CN111414917B - Identification method of low-pixel-density text - Google Patents
Identification method of low-pixel-density text Download PDFInfo
- Publication number
- CN111414917B CN111414917B CN202010190222.XA CN202010190222A CN111414917B CN 111414917 B CN111414917 B CN 111414917B CN 202010190222 A CN202010190222 A CN 202010190222A CN 111414917 B CN111414917 B CN 111414917B
- Authority
- CN
- China
- Prior art keywords
- text
- special symbol
- image information
- recognition
- result
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- 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
-
- 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
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/22—Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The invention belongs to the technical field of image recognition, and relates to a recognition method of a text with low pixel density. The method comprises the following specific steps: image information acquisition is carried out on the text with low pixel density to be identified, the acquired image information is input and preprocessed, and preliminary positioning is carried out; accurate positioning is carried out again; carrying out image information area identification according to the obtained accurate positioning result, classifying the identified result, and respectively inputting the classified data into a special symbol identification model and a common text identification model for identification; and summarizing and structuring the recognized result. The beneficial effects of the invention are as follows: the method is characterized in that the character area containing the special symbol in the low-pixel-density text is judged, and the special symbol is identified by using a trained model, so that the special symbol identification accuracy is improved, the identification process is carried out in multiple steps, but the detection and re-identification networks are relatively small, and the identification efficiency can be effectively improved.
Description
Technical Field
The invention belongs to the technical field of image recognition, and particularly relates to a recognition method of a low-pixel-density text.
Background
Currently, a large number of paper documents, such as credit cards, draft, invoices, contracts, shipping documents, etc., are generated in transactions, trade and business-to-business transactions. These paper documents often need to be entered into the corresponding business system, which requires a lot of manual and repetitive labor, and is inefficient and prone to error.
OCR (Optical Character Recognition) is a technique for converting text in an image into text by image recognition. Therefore, the OCR technology can effectively solve the problem caused by manual entry under the condition of processing various types of bill scenes.
In common document tickets there are a large number of special symbols such as radio boxes, check boxes, underlining and clause constraint footnotes, etc., whereas OCR recognition systems currently existing on the market are not able to process and recognize efficiently. Because the pixel density ratio of the areas is low, the recall rate of the common deep learning detection method is low, and the omission phenomenon is easy to occur; because the frequency of occurrence of special symbols is low and similar to the structure of Chinese characters, the transfer learning is difficult to achieve a better recognition effect by expanding a training data set.
Disclosure of Invention
The invention discloses a method for identifying low-pixel-density text, which aims to solve any one of the above and other potential problems in the prior art.
In order to achieve the above purpose, the technical scheme of the invention is as follows: a recognition method of low-pixel-density text specifically comprises the following steps:
s1) acquiring image information of a text with low pixel density to be identified, and inputting the acquired image information;
s2) preprocessing input image information, and processing an image information area by using SIFT (Scale-Invariant feature Transform, scale invariant feature transform) for preliminary positioning;
s3) carrying out accurate positioning again according to the result of the preliminary positioning obtained in the S2);
s4) carrying out image information area identification on the accurate positioning result obtained in the step S3), and classifying the identified result to obtain a common text area and a special symbol area;
s5) inputting the common text region and the special symbol region obtained in the S4) into a special symbol recognition model and a common text recognition model respectively for recognition;
s6) summarizing and structuring the recognized result.
The low pixel density text contains a large number of special symbols such as radio boxes, check boxes, underlining, and footnotes, etc., resulting in text recognition with very low local pixel density.
Further, the step S5) includes a correction step: and carrying out post correction on the recognized result through an error correction mechanism.
Further, the specific steps of S2 are as follows:
s2.1) adopting a characteristic point matching algorithm to convert the input image information into a matrix,
s2.2) calculating extreme points and corresponding coordinates by constructing a differential pyramid of a scale space and a Gaussian space,
s2.3) obtaining rough coordinates of all extreme point discrimination and positioning special symbols in the image information according to the S2.2), extracting corresponding image areas, and completing preliminary positioning.
Further, the specific steps of S3) are as follows:
s3.1) inputting the result subjected to the preliminary positioning of S2.3) into a CRAST deep learning detection model,
s3.2) calculating to obtain region coordinates containing a common text region and a special symbol region in the image information by using a CRAST deep learning detection model, extracting the image information in the region coordinates and extracting the directional gradient histogram characteristics of the corresponding image,
s3.3) carrying out two classification on the gradient histogram by using an SVM vector machine according to the gradient histogram characteristics obtained in the step S3.2), and dividing the gradient histogram into a common text region and a special symbol region.
