CN112257613A - Physical examination report information structured extraction method and device and computer equipment - Google Patents

Physical examination report information structured extraction method and device and computer equipment Download PDF

Info

Publication number
CN112257613A
CN112257613A CN202011149638.3A CN202011149638A CN112257613A CN 112257613 A CN112257613 A CN 112257613A CN 202011149638 A CN202011149638 A CN 202011149638A CN 112257613 A CN112257613 A CN 112257613A
Authority
CN
China
Prior art keywords
physical examination
examination report
report
target
picture set
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.)
Granted
Application number
CN202011149638.3A
Other languages
Chinese (zh)
Other versions
CN112257613B (en
Inventor
欧光礼
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ping An Life Insurance Company of China Ltd
Original Assignee
Ping An Life Insurance Company of China Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Ping An Life Insurance Company of China Ltd filed Critical Ping An Life Insurance Company of China Ltd
Priority to CN202011149638.3A priority Critical patent/CN112257613B/en
Publication of CN112257613A publication Critical patent/CN112257613A/en
Application granted granted Critical
Publication of CN112257613B publication Critical patent/CN112257613B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G06V30/40—Document-oriented image-based pattern recognition
    • 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
    • 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
    • G06V10/20—Image preprocessing
    • G06V10/24—Aligning, centring, orientation detection or correction of the image
    • G06V10/245—Aligning, centring, orientation detection or correction of the image by locating a pattern; Special marks for positioning
    • 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
    • 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
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Artificial Intelligence (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Character Input (AREA)
  • Character Discrimination (AREA)

Abstract

The invention discloses a physical examination report information structured extraction method, a physical examination report information structured extraction device, computer equipment and a storage medium, relates to an image recognition technology, can be applied to an intelligent medical scene, and comprises the steps of obtaining report source information corresponding to a physical examination report image picture set, and positioning a to-be-recognized area in the physical examination report image picture set to form a to-be-recognized area picture set; sequentially removing perspective deformation and character recognition to obtain a corresponding recognition text; obtaining a target field in the recognition text and a value of the target field through semantic analysis positioning to form a target text set; calling a pre-stored standard field set to correct each target field in the target text set to obtain a corrected target text set; and storing the corrected target text set into a correspondingly created storage area to obtain the physical examination report structured information. The method realizes the rapid positioning of the area to be identified, the accurate text identification of the area to be identified and the rapid structural extraction of the core physical examination data of the physical examination report.

Description

Physical examination report information structured extraction method and device and computer equipment
Technical Field
The invention relates to the technical field of artificial intelligence image recognition, in particular to a physical examination report information structured extraction method and device, computer equipment and a storage medium.
Background
In the field of insurance underwriting, physical examination report information entry is a very important link, and at present, manual entry is mainly adopted for processing, namely, physical examination report information corresponding to an applicant is entered by underwriting personnel after physical examination report original paper submitted by a user is checked, so that the efficiency of the whole entry process is low, the labor cost is high, and the underwriting efficiency is low.
Disclosure of Invention
The embodiment of the invention provides a structured extraction method, a structured extraction device, computer equipment and a storage medium for physical examination report information, and aims to solve the problems of low efficiency and high labor cost of the whole entry process caused by the adoption of a mode of manually entering key information of a physical examination report in a safety and insurance system in the prior art.
In a first aspect, an embodiment of the present invention provides a method for structured extraction of physical examination report information, including:
receiving a physical examination report image picture set uploaded by a user side, and acquiring report source information corresponding to the physical examination report image picture set through OCR recognition; wherein the report source information comprises a physical examination report issuing organization name and a physical examination report type;
calling a pre-stored physical examination report sample set to acquire sample report source information and a sample physical examination data distribution area corresponding to each physical examination report sample;
if sample report source information corresponding to a physical examination report sample in the physical examination report sample set is the same as the report source information of the physical examination report image picture set, acquiring a sample physical examination data distribution area of the physical examination report sample so as to position an area to be identified in the physical examination report image picture set to form an area to be identified picture set;
sequentially removing perspective deformation and character recognition on the picture set of the area to be recognized to obtain a recognition text corresponding to the picture set of the area to be recognized;
obtaining a target field and a target field value in the recognition text through semantic analysis positioning to form a target text set;
calling a pre-stored standard field set, acquiring an approximate field corresponding to each target field in the target text set in the standard field set, and correcting each target field in the target text set to obtain a corrected target text set; and
and storing the corrected target text set to a correspondingly created storage area to obtain the structural information of the physical examination report.
