 2, an embodiment of the present invention provides a multi-target two-dimensional code detection and identification system, including:
the acquisition unit is used for acquiring a real-time multi-two-dimension code scene image;
the enhancement unit is used for carrying out image enhancement processing on the multi-two-dimensional code scene image to obtain a processed enhanced image;
the positioning unit is used for positioning a plurality of two-dimensional codes in the enhanced image through a trained multi-two-dimensional code positioning model according to the enhanced image to obtain all two-dimensional code areas;
and the identification unit is used for carrying out two-dimensional code identification on the two-dimensional code area to obtain all two-dimensional code information.
Further, the method comprises a multi-two-dimension code positioning model establishing unit, wherein the multi-two-dimension code positioning model establishing unit specifically comprises:
the data set acquisition unit is used for acquiring a training image containing the two-dimensional code and labeling the training image to obtain a training data set;
and the training unit is used for training the established multi-two-dimensional code positioning model according to a preset loss function and a training data set to obtain the trained multi-two-dimensional code positioning model.
Further, the obtaining unit at least includes one of the following:
the first acquisition unit is used for acquiring images containing a plurality of two-dimensional codes through image acquisition equipment to obtain training images;
and the second acquisition unit is used for acquiring the two-dimensional code image and the background image through a web crawler and randomly synthesizing the two-dimensional code image and the background image to obtain a training image.
Further, the enhancement unit is specifically configured to:
and processing the multi-two-dimension code scene image by adopting a high-dynamic-range image processing algorithm of rapid bilateral filtering to obtain a processed enhanced image.
In a third aspect, an embodiment of the present invention provides a multi-target two-dimensional code detection and identification device, including:
at least one processor;
at least one memory for storing at least one program;
when the at least one program is executed by the at least one processor, the at least one processor is enabled to realize the multi-target two-dimensional code detection and identification method.
In a fourth aspect, an embodiment of the present invention provides a computer storage medium, which includes a computer program, and when the computer program runs on a computer, the method for detecting and identifying a multi-target two-dimensional code is executed.
From the above contents, the method is based on deep learning, multi-target detection, multi-environment image enhancement and two-dimensional code identification technologies, is faster and simpler compared with R-CNN, and has higher accuracy compared with YOLO, the problem that even though the traditional two-dimensional code identification can position a plurality of two-dimensional codes, the real-time positioning and identification of the plurality of two-dimensional codes cannot be realized is effectively solved, and the detection efficiency is greatly improved.
While the preferred embodiments of the present invention have been illustrated and described, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. A multi-target two-dimensional code detection and identification method is characterized by comprising the following steps:
acquiring a real-time multi-two-dimension code scene image;
performing image enhancement processing on the multi-two-dimension code scene image to obtain a processed enhanced image;
positioning a plurality of two-dimensional codes in the enhanced image through a trained multi-two-dimensional code positioning model according to the enhanced image to obtain all two-dimensional code areas;
and carrying out two-dimension code identification on the two-dimension code area to obtain all two-dimension code information.
2. The multi-target two-dimensional code detection and identification method according to claim 1, characterized in that: the method comprises the steps of establishing a multi-two-dimension code positioning model, wherein the step of establishing the multi-two-dimension code positioning model specifically comprises the following steps: acquiring a training image containing a two-dimensional code, and labeling the training image to obtain a training data set;
and training the established multi-two-dimension code positioning model according to a preset loss function and a training data set to obtain the trained multi-two-dimension code positioning model.
3. The multi-target two-dimensional code detection and identification method according to claim 1, characterized in that: the step of acquiring the training image containing the two-dimensional code at least comprises one of the following steps:
acquiring an image containing a plurality of two-dimensional codes through image acquisition equipment to obtain a training image;
the method comprises the steps of obtaining a two-dimensional code image and a background image through a web crawler, and randomly synthesizing the two-dimensional code image and the background image to obtain a training image.
4. The multi-target two-dimensional code detection and identification method according to claim 1, characterized in that: the image enhancement processing is carried out on the multi-two-dimensional code scene image to obtain a processed enhanced image, and the steps are specifically as follows:
and processing the multi-two-dimension code scene image by adopting a high-dynamic-range image processing algorithm of rapid bilateral filtering to obtain a processed enhanced image.