Further, the specific steps of S4) are as follows:
s4.1), firstly inputting the common text region obtained in the step S3.3) into a trained common text recognition model for recognition, and outputting a recognition result;
s4.2) inputting the special symbol area obtained in the step S3.3) into a trained special symbol recognition model for recognition, and outputting a recognition result.
Further, the specific steps of training the text recognition model and the special symbol recognition model are as follows:
step 1: taking text content in the millions of pictures as a training set of the network model;
step 2: uniformly processing all pictures to be trained in the training set, and converting the pictures to be trained into a matrix with the value range of [ -1,1 ];
step 3: randomly dividing all pictures to be trained in the training set into a plurality of groups, and combining the grouped pictures with a matrix obtained in S4.12) according to the groups to obtain batch batches;
step 4: inputting the batch obtained in the step 3 into an input layer of a corresponding identification model, and comparing an output result with an error of an actual value of a sample label in an output layer to calculate a final value;
step 5: if the output layer is subjected to CTC algorithm (Connectionist Temporal Classification, mainly used for processing alignment of input and output labels in sequence labeling problem), executing step 7 if the difference between the output result and the actual label result is smaller, otherwise executing step 6;
step 6: updating the network weight and threshold of each neuron in the hidden layer of the corresponding convolutional neural recognition model, enabling the network error function to descend along the negative gradient direction, enabling the output result to approach the expected output, and returning to the step 4;
step 7: and obtaining a trained ordinary text recognition model and a trained special symbol recognition model.
Further, the unified processing manner in the step 2 includes graying or binarization.
A computer program implementing the method of recognition of low pixel density text described above.
An information processing terminal for realizing the identification method of the low-pixel density text.
A computer readable storage medium comprising instructions which, when run on a computer, cause the computer to perform the method of the low pixel density text recognition method described above.
The invention has the beneficial effects that due to the adoption of the technical scheme, the invention is a multi-stage identification scheme and mainly comprises the steps of screening, identifying and positioning the characteristic symbol region, re-identifying and correcting the characteristic symbol region and the like, which is a step-by-step fine identification process. The method is characterized in that the character area containing the special symbol is judged, the special symbol is identified by using a trained model, so that the special symbol identification accuracy is improved, the identification process is carried out in multiple steps, but the detection and re-identification networks are relatively small, and the identification efficiency can be effectively improved.
Drawings
FIG. 1 is a block flow diagram of a method for identifying text with low pixel density according to the present invention.
Fig. 2 is a schematic diagram of a picture to be identified.
FIG. 3 is a schematic diagram of a detected region identified by the method of the present invention.
In the figure:
1. general text area, 2. Special symbol area.
Detailed Description
The technical scheme of the invention is further described below with reference to the accompanying drawings and specific embodiments.
As shown in fig. 1, a method for identifying low-pixel density text according to the present invention specifically includes the following steps:
s1) acquiring image information of a text with low pixel density to be identified, and inputting the acquired image information;
s2) preprocessing input image information, and processing an image information area by using a SIFT method to perform preliminary positioning;
s3) carrying out accurate positioning again according to the result of the preliminary positioning obtained in the S2);
s4) carrying out image information area identification on the accurate positioning result obtained in the step S3), and classifying the identified result to obtain a common text area and a special symbol area;
s5) inputting the common text region and the special symbol region obtained in the S4) into a special symbol recognition model and a common text recognition model respectively for recognition;
s6) summarizing and structuring the recognized result.
The step S5) also comprises a correction step: and carrying out post correction on the recognized result through an error correction mechanism.
The specific steps of the S2 are as follows:
s2.1) adopting a characteristic point matching algorithm to convert the input image information into a matrix,
s2.2) calculating extreme points and corresponding coordinates by constructing a differential pyramid of a scale space and a Gaussian space,
s2.3) obtaining rough coordinates of all extreme point discrimination and positioning special symbols in the image information according to the S2.2), extracting corresponding image areas, and completing preliminary positioning.
The specific steps of the S3) are as follows:
s3.1) inputting the result subjected to the preliminary positioning of S2.3) into a CRAST deep learning detection model,
s3.2) calculating to obtain region coordinates containing a common text region and a special symbol region in the image information by using a CRAST deep learning detection model, extracting the image information in the region coordinates and extracting the directional gradient histogram characteristics of the corresponding image,
s3.3) carrying out two classification on the gradient histogram by using an SVM vector machine according to the gradient histogram characteristics obtained in the step S3.2), and dividing the gradient histogram into a common text region and a special symbol region.
