CN112348024A - Image-text identification method and system based on deep learning optimization network - Google Patents
Image-text identification method and system based on deep learning optimization network Download PDFInfo
- Publication number
- CN112348024A CN112348024A CN202011178476.6A CN202011178476A CN112348024A CN 112348024 A CN112348024 A CN 112348024A CN 202011178476 A CN202011178476 A CN 202011178476A CN 112348024 A CN112348024 A CN 112348024A
- Authority
- CN
- China
- Prior art keywords
- deep learning
- recognition
- text
- picture
- correction
- 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
- 238000000034 method Methods 0.000 title claims abstract description 67
- 238000013135 deep learning Methods 0.000 title claims abstract description 58
- 238000005457 optimization Methods 0.000 title claims abstract description 24
- 238000003058 natural language processing Methods 0.000 claims abstract description 52
- 238000012937 correction Methods 0.000 claims abstract description 45
- 238000001514 detection method Methods 0.000 claims abstract description 15
- 238000005516 engineering process Methods 0.000 claims abstract description 13
- 238000012545 processing Methods 0.000 claims abstract description 10
- 238000006748 scratching Methods 0.000 claims abstract description 10
- 230000002393 scratching effect Effects 0.000 claims abstract description 10
- 230000008569 process Effects 0.000 claims description 30
- 238000013528 artificial neural network Methods 0.000 claims description 21
- 238000012549 training Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 4
- 238000012015 optical character recognition Methods 0.000 abstract description 46
- 238000013527 convolutional neural network Methods 0.000 description 6
- 238000010276 construction Methods 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/14—Image acquisition
- G06V30/148—Segmentation of character regions
- G06V30/153—Segmentation of character regions using recognition of characters or words
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Evolutionary Biology (AREA)
- Multimedia (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Character Discrimination (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method and a system for recognizing graphics and texts based on a deep learning optimization network, which belong to the technical field of optical character recognition and are characterized in that: at least comprises the following steps: the method comprises the following steps: identifying an object in a single-frame image through a deep learning target detection technology; step two: the method comprises the following steps of (1) scratching out a picture of an object through a scratching model and an aligning model, and aligning; step three: performing OCR recognition on the whole picture; step four: and sending the character recognition result obtained by OCR recognition into an NLP correction model established based on deep learning natural language processing for correction, and finally outputting the character recognition result. The invention can quickly identify the processing technology of the photo and the video of the whole block of characters by establishing an inaccurate text correction model by means of a deep learning target detection technology, and can mark the whole block of characters in the whole photo or the whole frame of video, thereby saving system resources of OCR processing and greatly improving character identification efficiency.
Description
Technical Field
The invention belongs to the technical field of optical character recognition, and particularly relates to a text-text recognition method and system based on a deep learning optimization network.
Background
As is well known, OCR (Optical Character Recognition) refers to a process in which an electronic device (e.g., a scanner or a digital camera) checks a Character printed on paper, determines its shape by detecting dark and light patterns, and then translates the shape into a computer text by a Character Recognition method; the method is characterized in that characters in a paper document are converted into an image file with a black-white dot matrix in an optical mode aiming at print characters, and the characters in the image are converted into a text format through recognition software for further editing and processing by word processing software. How to debug or use auxiliary information to improve recognition accuracy is the most important issue of OCR, and the term of icr (intelligent Character recognition) is generated accordingly. The main indicators for measuring the performance of an OCR system are: the rejection rate, the false recognition rate, the recognition speed, the user interface friendliness, the product stability, the usability, the feasibility and the like.
Referring to fig. 1, in the conventional OCR recognition technology, a single text block is usually found first, and the single text block is usually numerous and many small blocks are spliced, which results in a great waste of system resources and greatly reduces the text recognition efficiency.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a method and a system for recognizing pictures and texts based on a deep learning optimization network, which can quickly recognize the processing technology of photos and videos of a whole block of characters by establishing an inaccurate text correction model by means of a deep learning target detection technology, and can mark the whole block of characters in the whole photo or the whole frame of video, thereby saving the system resources of OCR processing and greatly improving the character recognition efficiency.