In a second aspect, an embodiment of the present invention provides a physical examination report information structured extraction apparatus, which includes:
the physical examination report picture receiving unit is used for receiving a physical examination report image picture set uploaded by a user terminal and acquiring report source information corresponding to the physical examination report image picture set through OCR recognition; wherein the report source information comprises a physical examination report issuing organization name and a physical examination report type;
the system comprises a sample set acquisition unit, a sample analysis unit and a sample analysis unit, wherein the sample set acquisition unit is used for calling a pre-stored physical examination report sample set and acquiring sample report source information and a sample physical examination data distribution area corresponding to each physical examination report sample;
the picture set positioning unit of the area to be identified is used for acquiring a sample physical examination data distribution area of the physical examination report sample if sample report source information corresponding to the physical examination report sample in the physical examination report sample set is the same as the report source information of the physical examination report picture set so as to position the area to be identified in the physical examination report picture set to form a picture set of the area to be identified;
the text recognition unit is used for sequentially removing perspective deformation and character recognition on the picture set of the area to be recognized to obtain a recognition text corresponding to the picture set of the area to be recognized;
the target text set acquisition unit is used for acquiring a target field and a target field value in the recognition text through semantic analysis positioning so as to form a target text set;
the target text correction unit is used for calling a pre-stored standard field set, acquiring approximate fields corresponding to all target fields in the target text set in the standard field set, and correcting all target fields in the target text set to obtain a corrected target text set; and
and the structural information acquisition unit is used for storing the corrected target text set into a correspondingly created storage area so as to obtain the structural information of the physical examination report.
In a third aspect, an embodiment of the present invention further provides a computer device, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, the processor implements the method for structured extraction of physical examination report information according to the first aspect.
In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and the computer program, when executed by a processor, causes the processor to execute the method for structured extraction of physical examination report information according to the first aspect.
The embodiment of the invention provides a method and a device for extracting physical examination report information in a structured manner, computer equipment and a storage medium, wherein the method comprises the steps of acquiring report source information corresponding to a physical examination report image picture set through OCR recognition; calling a pre-stored physical examination report sample set to acquire sample report source information and a sample physical examination data distribution area corresponding to each physical examination report sample; if sample report source information corresponding to a physical examination report sample in the physical examination report sample set is the same as the report source information of the physical examination report image picture set, acquiring a sample physical examination data distribution area of the physical examination report sample so as to position an area to be identified in the physical examination report image picture set to form an area to be identified picture set; sequentially removing perspective deformation and character recognition of the picture set of the area to be recognized to obtain a recognition text corresponding to the picture set of the area to be recognized; obtaining a target field and a target field value in the recognition text through semantic analysis positioning to form a target text set; calling a pre-stored standard field set, acquiring approximate fields corresponding to all target fields in the target text set in the standard field set, and correcting all target fields in the target text set to obtain a corrected target text set; and storing the corrected target text set into a correspondingly created storage area to obtain the physical examination report structured information. The method realizes the rapid positioning of the area to be identified, the accurate text identification of the area to be identified and the rapid structural extraction of the core physical examination data of the physical examination report.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a schematic view of an application scenario of a physical examination report information structured extraction method according to an embodiment of the present invention;
fig. 2 is a schematic flowchart of a structured extraction method for physical examination report information according to an embodiment of the present invention;
fig. 3 is a schematic sub-flow diagram of a physical examination report information structured extraction method according to an embodiment of the present invention;
fig. 4 is a schematic block diagram of a physical examination report information structured extraction device provided by the embodiment of the invention;
FIG. 5 is a block diagram schematically illustrating sub-elements of a physical examination report information structured extraction apparatus according to an embodiment of the present invention;
FIG. 6 is a schematic block diagram of a computer device provided by an embodiment 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 some, not all, embodiments of the present invention. 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.
It will be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
It is also to be understood that the terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in the specification of the present invention and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.
It should be further understood that the term "and/or" as used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.
Referring to fig. 1 and fig. 2, fig. 1 is a schematic view of an application scenario of a physical examination report information structured extraction method according to an embodiment of the present invention; fig. 2 is a schematic flowchart of a structured extraction method for physical examination report information according to an embodiment of the present invention, where the structured extraction method for physical examination report information is applied to a server, and the method is executed by application software installed in the server.