The specific steps of the S4) are as follows:
s4.1), firstly inputting the common text region obtained in the step S3.3) into a trained common text recognition model for recognition, and outputting a recognition result;
s4.2) inputting the special symbol area obtained in the step S3.3) into a trained special symbol recognition model for recognition, and outputting a recognition result.
The text recognition model and the special symbol recognition model training specifically comprises the following steps:
step 1: taking text content in the millions of pictures as a training set of the network model;
step 2: uniformly processing all pictures to be trained in the training set, and converting the pictures to be trained into a matrix with the value range of [ -1,1 ];
step 3: randomly dividing all pictures to be trained in the training set into a plurality of groups, and combining the grouped pictures with a matrix obtained in S4.12) according to the groups to obtain batch batches;
step 4: inputting the batch obtained in the step 3 into an input layer of a corresponding identification model, and comparing an output result with an error of an actual value of a sample label in an output layer to calculate a final value;
step 5: if the difference between the output result of the output layer after being processed by the CTC algorithm and the actual label result is smaller, executing the step 7, otherwise executing the step 6;
step 6: updating the network weight and threshold of each neuron in the hidden layer of the corresponding convolutional neural recognition model, enabling the network error function to descend along the negative gradient direction, enabling the output result to approach the expected output, and returning to the step 4;
step 7: and obtaining a trained ordinary text recognition model and a trained special symbol recognition model.
The unified processing mode in the step 2 comprises graying or binarization.
A computer program implementing the method of recognition of low pixel density text described above.
An information processing terminal for realizing the identification method of the low-pixel density text.
A computer readable storage medium comprising instructions that when executed on a computer cause the computer to perform the method of identifying low pixel density text described above.
Examples:
a recognition method of low-pixel-density text specifically comprises the following steps:
firstly, collecting images of a text (document bill) with low pixel density to be identified, and inputting the collected images;
secondly, adopting a feature matching algorithm, converting an input image into a matrix by utilizing two-dimensional Gaussian blur (formula 1, sigma is a standard deviation, r is a radius), constructing a scale space (formula 2, I is an original image, and is convolution operation) and a Gaussian space differential pyramid (formula 3, k is the reciprocal of the number of pyramid layers), calculating extreme points and corresponding coordinates according to the differential pyramid, and finally judging and positioning approximate coordinates of special symbols on the image according to all the extreme points and extracting a corresponding image area, wherein fig. 1 is the extracted image area;
L(x,y,σ)=G(x,y,σ)*I(x,y) (2)
inputting a picture to be detected by using a CRATT deep learning model (as shown in figure 2), obtaining picture coordinates containing a text region by model calculation, extracting an image in the coordinates, extracting a direction gradient histogram (Histogram of Oriented Gradient) characteristic, classifying and judging whether the image is a common text region or a special symbol region by using an SVM vector machine by using the vector, wherein the detected region is a text region in figure 3, and the detected region is a special symbol region in figure 1;
inputting the images extracted and classified in the step 2 into a trained special symbol recognition model (the model design core is shown as a formula 4), recognizing the images, and outputting a corresponding recognition result;
H l =Relu(b l *f l (H l-1 )+id(H l-1 )) (4)
correcting the error of the identification result according to the characteristics of the corresponding image to be identified;
inputting an image without special symbols into a trained ordinary text recognition model (a model design core is shown as a formula 5) for text recognition and post correction;
x i =H i ([x 0 ,x 1 ,...,x i-1 ]) (5)
and finally, merging and sequencing the identification results and outputting the identification results.
The above describes in detail the method for identifying the text with low pixel density provided in the embodiment of the present application. The above description of embodiments is only for aiding in understanding the method of the present application and its core ideas; meanwhile, as those skilled in the art will have modifications in the specific embodiments and application scope in accordance with the ideas of the present application, the present description should not be construed as limiting the present application in view of the above.
Certain terms are used throughout the description and claims to refer to particular components. Those of skill in the art will appreciate that a hardware manufacturer may refer to the same component by different names. The description and claims do not take the form of an element differentiated by name, but rather by functionality. As referred to throughout the specification and claims, the terms "comprising," including, "and" includes "are intended to be interpreted as" including/comprising, but not limited to. By "substantially" is meant that within an acceptable error range, a person skilled in the art is able to solve the technical problem within a certain error range, substantially achieving the technical effect. The description hereinafter sets forth the preferred embodiment for carrying out the present application, but is not intended to limit the scope of the present application in general, for the purpose of illustrating the general principles of the present application. The scope of the present application is defined by the appended claims.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a product or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such product or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a commodity or system comprising such elements.