One of the purposes of the invention is to provide an image-text identification method based on a deep learning optimization network, which comprises the following steps:
the method comprises the following steps: identifying an object in a single-frame image through a deep learning target detection technology;
step two: the method comprises the following steps of (1) scratching out a picture of an object through a scratching model and an aligning model, and aligning;
step three: performing OCR recognition on the whole picture;
step four: and sending the character recognition result obtained by OCR recognition into an NLP correction model established based on deep learning natural language processing for correction, and finally outputting the character recognition result.
Preferably, the specific steps of establishing the NLP correction model based on deep learning natural language processing are as follows:
firstly, initializing a deep artificial neural network by utilizing a corpus accumulated in the early stage;
then, the sorted whole block is used for carrying out recognition process information of OCR recognition, error information of an input text manually corrected by NLP, relevant information of a correction process record, and the text which is used as a data set and input into aligned target object information and has low accuracy is used for training, and weight adjustment is carried out on the deep artificial neural network through a reasonably set loss function.
Preferably, the single frame image is a single picture in a photo album or a single frame picture in a video.
The invention also provides a system for identifying graphics and texts based on deep learning optimization network, which at least comprises:
an object identification module: identifying an object in a single-frame image through a deep learning target detection technology;
an alignment module: the method comprises the following steps of (1) scratching out a picture of an object through a scratching model and an aligning model, and aligning;
an OCR recognition module: performing OCR recognition on the whole picture;
a correction module: and sending the character recognition result obtained by OCR recognition into an NLP correction model established based on deep learning natural language processing for correction, and finally outputting the character recognition result.
Preferably, the specific steps of establishing the NLP correction model based on deep learning natural language processing are as follows:
firstly, initializing a deep artificial neural network by utilizing a corpus accumulated in the early stage;
then, the sorted whole block is used for carrying out recognition process information of OCR recognition, error information of an input text manually corrected by NLP, relevant information of a correction process record, and the text which is used as a data set and input into aligned target object information and has low accuracy is used for training, and weight adjustment is carried out on the deep artificial neural network through a reasonably set loss function.
Preferably, the single frame image is a single picture of a photo album or a single frame picture in a video.
The invention also aims to provide a computer program for realizing the image-text recognition method based on the deep learning optimization network.
The fourth purpose of the invention is to provide an information data processing terminal for realizing the image-text identification method based on the deep learning optimization network.
The fifth object of the present invention is to provide a computer-readable storage medium, which includes instructions that, when executed on a computer, cause the computer to execute a deep learning optimization network-based text-text recognition method.
In summary, the advantages and positive effects of the invention are:
by using the technical scheme of the invention, the processing technology of the photo and the video of the whole block of characters can be rapidly identified, and the whole block of characters in the whole photo or the whole frame of video can be marked, thereby saving the system resource of OCR processing and greatly improving the character identification efficiency.
Drawings
FIG. 1 is a flow chart of a conventional solution;
FIG. 2 is a flow chart of a preferred embodiment of the present invention;
FIG. 3 is a flow chart of the establishment of the NLP correction model in the preferred embodiment of the present invention;
fig. 4 is a flow chart of NLP application in the preferred embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 2, a method for identifying an image-text based on a deep learning optimization network includes the following steps:
1) objects are first identified in an image or video frame by a deep learning object detection technique.
2) And through a plurality of background target object matting models and alignment models, each identified object matting picture is aligned.
3) And performing OCR recognition on the aligned object picture instead of single character OCR recognition. In the process, because a large amount of interference and distortion exist in the OCR recognition process of the whole picture, the recognized characters can be extremely undesirable.
4) And (3) sending the inaccurate character recognition result in the previous step into an NLP (Natural Language Processing) correction model established based on deep learning Natural Language Processing for correction, and finally outputting a more accurate character recognition result.
Wherein: the specific steps of establishing the NLP correction model based on deep learning natural language processing are as follows:
1) first, a corpus accumulated in a previous stage is used to initialize a deep artificial neural network (DNN).
2) And training the recognition process information of OCR recognition, the error information of the input text manually corrected by NLP, the relevant information of the correction process record, the aligned target object information input as a data set and the text with low accuracy by using the sorted whole block, and adjusting the weight of DNN by using a reasonably set loss function.
The process of using the trained "NLP correction model built based on deep learning natural language processing" is shown in fig. 4: and inputting the image or video frame needing character recognition as an input into the trained DNN, and outputting corrected and more accurate text by the DNN.