As shown in fig. 2, the method includes steps S110 to S170.
S110, receiving a physical examination report image picture set uploaded by a user side, and acquiring report source information corresponding to the physical examination report image picture set through OCR recognition; wherein the report source information comprises a physical examination report issuing organization name and a physical examination report type.
In this embodiment, both the application and underwriting systems may be deployed in the server, and when a user uses a client to establish a communication connection with the server and logs in the application system, a user interaction interface of the application system is displayed on the client. At this time, after selecting a risk category (e.g., a serious disease risk) previously applied to the insurance, the user uploads a claim data for the risk category (e.g., uploads a physical examination report image picture set as the claim data, the physical examination report image picture set includes a plurality of physical examination report image pictures, generally, the user uses a photographing function of the user terminal to photograph a plurality of paper-based physical examination reports to form a physical examination report image picture set), and the claim data is uploaded to the server and then transferred to the underwriting system for claim data examination.
After the underwriting system deployed in the server receives the physical examination report image picture set uploaded by the user side, in order to determine which type of physical examination report template of which physical examination report issuing organization is, the text of the local area of the physical examination report image picture set can be acquired through OCR recognition, so as to acquire corresponding report source information.
In one embodiment, step S110 includes:
and acquiring the upper half area of the top page of the physical examination report image picture set, and acquiring the first three lines of characters corresponding to the upper half area of the top page through OCR recognition to acquire report source information corresponding to the physical examination report image picture set.
In this embodiment, different hospitals or physical examination institutions have different physical examination report templates, and the physical examination data arrangement areas of the physical examination report templates are different, more specifically, the first type of physical examination data (e.g., blood routine physical examination data) of hospital a is generally distributed in the lower half of the first page of the physical examination report, the second type of physical examination data (e.g., routine whole body physical examination data) of hospital a is generally distributed in the second to fifth pages of the physical examination report, and the physical examination data of hospital B is generally distributed in the lower half of the second page and the upper half of the third page of the physical examination report.
The reason why the text of the local area of the physical examination report image picture set (generally, the head-up part of the first page of the physical examination report image picture set, for example, the first 3 lines of text of the first page generally include the name of the hospital or the name of the physical examination institution and the type of the physical examination report, that is, the text is distributed in the head-up part of the first page of the physical examination report image picture set) is recognized is that the physical examination institution and the type of the physical examination report can be recognized in this area, so that the distribution area of the physical examination data can be rapidly determined.
S120, calling a pre-stored physical examination report sample set, and acquiring sample report source information and a sample physical examination data distribution area corresponding to each physical examination report sample.
In the embodiment, since the physical examination report samples respectively corresponding to the plurality of physical examination report issuing mechanisms are stored in the underwriting system of the server, the sample report source information and the sample physical examination data distribution area corresponding to each physical examination report sample are known. Thus, the report source information corresponding to the physical examination report image picture set is compared with the sample report source information of the physical examination report sample set one by one, and the distribution area of the examination data corresponding to the physical examination report image picture set can be rapidly judged and obtained.
S130, if the sample report source information corresponding to the physical examination report samples in the physical examination report sample set is the same as the report source information of the physical examination report image picture set, acquiring a sample physical examination data distribution area of the physical examination report samples so as to position the to-be-identified area in the physical examination report image picture set to form a to-be-identified area picture set.
In this embodiment, for example, when the report source information of the physical examination report picture set uploaded by the user terminal corresponds to a1 with the mechanism B1 physical examination report type, and the sample report source information corresponding to the physical examination report sample in the physical examination report sample set also corresponds to a1 with the mechanism B1 physical examination report type, sample physical examination data distribution regions of the physical examination report sample (specifically, physical examination data is distributed in the lower half of the second page and the upper half of the third page of the physical examination report) can be obtained, the same region in the physical examination report picture set is directly located according to the sample physical examination data distribution regions as the to-be-identified region, and the to-be-identified region picture sets are composed of to-be-identified region pictures corresponding to the to-be-identified regions. The areas to be identified are in sequence, that is, each area to be identified corresponds to a page attribute value, for example, if physical examination data is distributed on the lower half of the second page of the physical examination report, the page attribute value corresponding to the area is equal to 2, and if physical examination data is distributed on the upper half of the third page of the physical examination report, the page attribute value corresponding to the area is equal to 3.
For example, if the physical examination report image picture set corresponds to a1 physical examination report type with a mechanism B1, the lower half area of the second page in the physical examination report image picture set is located as an area to be identified 1 (the picture corresponding to the area to be identified 1 is referred to as an area picture to be identified 1), the upper half area of the third page in the physical examination report image picture set is located as an area to be identified 2 (the picture corresponding to the area to be identified 2 is referred to as an area picture to be identified 2), and the area picture set to be identified is composed of the area picture to be identified 1 and the area picture to be identified 2. By the mode of quickly positioning the image set of the physical examination report according to the physical examination report issuing mechanism and the physical examination report type, the accurate positioning area for extracting the physical examination report data is quickly screened out, so that the area for subsequently performing text recognition is reduced, and the efficiency of text recognition and data extraction is improved.
S140, removing perspective deformation and character recognition of the picture set of the area to be recognized in sequence to obtain a recognition text corresponding to the picture set of the area to be recognized.
In this embodiment, after the picture set of the region to be identified is obtained through the report source information of the image set of the physical examination report, in order to perform text identification more accurately, preprocessing for removing perspective distortion needs to be performed on the picture set before text identification, so that the text in the picture can be extracted more accurately and quickly.
In one embodiment, as shown in fig. 3, step S140 includes:
s141, removing perspective deformation of the picture set of the region to be identified through a Warping algorithm to obtain a first processing picture set;
s142, performing character cutting on each first processed picture in the first processed picture set to obtain a plurality of character cutting sub-pictures to form a character cutting sub-picture set;
s143, calling a pre-trained CRNN-CTC character recognition model, and performing character recognition on each character cutting sub-picture in the character cutting sub-picture set through the CRNN-CTC character recognition model to obtain character recognition results respectively corresponding to each character cutting sub-picture;
s144, sequentially connecting and combining the character recognition results corresponding to the character cutting sub-pictures in series to obtain a recognition text corresponding to the picture set of the area to be recognized.
In this embodiment, since the to-be-identified region picture set corresponding to the physical examination report image picture set is already acquired, and the to-be-identified region pictures in the to-be-identified region picture set are arranged in an ascending order according to the size of the page attribute value, at this time, the perspective deformation processing is sequentially removed from each to-be-identified region picture in the to-be-identified region picture set through a Warping algorithm (i.e., an image deformation algorithm), so as to obtain first processing pictures respectively corresponding to each to-be-identified region picture, so as to form a first processing picture set.
In one embodiment, step S141 includes:
and converting all intersected line segments of the pictures of the areas to be identified in the picture set of the areas to be identified into parallel line segments through a homography matrix in a Warping algorithm to remove perspective deformation, thereby obtaining a first processing picture set.
In the present embodiment, the removal of perspective distortion can be achieved by a homography matrix (homography). In computer vision, the homography of a plane is defined as the projection mapping from one plane to another. Through a function findHomography () provided by OpenCV (which is a cross-platform computer vision and machine learning software library based on BSD licensing), corresponding point sequences are used as input, homography matrixes which best describe the corresponding points are returned, and therefore the solution of the homography matrixes is achieved. All intersecting line segments of the pictures of the areas to be identified in the picture set of the areas to be identified can be converted into parallel line segments through the homography matrix in the Warping algorithm, and then the processing of removing geometric deformation can be added, so that a first processed picture set is obtained.
After the first processed picture set is obtained, in order to perform text recognition more accurately, character cutting needs to be performed on each first processed picture in the first processed picture set, that is, each first processed picture is refined and cut into a plurality of character cutting sub-pictures again to form a character cutting sub-picture set. Through the character cutting mode with finer granularity, the text recognition of small areas can be performed more accurately in the follow-up process. In specific implementation, a statistical segmentation method or a horizontal/vertical projection character cutting method can be selected to perform character cutting on each first processed picture in the first processed picture set so as to obtain a character cutting sub-picture set.
And calling a locally stored and trained CRNN-CTC character recognition model in the server after the character cutting sub-picture set is obtained so as to perform text recognition on each character cutting sub-picture in the character cutting sub-picture set. The CRNN network in the CRNN-CTC character recognition model is a neural network formed by mixing CNN (convolutional neural network) and RNN (recurrent neural network), the convolutional layer of the CRNN network is a CNN network (convolutional characteristic matrix for extracting input images), the recurrent network layer of the CRNN network is a deep bidirectional LSTM network (used for continuously extracting character sequence characteristics on the basis of the convolutional characteristic matrix), and extracted texts corresponding to the input images are output after the output of the recurrent network layer of the CRNN network is softmax. CTC (the english name of CTC is connection Temporal Classification, which represents connection timing Classification), that is, CTC is used to solve the problem that input sequences and output sequences are difficult to correspond one to one, and can improve the robustness of single-line text recognition.