It should be understood that the term "and/or" as used herein is merely one relationship describing the association of the associated objects, meaning that there may be three relationships, e.g., a and/or B, may represent: a exists alone, A and B exist together, and B exists alone. In addition, the character "/" herein generally indicates that the front and rear associated objects are an "or" relationship.
While the foregoing description illustrates and describes the preferred embodiments of the present application, it is to be understood that this application is not limited to the forms disclosed herein, but is not to be construed as an exclusive use of other embodiments, and is capable of many other combinations, modifications and environments, and adaptations within the scope of the teachings described herein, through the foregoing teachings or through the knowledge or skills of the relevant art. And that modifications and variations which do not depart from the spirit and scope of the present invention are intended to be within the scope of the appended claims.
Claims (4)
1. A method for identifying low-pixel density text specifically comprises the following steps:
s1) acquiring image information of a text with low pixel density to be identified, and inputting the acquired image information;
s2) preprocessing input image information, and processing an image information area by using a SIFT method to perform preliminary positioning;
the method comprises the following specific steps:
s2.1) adopting a characteristic point matching algorithm to convert the input image information into a matrix,
s2.2) calculating extreme points and corresponding coordinates by constructing a differential pyramid of a scale space and a Gaussian space,
s2.3) obtaining rough coordinates of all extreme point discrimination and positioning special symbols in the image information according to the S2.2), extracting corresponding image areas, and completing preliminary positioning;
s3) carrying out accurate positioning again according to the result of the preliminary positioning obtained in the S2);
the method comprises the following steps: s3.1) inputting the result subjected to the preliminary positioning of S2.3) into a CRAST deep learning detection model,
s3.2) calculating to obtain region coordinates containing a common text region and a special symbol region in the image information by using a CRAST deep learning detection model, extracting the image information in the region coordinates and extracting the directional gradient histogram characteristics of the corresponding image,
s3.3) carrying out two classification on the gradient histogram by using an SVM vector machine according to the gradient histogram characteristics obtained in the step S3.2), and dividing the gradient histogram into a common text region and a special symbol region;
s4) carrying out image information area identification on the accurate positioning result obtained in the step S3), and classifying the identified result to obtain a common text area and a special symbol area;
s5) inputting the common text region and the special symbol region obtained in the S4) into a special symbol recognition model and a common text recognition model respectively for recognition;
s6) summarizing and structuring the identified results and outputting the results; the method is characterized by comprising the following specific steps of S4):
s4.1), firstly inputting the common text region obtained in the step S3.3) into a trained common text recognition model for recognition, and outputting a recognition result;
s4.2) inputting the special symbol area obtained in the step S3.3) into a trained special symbol recognition model for recognition, and outputting a recognition result;
the text recognition model and the special symbol recognition model training specifically comprises the following steps:
step 1: taking text content in the millions of pictures as a training set of the network model;
step 2: uniformly processing all pictures to be trained in the training set, and converting the pictures to be trained into a matrix with the value range of [ -1,1 ];
step 3: randomly dividing all the pictures to be trained in the training set into a plurality of groups, and combining the grouped pictures with the matrix obtained in the step 2 according to the groups to obtain batch batches;
step 4: inputting the batch obtained in the step 3 into an input layer of a corresponding identification model, and comparing an output result with an error of an actual value of a sample label in an output layer to calculate a final value;
step 5: if the difference between the output result of the output layer after being processed by the CTC algorithm and the actual label result is smaller, executing the step 7, otherwise executing the step 6;
step 6: updating the network weight and threshold of each neuron in the hidden layer of the corresponding convolutional neural recognition model, enabling the network error function to descend along the negative gradient direction, enabling the output result to approach the expected output, and returning to the step 4;
step 7: and obtaining a trained ordinary text recognition model and a trained special symbol recognition model.
2. The method according to claim 1, wherein the unified processing means in step 2 includes graying or binarizing.
3. An information processing terminal implementing the low-pixel-density text recognition method according to any one of claims 1 to 2.