The artificial neural network (DNN) referred to in fig. 3 and 4 of the present invention includes, but is not limited to, the following networks or a combination of networks: CNN (convolutional neural network), RNN (Recurrent neural network), GAN (generic adaptive network generation countermeasure network), LSTM (Long Short-Term Memory), etc., inclusive subnetworks including but not limited to a combination of the following methods: R-CNN (Region-CNN, meaning of CNN is described above), fast-RCNN (RCNN is the same as R-CNN, meaning is described above), MASK-RCNN, etc. Such networks or sub-networks, with or without attentional mechanisms, are encompassed by the present invention.
1. The invention provides a technology for realizing character recognition in an image or a video through deep learning target detection. It is completely different from the traditional OCR technology in recognizing characters in images or videos.
2. The traditional OCR method directly carries out the character block recognition step on the image or the video frame, a target detection method is not used for recognizing specific objects (except the situation of recognizing the character blocks), and the method for recognizing the specific objects by using the target detection method in the OCR task is the innovation of the invention. Are within the scope of the invention.
3. Since the object is not recognized, the conventional OCR method will not scratch out and align the specific object image in the image (the operations of rotating, aligning, etc. the whole image are not listed here).
4. The traditional OCR method only focuses on each single character in an image or a video frame, and does not combine with object information to perform whole block recognition on the characters in an object.
The invention designs a special deep neural network construction method aiming at detecting a whole block of recognized characters by combining a target, the construction steps, the network structure or the substructure, and the training and application methods of the deep neural network construction method are innovative, and for image-text recognition tasks and OCR tasks in images or video frames, if the steps similar to the invention are used, and if the artificial neural network structure and the artificial neural network training application method similar to the invention shown in the figures 3 and 4 are used when the artificial neural network is established, the deep neural network construction method is within the protection scope of the patent of the invention.
Referring to fig. 3, a second preferred embodiment is a system for identifying texts based on a deep learning optimization network, including:
an object identification module: objects are first identified in an image or video frame by a deep learning object detection technique.
An alignment module: and through a plurality of background target object matting models and alignment models, each identified object matting picture is aligned.
An OCR recognition module: and performing OCR recognition on the aligned object picture instead of single character OCR recognition. In the process, because a large amount of interference and distortion exist in the OCR recognition process of the whole picture, the recognized characters can be extremely undesirable.
A correction module: and (4) sending the inaccurate character recognition result in the last step into an NLP correction model established based on deep learning natural language processing for correction, and finally outputting a more accurate character recognition result.
Wherein: the specific steps of establishing the NLP correction model based on deep learning natural language processing are as follows:
1) first, a corpus accumulated in a previous stage is used to initialize a deep artificial neural network (DNN).
2) And training the recognition process information of OCR recognition, the error information of the input text manually corrected by NLP, the relevant information of the correction process record, the aligned target object information input as a data set and the text with low accuracy by using the sorted whole block, and adjusting the weight of DNN by using a reasonably set loss function.
In a third preferred embodiment, a computer program for implementing a deep learning optimization network-based image-text recognition method includes the following steps:
1) objects are first identified in an image or video frame by a deep learning object detection technique.
2) And through a plurality of background target object matting models and alignment models, each identified object matting picture is aligned.
3) And performing OCR recognition on the aligned object picture instead of single character OCR recognition. In the process, because a large amount of interference and distortion exist in the OCR recognition process of the whole picture, the recognized characters can be extremely undesirable.
4) And (4) sending the inaccurate character recognition result in the last step into an NLP correction model established based on deep learning natural language processing for correction, and finally outputting a more accurate character recognition result.
Wherein: the specific steps of establishing the NLP correction model based on deep learning natural language processing are as follows:
1) first, initialization of a deep artificial neural network (DNN) is performed using a corpus accumulated in the early stage.
2) And training the recognition process information of OCR recognition, the error information of the input text manually corrected by NLP, the relevant information of the correction process record, the aligned target object information input as a data set and the text with low accuracy by using the sorted whole block, and adjusting the weight of DNN by using a reasonably set loss function.