After the character recognition results are sequentially recognized according to the sequence of cutting the sub-pictures by the characters, the character recognition results and the character recognition results can be sequentially connected in series and combined to obtain a recognition text corresponding to the picture set of the region to be recognized, wherein the recognition text comprises a plurality of characters (possibly English characters, numeric characters, symbolic characters and/or Chinese characters). After the identification of the picture set of the area to be identified is completed, the key data required by the subsequent user can be extracted.
S150, obtaining a target field and a target field value in the recognition text through semantic analysis positioning to form a target text set.
In this embodiment, sentence-level semantic analysis is used to extract a target field and a value of the target field from the recognition text, and more specifically, a semantic role labeling method based on complete syntactic analysis may be used to perform semantic analysis. After the recognition text is given, syntactic analysis, candidate argument pruning, argument recognition, argument labeling and post-processing (processing the arguments after labeling or adding some richer information) are sequentially performed to obtain a target text set, so that the extraction of the target field and the value of the target field is realized.
In one embodiment, the step S150 includes:
dividing the recognition text into a plurality of sentences to be recognized according to separators;
and obtaining a target field and a target field value respectively included by each sentence to be recognized through semantic analysis to form a target text set.
In this embodiment, since sentence-level semantic analysis is used to extract the target field and the target field value of the recognition text, before the semantic analysis, the recognition text may be divided into a plurality of sentences to be recognized according to separators (for example, periods, twiddle symbols, or the last end of each character cut sub-picture is automatically regarded as a separator), and then sentence-level semantic analysis may be used to extract the target field and the target field value of the recognition text.
And S160, calling a pre-stored standard field set, acquiring an approximate field corresponding to each target field in the target text set in the standard field set, and correcting each target field in the target text set to obtain a corrected target text set.
In this embodiment, since the identified target text set may have target fields that are not standard physical examination terms, at this time, an approximate field to be corresponding may be searched in the standard field set for each target field in the target text set (this approximate field may be completely the same as the target field, or may have a difference of 1-2 characters), and each target field in the target text set is corrected by the approximate field of each target field, so as to obtain a corrected target text set. Through the field correction process, the obtained text recognition result is more accurate.
In one embodiment, step S160 includes:
acquiring the character string editing distance between each target field in the target text set and each standard field in the standard field set, and taking the standard field with the minimum character string editing distance with each target field as the approximate field corresponding to each target field;
judging whether each target field is the same as the corresponding approximate field;
and if the target field is different from the corresponding approximate field, replacing the target field with the corresponding approximate field to obtain the corrected target text set.
In the present embodiment, the character string edit distance is used to determine the similarity between two characters. The string edit distance refers to the minimum number of operands required to convert string a to string B using character manipulation. Wherein the character operation includes: deleting a character; inserting a character; a character is modified. For example, for the character strings "if" and "iff", the purpose can be achieved by inserting one 'f' or deleting one 'f'.
And if all the target fields in the target text set are the same as the corresponding approximate fields, the target fields do not need to be corrected at the moment, and the target text set is directly used as a corrected target text set. If some target fields are different from the corresponding approximate fields, which indicates that the target fields are caused by non-standard expressions or recognition errors, the target fields can be replaced by the corresponding approximate fields to obtain the corrected target text set. By the approximate field replacement method based on the character string editing distance, the text recognition result can be corrected more quickly and accurately.
S170, storing the corrected target text set into a correspondingly created storage area to obtain the physical examination report structured information.
In this embodiment, the corrected target text set obtained after the field correction may be stored locally in the server as the final physical examination report structured information. When the storage area is created in advance, the name of the storage area can be the same as the terminal unique identification code of the user side, and a plurality of subfolders are arranged in the storage area to respectively store the physical examination report structured information corresponding to the physical examination report image picture sets uploaded by the user side in different operation time periods. By the aid of the partitioned storage mode, partitioned retrieval of the physical examination report structured information of each user is facilitated, and retrieval efficiency is improved.
In an embodiment, step S170 is followed by:
and uploading the physical examination report structured information to a blockchain network.