4. A computer readable storage medium comprising instructions which, when run on a computer, cause the computer to perform the method of identifying low pixel density text as claimed in any one of claims 1-2.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010190222.XA CN111414917B (en) | 2020-03-18 | 2020-03-18 | Identification method of low-pixel-density text |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010190222.XA CN111414917B (en) | 2020-03-18 | 2020-03-18 | Identification method of low-pixel-density text |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111414917A CN111414917A (en) | 2020-07-14 |
CN111414917B true CN111414917B (en) | 2023-05-12 |
Family
ID=71491107
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010190222.XA Active CN111414917B (en) | 2020-03-18 | 2020-03-18 | Identification method of low-pixel-density text |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111414917B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112514970B (en) * | 2020-12-08 | 2021-11-05 | 烟台海裕食品有限公司 | Self-adaptive fish scale removing platform and method |
CN113642556A (en) * | 2021-08-04 | 2021-11-12 | 五八有限公司 | Image processing method and device, electronic equipment and storage medium |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2019232874A1 (en) * | 2018-06-04 | 2019-12-12 | 平安科技(深圳)有限公司 | Chinese character model training method, chinese character recognition method, apparatus, device, and medium |
WO2019232872A1 (en) * | 2018-06-04 | 2019-12-12 | 平安科技(深圳)有限公司 | Handwritten character model training method, chinese character recognition method, apparatus, device, and medium |
Family Cites Families (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9984471B2 (en) * | 2016-07-26 | 2018-05-29 | Intuit Inc. | Label and field identification without optical character recognition (OCR) |
US10013643B2 (en) * | 2016-07-26 | 2018-07-03 | Intuit Inc. | Performing optical character recognition using spatial information of regions within a structured document |
US10776903B2 (en) * | 2017-07-17 | 2020-09-15 | Open Text Corporation | Systems and methods for image modification and image based content capture and extraction in neural networks |
CN108038476B (en) * | 2018-01-03 | 2019-10-11 | 东北大学 | A kind of facial expression recognition feature extracting method based on edge detection and SIFT |
CN108446621A (en) * | 2018-03-14 | 2018-08-24 | 平安科技(深圳)有限公司 | Bank slip recognition method, server and computer readable storage medium |
CN108764226B (en) * | 2018-04-13 | 2022-05-03 | 顺丰科技有限公司 | Image text recognition method, device, equipment and storage medium thereof |
CN109376658B (en) * | 2018-10-26 | 2022-03-08 | 信雅达科技股份有限公司 | OCR method based on deep learning |
CN109919014B (en) * | 2019-01-28 | 2023-11-03 | 平安科技(深圳)有限公司 | OCR (optical character recognition) method and electronic equipment thereof |
CN110263694A (en) * | 2019-06-13 | 2019-09-20 | 泰康保险集团股份有限公司 | A kind of bank slip recognition method and device |
-
2020
- 2020-03-18 CN CN202010190222.XA patent/CN111414917B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2019232874A1 (en) * | 2018-06-04 | 2019-12-12 | 平安科技(深圳)有限公司 | Chinese character model training method, chinese character recognition method, apparatus, device, and medium |
WO2019232872A1 (en) * | 2018-06-04 | 2019-12-12 | 平安科技(深圳)有限公司 | Handwritten character model training method, chinese character recognition method, apparatus, device, and medium |
Also Published As
Publication number | Publication date |
---|---|
CN111414917A (en) | 2020-07-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Gonçalves et al. | Benchmark for license plate character segmentation | |
Cheriet et al. | Character recognition systems: a guide for students and practitioners | |
Shanthi et al. | A novel SVM-based handwritten Tamil character recognition system | |
US8750619B2 (en) | Character recognition | |
Bataineh et al. | A novel statistical feature extraction method for textual images: Optical font recognition | |
Chim et al. | Character recognition using statistical moments | |
US9224207B2 (en) | Segmentation co-clustering | |
CN112508011A (en) | OCR (optical character recognition) method and device based on neural network | |
CN113963147B (en) | Key information extraction method and system based on semantic segmentation | |
Hossain et al. | Recognition and solution for handwritten equation using convolutional neural network | |
CN111414917B (en) | Identification method of low-pixel-density text | |
Caldeira et al. | Industrial optical character recognition system in printing quality control of hot-rolled coils identification | |
Barnouti et al. | An efficient character recognition technique using K-nearest neighbor classifier | |
Neri et al. | A Convolutional Neural Network for Handwritten Digit Recognition. | |
Adyanthaya | Text recognition from images: a study | |
Kumar et al. | Online handwritten character recognition for Telugu language using support vector machines | |
CN115376149A (en) | Reimbursement invoice identification method | |
CN114581928A (en) | Form identification method and system | |
Shirdhonkar et al. | Discrimination between printed and handwritten text in documents | |
CN113111882A (en) | Card identification method and device, electronic equipment and storage medium | |
Gaceb et al. | Classification of business documents for real-time application | |
Paul et al. | Text localization in camera captured images using adaptive stroke filter | |
Halder et al. | Comparison of the classifiers in Bangla handwritten numeral recognition | |
Kadam et al. | Handwritten Marathi Compound Character Recognition | |
Rajdev et al. | Optical Character and Font Recognizer |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |