CN110782402A - Method for deblurring invoice text - Google Patents
Method for deblurring invoice text Download PDFInfo
- Publication number
- CN110782402A CN110782402A CN201910924838.2A CN201910924838A CN110782402A CN 110782402 A CN110782402 A CN 110782402A CN 201910924838 A CN201910924838 A CN 201910924838A CN 110782402 A CN110782402 A CN 110782402A
- Authority
- CN
- China
- Prior art keywords
- fuzzy
- invoice
- clear
- text information
- invoice text
- 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 32
- 238000013528 artificial neural network Methods 0.000 claims abstract description 13
- 230000000694 effects Effects 0.000 claims abstract description 10
- 238000012545 processing Methods 0.000 claims abstract description 10
- 238000005516 engineering process Methods 0.000 claims abstract description 9
- 238000012549 training Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000013135 deep learning Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 7
- 238000001914 filtration Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000009467 reduction Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/73—Deblurring; Sharpening
-
- 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
-
- 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
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20172—Image enhancement details
- G06T2207/20201—Motion blur correction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30176—Document
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Character Input (AREA)
Abstract
The invention discloses a fuzzy invoice identification technical field based on a computer, in particular to an invoice text deblurring method, which comprises the following steps: s1: making clear and fuzzy one-to-one corresponding invoice text information: firstly, processing clear invoice text information by a fuzzy effect through a traditional computer vision technology; s2: calculating a fuzzy core of fuzzy invoice text information; s3: the fuzzy check is converted into clear invoice text information, the deblurring capability is strong, and compared with the traditional method, the method can effectively convert the real fuzzy information into clear information; simulating real fuzzy style data as much as possible by an unsupervised neural network algorithm, thereby calculating a fuzzy core of a fuzzy invoice text, and finally converting the fuzzy invoice text information into clear invoice text information by using the fuzzy core; the application range is wide, and the deblurring processing can be performed on different fuzzy conditions of the invoice.
Description
Technical Field
The invention relates to the technical field of fuzzy invoice recognition based on a computer, in particular to an invoice text deblurring method.
Background
In the process of bill identification of invoice data, the key text of the invoice is identified wrongly due to the problems that the invoice is not printed clearly or is shielded by a red stamp and the like by a photographing technology. How to convert the complicated and fuzzy invoice texts into clear text is the key of the invoice text deblurring technology.
One of the existing methods is to use the traditional computer vision technology to adjust the brightness and contrast of text data to improve the significance of text information in the background, but the processing is ineffective to the complex and unknown fuzzy conditions of motion blur, Gaussian blur, unclear invoice printing, red seal shielding and the like caused by photographing.
The second method is to carry out noise reduction processing through different fuzzy filtering, but the fuzzy situation is very complex in reality, and the processing can be successfully carried out without combining one or more filtering.
Disclosure of Invention
The invention aims to provide a method for deblurring an invoice text, which aims to solve the problems that the background technology provides a method for adjusting the brightness and contrast of text data by using the traditional computer vision technology, the motion blur, Gaussian blur, unclear invoice printing, red seal shielding and other complex and unknown blurring conditions are invalid, and the deblurring effect is weak in invoice fuzzy text application by performing noise reduction treatment through different fuzzy filtering, so that the method is not suitable for commercial use.
In order to achieve the purpose, the invention provides the following technical scheme: an invoice text deblurring method comprises the following specific steps:
s1: making clear and fuzzy one-to-one corresponding invoice text information: firstly, processing clear invoice text information by a fuzzy effect through a traditional computer vision technology;
cutting out a complete red seal from clear invoice text information, and calculating a binary image of the red seal by a traditional method;
then, a screenshot A of a stamp and a screenshot B of a corresponding area on the red stamp are intercepted from clear invoice text information, and then the background of the screenshot B on the intercepted red stamp is replaced by the screenshot A on the invoice text information to obtain a replaced graph C;
finally, fusing the replaced graph C with the intercepted screenshot A to obtain a graph D, and splicing the graph D back;
the method comprises the steps that an unsupervised neural network algorithm is adopted, real data are divided into three types of clear, real and fuzzy and real red seal connection, and then clear text data are learned and converted into fuzzy data and the style of red seal shielding is fuzzy through the algorithm;
then obtaining a large amount of clear invoice text information and fuzzy invoice text information corresponding to each other, wherein the data are applied to the following calculation fuzzy core;
s2: calculating a fuzzy core of fuzzy invoice text information: acquiring the clear invoice text information in the step S1 and fuzzy invoice text information corresponding to each other, and continuously learning, continuously calculating and adjusting the fuzzy core from the corresponding clear information through the fuzzy information by adopting a supervised deep learning neural network;
finally, a fitted fuzzy kernel is obtained through continuous training;
s3: and (3) converting fuzzy verification into clear invoice text information: inputting an unknown invoice text message, and calling the fuzzy core calculated in the last step to obtain clear invoice text message.