The fourth preferred embodiment is an information data processing terminal for realizing the image-text identification method based on the deep learning optimization network. The image-text identification method based on the deep learning optimization network comprises the following steps:
1) objects are first identified in an image or video frame by a deep learning object detection technique.
2) And through a plurality of background target object matting models and alignment models, each identified object matting picture is aligned.
3) And performing OCR recognition on the aligned object picture instead of single character OCR recognition. In the process, because a large amount of interference and distortion exist in the OCR recognition process of the whole picture, the recognized characters can be extremely undesirable.
4) And (4) sending the inaccurate character recognition result in the last step into an NLP correction model established based on deep learning natural language processing for correction, and finally outputting a more accurate character recognition result.
Wherein: the specific steps of establishing the NLP correction model based on deep learning natural language processing are as follows:
1) first, initialization of a deep artificial neural network (DNN) is performed using a corpus accumulated in the early stage.
2) And training the recognition process information of OCR recognition, the error information of the input text manually corrected by NLP, the relevant information of the correction process record, the aligned target object information input as a data set and the text with low accuracy by using the sorted whole block, and adjusting the weight of DNN by using a reasonably set loss function.
A fifth preferred embodiment is a computer-readable storage medium comprising instructions which, when executed on a computer, cause the computer to perform the following method for deep learning optimized network based teletext recognition:
1) objects are first identified in an image or video frame by a deep learning object detection technique.
2) And through a plurality of background target object matting models and alignment models, each identified object matting picture is aligned.
3) And performing OCR recognition on the aligned object picture instead of single character OCR recognition. In the process, because a large amount of interference and distortion exist in the OCR recognition process of the whole picture, the recognized characters can be extremely undesirable.
4) And (4) sending the inaccurate character recognition result in the last step into an NLP correction model established based on deep learning natural language processing for correction, and finally outputting a more accurate character recognition result.
Wherein: the specific steps of establishing the NLP correction model based on deep learning natural language processing are as follows:
1) first, a corpus accumulated in a previous stage is used to initialize a deep artificial neural network (DNN).
2) And training the recognition process information of OCR recognition, the error information of the input text manually corrected by NLP, the relevant information of the correction process record, the aligned target object information input as a data set and the text with low accuracy by using the sorted whole block, and adjusting the weight of DNN by using a reasonably set loss function.
In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When used in whole or in part, can be implemented in a computer program product that includes one or more computer instructions. When loaded or executed on a computer, cause the flow or functions according to embodiments of the invention to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, the computer instructions may be transmitted from one website site, computer, server, or data center to another website site, computer, server, or data center via wire (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL), or wireless (e.g., infrared, wireless, microwave, etc.)). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that includes one or more of the available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., Solid State Disk (SSD)), among others.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (9)
1. A graphic identification method based on a deep learning optimization network is characterized in that: at least comprises the following steps:
the method comprises the following steps: identifying an object in a single-frame image through a deep learning target detection technology;
step two: the method comprises the following steps of (1) scratching out a picture of an object through a scratching model and an aligning model, and aligning;
step three: performing OCR recognition on the whole picture;
step four: and sending the character recognition result obtained by OCR recognition into an NLP correction model established based on deep learning natural language processing for correction, and finally outputting the character recognition result.
2. The image-text recognition method based on the deep learning optimization network of claim 1 is characterized in that the specific steps of establishing the NLP correction model based on the deep learning natural language processing are as follows:
firstly, initializing a deep artificial neural network by utilizing a corpus accumulated in the early stage;
then, the sorted whole block is used for carrying out recognition process information of OCR recognition, error information of an input text manually corrected by NLP, relevant information of a correction process record, and the text which is used as a data set and input into aligned target object information and has low accuracy is used for training, and weight adjustment is carried out on the deep artificial neural network through a reasonably set loss function.
3. The image-text recognition method based on the deep learning optimization network of claim 1 or 2, wherein the single-frame image is a single picture in a photo set or a single picture in a video.
4. A picture and text recognition system based on deep learning optimization network is characterized in that: at least comprises the following steps:
an object identification module: identifying an object in a single-frame image through a deep learning target detection technology;
an alignment module: the method comprises the following steps of (1) scratching out a picture of an object through a scratching model and an aligning model, and aligning;
an OCR recognition module: performing OCR recognition on the whole picture;
a correction module: and sending the character recognition result obtained by OCR recognition into an NLP correction model established based on deep learning natural language processing for correction, and finally outputting the character recognition result.