5. A multi-target two-dimensional code detection and identification system is characterized by comprising:
the acquisition unit is used for acquiring a real-time multi-two-dimension code scene image;
the enhancement unit is used for carrying out image enhancement processing on the multi-two-dimensional code scene image to obtain a processed enhanced image;
the positioning unit is used for positioning a plurality of two-dimensional codes in the enhanced image through a trained multi-two-dimensional code positioning model according to the enhanced image to obtain all two-dimensional code areas;
and the identification unit is used for carrying out two-dimensional code identification on the two-dimensional code area to obtain all two-dimensional code information.
6. The multi-target two-dimensional code detection and identification system according to claim 5, wherein: the method comprises a multi-two-dimension code positioning model establishing unit, wherein the multi-two-dimension code positioning model establishing unit specifically comprises the following steps: the data set acquisition unit is used for acquiring a training image containing the two-dimensional code and labeling the training image to obtain a training data set;
and the training unit is used for training the established multi-two-dimensional code positioning model according to a preset loss function and a training data set to obtain the trained multi-two-dimensional code positioning model.
7. The multi-target two-dimensional code detection and identification system according to claim 5, wherein: the acquisition unit at least comprises one of the following components:
the first acquisition unit is used for acquiring images containing a plurality of two-dimensional codes through image acquisition equipment to obtain training images;
and the second acquisition unit is used for acquiring the two-dimensional code image and the background image through a web crawler and randomly synthesizing the two-dimensional code image and the background image to obtain a training image.
8. The multi-target two-dimensional code detection and identification system according to claim 5, wherein: the enhancement unit is specifically configured to:
and processing the multi-two-dimension code scene image by adopting a high-dynamic-range image processing algorithm of rapid bilateral filtering to obtain a processed enhanced image.
9. The utility model provides a multi-target two-dimensional code detects recognition device which characterized in that includes:
at least one processor;
at least one memory for storing at least one program;
when the at least one program is executed by the at least one processor, the at least one processor is enabled to realize the multi-target two-dimensional code detection and identification method as claimed in any one of claims 1 to 4.
10. A computer storage medium, comprising a computer program that, when run on a computer, causes a multi-target two-dimensional code detection recognition method according to any one of claims 1 to 4 to be executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011335084.6A CN112347805A (en) | 2020-11-25 | 2020-11-25 | Multi-target two-dimensional code detection and identification method, system, device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011335084.6A CN112347805A (en) | 2020-11-25 | 2020-11-25 | Multi-target two-dimensional code detection and identification method, system, device and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112347805A true CN112347805A (en) | 2021-02-09 |
Family
ID=74364766
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011335084.6A Pending CN112347805A (en) | 2020-11-25 | 2020-11-25 | Multi-target two-dimensional code detection and identification method, system, device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112347805A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113469937A (en) * | 2021-05-25 | 2021-10-01 | 长兴云尚科技有限公司 | Pipe gallery abnormal position positioning method and system based on pipe gallery video and two-dimensional code detection |
| CN114022558A (en) * | 2022-01-05 | 2022-02-08 | 深圳思谋信息科技有限公司 | Image positioning method and device, computer equipment and storage medium |
| CN114021596A (en) * | 2021-09-22 | 2022-02-08 | 厦门华联电子股份有限公司 | Bar code identification method and device based on deep learning |
| CN116882433A (en) * | 2023-09-07 | 2023-10-13 | 无锡维凯科技有限公司 | Machine vision-based code scanning identification method and system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109190439A (en) * | 2018-09-21 | 2019-01-11 | 南京机灵侠软件技术有限公司 | A kind of image-recognizing method of optical splitter port lines and its two-dimension code label |
| CN109800616A (en) * | 2019-01-17 | 2019-05-24 | 柳州康云互联科技有限公司 | A kind of two dimensional code positioning identification system based on characteristics of image |
| CN110941970A (en) * | 2019-12-05 | 2020-03-31 | 深圳牛图科技有限公司 | High-speed dimension code positioning and identifying system based on full convolution neural network |
| CN110991457A (en) * | 2019-11-26 | 2020-04-10 | 北京达佳互联信息技术有限公司 | Two-dimensional code processing method and device, electronic equipment and storage medium |
| CN111597847A (en) * | 2019-02-20 | 2020-08-28 | 中科院微电子研究所昆山分所 | Two-dimensional code identification method, device and equipment and readable storage medium |
-
2020
- 2020-11-25 CN CN202011335084.6A patent/CN112347805A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109190439A (en) * | 2018-09-21 | 2019-01-11 | 南京机灵侠软件技术有限公司 | A kind of image-recognizing method of optical splitter port lines and its two-dimension code label |
| CN109800616A (en) * | 2019-01-17 | 2019-05-24 | 柳州康云互联科技有限公司 | A kind of two dimensional code positioning identification system based on characteristics of image |
| CN111597847A (en) * | 2019-02-20 | 2020-08-28 | 中科院微电子研究所昆山分所 | Two-dimensional code identification method, device and equipment and readable storage medium |
| CN110991457A (en) * | 2019-11-26 | 2020-04-10 | 北京达佳互联信息技术有限公司 | Two-dimensional code processing method and device, electronic equipment and storage medium |
| CN110941970A (en) * | 2019-12-05 | 2020-03-31 | 深圳牛图科技有限公司 | High-speed dimension code positioning and identifying system based on full convolution neural network |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113469937A (en) * | 2021-05-25 | 2021-10-01 | 长兴云尚科技有限公司 | Pipe gallery abnormal position positioning method and system based on pipe gallery video and two-dimensional code detection |
| CN114021596A (en) * | 2021-09-22 | 2022-02-08 | 厦门华联电子股份有限公司 | Bar code identification method and device based on deep learning |
| CN114022558A (en) * | 2022-01-05 | 2022-02-08 | 深圳思谋信息科技有限公司 | Image positioning method and device, computer equipment and storage medium |
| CN114022558B (en) * | 2022-01-05 | 2022-08-26 | 深圳思谋信息科技有限公司 | Image positioning method, image positioning device, computer equipment and storage medium |
| CN116882433A (en) * | 2023-09-07 | 2023-10-13 | 无锡维凯科技有限公司 | Machine vision-based code scanning identification method and system |
| CN116882433B (en) * | 2023-09-07 | 2023-12-08 | 无锡维凯科技有限公司 | Machine vision-based code scanning identification method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11681418B2 (en) | Multi-sample whole slide image processing in digital pathology via multi-resolution registration and machine learning | |
| CN112347805A (en) | Multi-target two-dimensional code detection and identification method, system, device and storage medium | |
| CN112150493B (en) | Semantic guidance-based screen area detection method in natural scene | |
| CN112614136B (en) | Infrared small target real-time instance segmentation method and device | |
| CN112215795B (en) | Intelligent detection method for server component based on deep learning | |
| CN113052170B (en) | Small target license plate recognition method under unconstrained scene | |
| CN113361645B (en) | Target detection model construction method and system based on meta learning and knowledge memory | |
| CN111539957A (en) | Image sample generation method, system and detection method for target detection | |
| CN116051820A (en) | Single target detection method based on multiple templates | |
| CN114781514A (en) | Floater target detection method and system integrating attention mechanism | |
| CN114283431B (en) | Text detection method based on differentiable binarization | |
| CN115482529A (en) | Method, equipment, storage medium and device for recognizing fruit image in near scene | |
| CN116543325A (en) | Unmanned aerial vehicle image-based crop artificial intelligent automatic identification method and system | |
| CN115115950A (en) | Unmanned aerial vehicle image duplicate checking method based on image histogram features | |
| CN111597875A (en) | Traffic sign identification method, device, equipment and storage medium | |
| CN111950357A (en) | Marine water surface garbage rapid identification method based on multi-feature YOLOV3 | |
| CN109165592B (en) | Real-time rotatable face detection method based on PICO algorithm | |
| CN118212572A (en) | Road damage detection method based on improvement YOLOv7 | |
| CN110766001B (en) | Bank card number positioning and end-to-end identification method based on CNN and RNN | |
| CN115861922B (en) | Sparse smoke detection method and device, computer equipment and storage medium | |
| CN114445689A (en) | Multi-scale weighted fusion target detection method and system guided by target prior information | |
| CN113128492A (en) | Bill text positioning method and device | |
| CN118379696B (en) | Ship target detection method and device and readable storage medium | |
| CN112419227B (en) | Underwater target detection method and system based on small target search scaling technology | |
| CN115205853B (en) | Image-based citrus fruit detection and identification method and system |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210209 |
|
| RJ01 | Rejection of invention patent application after publication |