Preferably, the blurring effect in step S1 includes noise, defocus blur, mirror blur, motion blur, snow, fog, down-sampling, median blur, and the like.
Preferably, the unknown invoice text information in step S3 is clear unknown invoice text information or fuzzy unknown invoice text information.
Compared with the prior art, the invention has the beneficial effects that:
1) the deblurring capability is strong, and compared with the traditional method, the method can effectively convert the real fuzzy information into clear information;
2) simulating real fuzzy style data as much as possible by an unsupervised neural network algorithm, calculating a fuzzy core of a fuzzy invoice text by using a deep convolution and deconvolution neural network algorithm, and finally converting the fuzzy invoice text information into clear invoice text information by using the fuzzy core;
3) the application range is wide, and the deblurring processing can be performed on different fuzzy conditions of the invoice.
Drawings
FIG. 1 is a flow chart of invoice text deblurring according to the present invention;
FIG. 2 is an operational diagram of the present invention with blurring effect;
FIG. 3 is a binary diagram of a red stamp according to the present invention;
FIG. 4 is an operational diagram of an alternative background of the present invention;
FIG. 5 is an operational diagram of a splice screenshot of the present invention;
FIG. 6 is a schematic diagram of an unsupervised neural network algorithm process of the present invention;
FIG. 7 is a schematic diagram of the present invention to obtain fuzzy kernels;
FIG. 8 is a schematic diagram of the deblurring process of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In the description of the present invention, it is to be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", and the like, indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, are merely for convenience in describing the present invention and simplifying the description, and do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be construed as limiting the present invention.
Example (b):
referring to fig. 1-8, the present invention provides a technical solution: an invoice text deblurring method comprises the following specific steps:
s1: making clear and fuzzy one-to-one corresponding invoice text information: firstly, carrying out fuzzy effect processing on clear invoice text information through a traditional computer vision technology, wherein the fuzzy effect processing mode comprises noise, defocus fuzzy, mirror image fuzzy, dynamic fuzzy, snowing, heavy fog, down sampling, median fuzzy and the like, and is shown in figure 2;
in order to increase the effect of stamping a red stamp on a clear picture, a complete red stamp is cut out from clear invoice text information, and a binary image of the red stamp is calculated by a traditional method, as shown in fig. 3;
then, a screenshot A of a stamp is intercepted from clear invoice text information, a screenshot B of a corresponding area is intercepted on the red stamp, then the background of the screenshot B on the intercepted red stamp is replaced by the screenshot A on the invoice text information, and a replaced graph C is obtained, wherein the screenshot A is shown in figure 4;
finally, fusing the replaced graph C with the intercepted screenshot A to obtain a graph D, and splicing the graph D back as shown in FIG. 5;
although fuzzy data manufactured by the traditional method is relatively close to real fuzzy data, human factors are too large, the fuzzy situation cannot be simulated well completely, in order to better generate the data which is most like the real fuzzy data, an unsupervised neural network algorithm is adopted, the real data is divided into three types of clear, real and fuzzy data connected with real red chapters, and then the clear text data is learned and converted into fuzzy data and fuzzy style of red chapter shielding by the algorithm, as shown in fig. 6;
then obtaining a large amount of clear invoice text information and fuzzy invoice text information corresponding to one, applying the data to the following calculation fuzzy kernel, and simulating real fuzzy style data corresponding to one as much as possible through an unsupervised neural network algorithm;
s2: calculating a fuzzy core of fuzzy invoice text information: acquiring the clear invoice text information in the step S1 and fuzzy invoice text information corresponding to each other, and continuously learning, continuously calculating and adjusting the fuzzy core from the corresponding clear information through the fuzzy information by adopting a supervised deep learning neural network;
finally, a fitted fuzzy kernel is obtained through continuous training, as shown in fig. 7;
s3: and (3) converting fuzzy verification into clear invoice text information: inputting an unknown invoice text message, wherein the unknown invoice text message is clear unknown invoice text message or fuzzy unknown invoice text message, and obtaining clear invoice text message after calling the fuzzy core calculated in the last step, as shown in fig. 8.