5. The deep learning optimization network-based image-text recognition system based on claim 4 is characterized in that the specific steps of establishing the NLP correction model based on deep learning natural language processing are as follows:
firstly, initializing a deep artificial neural network by utilizing a corpus accumulated in the early stage;
then, the sorted whole block is used for carrying out recognition process information of OCR recognition, error information of an input text manually corrected by NLP, relevant information of a correction process record, and the text which is used as a data set and input into aligned target object information and has low accuracy is used for training, and weight adjustment is carried out on the deep artificial neural network through a reasonably set loss function.
6. The deep learning optimization network-based image-text recognition system based on claim 4 or 5, wherein the single frame image is a single picture in a photo set or a single frame picture in a video.
7. A computer program for implementing the deep learning optimization network-based teletext recognition method according to any one of claims 1-3.
8. An information data processing terminal for implementing the image-text identification method based on the deep learning optimization network of any one of claims 1 to 3.
9. A computer-readable storage medium comprising instructions which, when executed on a computer, cause the computer to perform the deep learning optimization network-based teletext recognition method according to any one of claims 1-3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011178476.6A CN112348024A (en) | 2020-10-29 | 2020-10-29 | Image-text identification method and system based on deep learning optimization network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011178476.6A CN112348024A (en) | 2020-10-29 | 2020-10-29 | Image-text identification method and system based on deep learning optimization network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112348024A true CN112348024A (en) | 2021-02-09 |
Family
ID=74355852
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011178476.6A Pending CN112348024A (en) | 2020-10-29 | 2020-10-29 | Image-text identification method and system based on deep learning optimization network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112348024A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117037049A (en) * | 2023-10-10 | 2023-11-10 | 武汉博特智能科技有限公司 | Image content detection method and system based on YOLOv5 deep learning |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20090132482A (en) * | 2008-06-19 | 2009-12-30 | 삼성전자주식회사 | Character recognition method and apparatus |
| CN105740858A (en) * | 2016-01-26 | 2016-07-06 | 南京风力舰信息技术有限公司 | Region-of-interest extraction based image copy detection method |
| CN110348346A (en) * | 2019-06-28 | 2019-10-18 | 苏宁云计算有限公司 | A kind of bill classification recognition methods and system |
| CN110501018A (en) * | 2019-08-13 | 2019-11-26 | 广东星舆科技有限公司 | A kind of traffic mark board information collecting method for serving high-precision map producing |
| CN110866457A (en) * | 2019-10-28 | 2020-03-06 | 世纪保众(北京)网络科技有限公司 | Electronic insurance policy obtaining method and device, computer equipment and storage medium |
| US20200159820A1 (en) * | 2018-11-15 | 2020-05-21 | International Business Machines Corporation | Extracting structured information from a document containing filled form images |
| CN111291726A (en) * | 2020-03-12 | 2020-06-16 | 泰康保险集团股份有限公司 | Medical bill sorting method, device, equipment and medium |
| CN111368852A (en) * | 2018-12-26 | 2020-07-03 | 沈阳新松机器人自动化股份有限公司 | Article identification and pre-sorting system and method based on deep learning and robot |
| CN111428656A (en) * | 2020-03-27 | 2020-07-17 | 信雅达系统工程股份有限公司 | Mobile terminal identity card identification method based on deep learning and mobile device |
| KR102149051B1 (en) * | 2020-04-24 | 2020-08-28 | 주식회사 애자일소다 | System and method for analyzing document using self confidence based on ocr |
| WO2020173008A1 (en) * | 2019-02-27 | 2020-09-03 | 平安科技（深圳）有限公司 | Text recognition method and apparatus |
| CN111814797A (en) * | 2020-07-13 | 2020-10-23 | 邓兴尧 | Picture character recognition method and device and computer readable storage medium |
-
2020
- 2020-10-29 CN CN202011178476.6A patent/CN112348024A/en active Pending
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20090132482A (en) * | 2008-06-19 | 2009-12-30 | 삼성전자주식회사 | Character recognition method and apparatus |