The deblurring of the invoice text is a very complicated problem because the fuzzy types in reality are very complicated and are difficult to simulate artificially. Thus, deblurring invoice text information is analogous to a blind super-resolution problem. After simulating the data of the real fuzzy style as much as possible through an unsupervised neural network algorithm, calculating a fuzzy core of the fuzzy invoice text by utilizing a convolution-deconvolution neural network, and finally converting the fuzzy invoice text information into clear invoice text information by utilizing the fuzzy core.
Compared with the traditional method, the method can effectively convert the real fuzzy information into clear information.
While there have been shown and described the fundamental principles and essential features of the invention and advantages thereof, it will be apparent to those skilled in the art that the invention is not limited to the details of the foregoing exemplary embodiments, but is capable of other specific forms without departing from the spirit or essential characteristics thereof; the present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein, and any reference signs in the claims are not intended to be construed as limiting the claim concerned.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (3)
1. A method for deblurring invoice texts is characterized by comprising the following steps: training a deblurring core of a fuzzy condition, wherein the method for deblurring the invoice text comprises the following specific steps:
s1: making clear and fuzzy one-to-one corresponding invoice text information: firstly, processing clear invoice text information by a fuzzy effect through a traditional computer vision technology;
cutting out a complete red seal from clear invoice text information, and calculating a binary image of the red seal by a traditional method;
then, a screenshot A of a stamp and a screenshot B of a corresponding area on the red stamp are intercepted from clear invoice text information, and then the background of the screenshot B on the intercepted red stamp is replaced by the screenshot A on the invoice text information to obtain a replaced graph C;
finally, fusing the replaced graph C with the intercepted screenshot A to obtain a graph D, and splicing the graph D back;
the method comprises the steps that an unsupervised neural network algorithm is adopted, real data are divided into three types of clear, real and fuzzy and real red seal connection, and then clear text data are learned and converted into fuzzy data and the style of red seal shielding is fuzzy through the algorithm;
then obtaining a large amount of clear invoice text information and fuzzy invoice text information corresponding to each other, wherein the data are applied to the following calculation fuzzy core;
s2: calculating a fuzzy core of fuzzy invoice text information: acquiring the clear invoice text information in the step S1 and fuzzy invoice text information corresponding to each other, and continuously learning, continuously calculating and adjusting the fuzzy core from the corresponding clear information through the fuzzy information by adopting a supervised deep learning neural network;
finally, a fitted fuzzy kernel is obtained through continuous training;
s3: and (3) converting fuzzy verification into clear invoice text information: inputting an unknown invoice text message, and calling the fuzzy core calculated in the last step to obtain clear invoice text message.
2. The method of claim 1, wherein the method for deblurring invoice text comprises the following steps: the blurring effect in step S1 may be added in a manner of noise, defocus blur, mirror blur, motion blur, snow, fog, down-sampling, median blur, or the like.