| CN105740858A (en) * | 2016-01-26 | 2016-07-06 | 南京风力舰信息技术有限公司 | Region-of-interest extraction based image copy detection method |
| US20200159820A1 (en) * | 2018-11-15 | 2020-05-21 | International Business Machines Corporation | Extracting structured information from a document containing filled form images |
| CN111368852A (en) * | 2018-12-26 | 2020-07-03 | 沈阳新松机器人自动化股份有限公司 | Article identification and pre-sorting system and method based on deep learning and robot |
| WO2020173008A1 (en) * | 2019-02-27 | 2020-09-03 | 平安科技（深圳）有限公司 | Text recognition method and apparatus |
| CN110348346A (en) * | 2019-06-28 | 2019-10-18 | 苏宁云计算有限公司 | A kind of bill classification recognition methods and system |
| CN110501018A (en) * | 2019-08-13 | 2019-11-26 | 广东星舆科技有限公司 | A kind of traffic mark board information collecting method for serving high-precision map producing |
| CN110866457A (en) * | 2019-10-28 | 2020-03-06 | 世纪保众(北京)网络科技有限公司 | Electronic insurance policy obtaining method and device, computer equipment and storage medium |
| CN111291726A (en) * | 2020-03-12 | 2020-06-16 | 泰康保险集团股份有限公司 | Medical bill sorting method, device, equipment and medium |
| CN111428656A (en) * | 2020-03-27 | 2020-07-17 | 信雅达系统工程股份有限公司 | Mobile terminal identity card identification method based on deep learning and mobile device |
| KR102149051B1 (en) * | 2020-04-24 | 2020-08-28 | 주식회사 애자일소다 | System and method for analyzing document using self confidence based on ocr |
| CN111814797A (en) * | 2020-07-13 | 2020-10-23 | 邓兴尧 | Picture character recognition method and device and computer readable storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 周珍娟;韩金华;: "舰船遥感图像的目标识别研究", 舰船科学技术, no. 12, 15 December 2014 (2014-12-15) * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117037049A (en) * | 2023-10-10 | 2023-11-10 | 武汉博特智能科技有限公司 | Image content detection method and system based on YOLOv5 deep learning |
| CN117037049B (en) * | 2023-10-10 | 2023-12-15 | 武汉博特智能科技有限公司 | Image content detection method and system based on YOLOv5 deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10832046B1 (en) | Systems and methods for processing document images | |
| CN112766255A (en) | Optical character recognition method, device, equipment and storage medium | |
| Kaur | Text recognition applications for mobile devices | |
| US8773733B2 (en) | Image capture device for extracting textual information | |
| KR102610456B1 (en) | Object recognition methods and devices, electronic devices, and storage media | |
| Tymoshenko et al. | Real-Time Ukrainian Text Recognition and Voicing. | |
| CN111539417A (en) | Text recognition training optimization method based on deep neural network | |
| CN108304815A (en) | A kind of data capture method, device, server and storage medium | |
| CN112348024A (en) | Image-text identification method and system based on deep learning optimization network | |
| US8768058B2 (en) | System for extracting text from a plurality of captured images of a document | |
| WO2013177240A1 (en) | Textual information extraction method using multiple images | |
| US20190073571A1 (en) | Method for improving quality of recognition of a single frame | |
| CN113971810A (en) | Document generation method, device, platform, electronic equipment and storage medium | |
| Pattnaik et al. | A Framework to Detect Digital Text Using Android Based Smartphone | |
| CN111046864A (en) | Method and system for automatically extracting five elements of contract scanning piece | |
| CN111556251A (en) | Electronic book generation method, device and medium | |
| JP2012049860A (en) | Image processor, image processing method and program | |
| CN112100630A (en) | Identification method for confidential document | |
| US11995905B2 (en) | Object recognition method and apparatus, and electronic device and storage medium | |
| US11451695B2 (en) | System and method to configure an image capturing device with a wireless network | |
| TWI703504B (en) | Serial number detecting system | |
| HR | OCR Oriented Reading System for Blind People | |
| CN115147853A (en) | OCR system and method for dynamically analyzing form image characteristics | |
| KR20230062260A (en) | Method, apparatus, system and computer program for high quality transformation model learning for document image and document recognition | |
| Ilin et al. | Fast words boundaries localization in text fields for low quality document images |
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 |