3. The method of claim 1, wherein the method for deblurring invoice text comprises the following steps: the unknown invoice text information in the step S3 is clear unknown invoice text information or fuzzy unknown invoice text information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910924838.2A CN110782402A (en) | 2019-09-27 | 2019-09-27 | Method for deblurring invoice text |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910924838.2A CN110782402A (en) | 2019-09-27 | 2019-09-27 | Method for deblurring invoice text |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110782402A true CN110782402A (en) | 2020-02-11 |
Family
ID=69384636
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910924838.2A Pending CN110782402A (en) | 2019-09-27 | 2019-09-27 | Method for deblurring invoice text |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110782402A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114049641A (en) * | 2022-01-13 | 2022-02-15 | 中国电子科技集团公司第十五研究所 | Character recognition method and system based on deep learning |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107730458A (en) * | 2017-09-05 | 2018-02-23 | 北京飞搜科技有限公司 | A kind of fuzzy facial reconstruction method and system based on production confrontation network |
| CN108549892A (en) * | 2018-06-12 | 2018-09-18 | 东南大学 | A kind of license plate image clarification method based on convolutional neural networks |
| CN109727201A (en) * | 2017-10-30 | 2019-05-07 | 富士通株式会社 | Information processing equipment, image processing method and storage medium |
| CN109934778A (en) * | 2019-01-30 | 2019-06-25 | 长视科技股份有限公司 | A kind of blind deblurring method of household monitor video screenshot |
| CN110276253A (en) * | 2019-05-15 | 2019-09-24 | 中国科学院信息工程研究所 | A kind of fuzzy literal detection recognition method based on deep learning |
-
2019
- 2019-09-27 CN CN201910924838.2A patent/CN110782402A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107730458A (en) * | 2017-09-05 | 2018-02-23 | 北京飞搜科技有限公司 | A kind of fuzzy facial reconstruction method and system based on production confrontation network |
| CN109727201A (en) * | 2017-10-30 | 2019-05-07 | 富士通株式会社 | Information processing equipment, image processing method and storage medium |
| CN108549892A (en) * | 2018-06-12 | 2018-09-18 | 东南大学 | A kind of license plate image clarification method based on convolutional neural networks |
| CN109934778A (en) * | 2019-01-30 | 2019-06-25 | 长视科技股份有限公司 | A kind of blind deblurring method of household monitor video screenshot |
| CN110276253A (en) * | 2019-05-15 | 2019-09-24 | 中国科学院信息工程研究所 | A kind of fuzzy literal detection recognition method based on deep learning |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114049641A (en) * | 2022-01-13 | 2022-02-15 | 中国电子科技集团公司第十五研究所 | Character recognition method and system based on deep learning |
| CN114049641B (en) * | 2022-01-13 | 2022-03-15 | 中国电子科技集团公司第十五研究所 | Character recognition method and system based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| TWI728465B (en) | Method, device and electronic apparatus for image processing and storage medium thereof | |
| CN107679525B (en) | Image classification method and device and computer readable storage medium | |
| CN110399929B (en) | Fundus image classification method, fundus image classification apparatus, and computer-readable storage medium | |
| CN111488865B (en) | Image optimization method and device, computer storage medium and electronic equipment | |
| Wu et al. | Accurate transmission estimation for removing haze and noise from a single image | |
| CN111539290B (en) | Video motion recognition method and device, electronic equipment and storage medium | |
| Panetta et al. | Tmo-net: A parameter-free tone mapping operator using generative adversarial network, and performance benchmarking on large scale hdr dataset | |
| CN109447001B (en) | Dynamic emotion recognition method | |
| CN110781980B (en) | Training method of target detection model, target detection method and device | |
| Yuan et al. | Single image dehazing via NIN-DehazeNet | |
| CN111680702A (en) | Method for realizing weak supervision image significance detection by using detection frame | |
| CN111275686A (en) | Method and device for generating medical image data for artificial neural network training | |
| CN113505854A (en) | Method, device, equipment and medium for constructing facial image quality evaluation model | |
| CN114140844A (en) | Face silence living body detection method and device, electronic equipment and storage medium | |
| CN112700460A (en) | Image segmentation method and system | |
| Yang et al. | Low‐light image enhancement based on Retinex decomposition and adaptive gamma correction | |
| Jin et al. | Vehicle license plate recognition for fog‐haze environments | |
| Zhao et al. | A deep variational Bayesian framework for blind image deblurring | |
| Yu et al. | FS-GAN: Fuzzy Self-guided structure retention generative adversarial network for medical image enhancement | |
| CN110782402A (en) | Method for deblurring invoice text | |
| Fridman et al. | Sideeye: A generative neural network based simulator of human peripheral vision | |
| Liu et al. | Dual UNet low-light image enhancement network based on attention mechanism | |
| CN115358952A (en) | Image enhancement method, system, equipment and storage medium based on meta-learning | |
| Zhou et al. | An improved algorithm using weighted guided coefficient and union self‐adaptive image enhancement for single image haze removal | |
| CN114283087A (en) | Image denoising method and related equipment |
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 |