In this embodiment, the server may serve as a blockchain link point device to upload the physical examination report structured information to a blockchain network, and make full use of the characteristic that blockchain data cannot be tampered with, so as to implement data evidence solidification.
The physical examination report structured information is obtained by obtaining corresponding summary information, specifically, the summary information is obtained by performing hash processing on the physical examination report structured information, for example, by using a sha256 algorithm. Uploading summary information to the blockchain can ensure the safety and the fair transparency of the user. The user equipment can download the summary information from the blockchain so as to verify whether the physical examination report structured information is tampered. The blockchain referred to in this example is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, encryption algorithm, and the like. A block chain (Blockchain), which is essentially a decentralized database, is a series of data blocks associated by using a cryptographic method, and each data block contains information of a batch of network transactions, so as to verify the validity (anti-counterfeiting) of the information and generate a next block. The blockchain may include a blockchain underlying platform, a platform product service layer, an application service layer, and the like.
The method realizes the rapid positioning of the area to be identified, the accurate text identification of the area to be identified and the rapid structural extraction of the core physical examination data of the physical examination report.
The embodiment of the invention also provides a physical examination report information structured extraction device, which is used for executing any embodiment of the physical examination report information structured extraction method. Specifically, referring to fig. 4, fig. 4 is a schematic block diagram of a physical examination report information structured extraction apparatus according to an embodiment of the present invention. The physical examination report information structured extraction apparatus 100 may be configured in a server.
As shown in fig. 4, the physical examination report information structured extraction apparatus 100 includes: the physical examination report picture receiving unit 110, the sample set acquiring unit 120, the picture set locating unit 130 of the area to be identified, the text identifying unit 140, the target text set acquiring unit 150, the target text correcting unit 160 and the structured information acquiring unit 170.
The physical examination report picture receiving unit 110 is configured to receive a physical examination report image picture set uploaded by a user terminal, and acquire report source information corresponding to the physical examination report image picture set through OCR recognition; wherein the report source information comprises a physical examination report issuing organization name and a physical examination report type.
In this embodiment, both the application and underwriting systems may be deployed in the server, and when a user uses a client to establish a communication connection with the server and logs in the application system, a user interaction interface of the application system is displayed on the client. At this time, after selecting a risk category (e.g., a serious disease risk) previously applied to the insurance, the user uploads a claim data for the risk category (e.g., uploads a physical examination report image picture set as the claim data, the physical examination report image picture set includes a plurality of physical examination report image pictures, generally, the user uses a photographing function of the user terminal to photograph a plurality of paper-based physical examination reports to form a physical examination report image picture set), and the claim data is uploaded to the server and then transferred to the underwriting system for claim data examination.
After the underwriting system deployed in the server receives the physical examination report image picture set uploaded by the user side, in order to determine which type of physical examination report template of which physical examination report issuing organization is, the text of the local area of the physical examination report image picture set can be acquired through OCR recognition, so as to acquire corresponding report source information.
In one embodiment, the physical examination report picture receiving unit 110 includes:
and the report source information identification unit is used for acquiring the upper half area of the top page of the physical examination report image picture set, and acquiring the first three lines of characters corresponding to the upper half area of the top page through OCR identification so as to acquire the report source information corresponding to the physical examination report image picture set.
In this embodiment, different hospitals or physical examination institutions have different physical examination report templates, and the physical examination data arrangement areas of the physical examination report templates are different, more specifically, the first type of physical examination data (e.g., blood routine physical examination data) of hospital a is generally distributed in the lower half of the first page of the physical examination report, the second type of physical examination data (e.g., routine whole body physical examination data) of hospital a is generally distributed in the second to fifth pages of the physical examination report, and the physical examination data of hospital B is generally distributed in the lower half of the second page and the upper half of the third page of the physical examination report.
The reason why the text of the local area of the physical examination report image picture set (generally, the head-up part of the first page of the physical examination report image picture set, for example, the first 3 lines of text of the first page generally include the name of the hospital or the name of the physical examination institution and the type of the physical examination report, that is, the text is distributed in the head-up part of the first page of the physical examination report image picture set) is recognized is that the physical examination institution and the type of the physical examination report can be recognized in this area, so that the distribution area of the physical examination data can be rapidly determined.
The sample set obtaining unit 120 is configured to call a pre-stored physical examination report sample set, and obtain sample report source information and a sample physical examination data distribution area corresponding to each physical examination report sample.
In the embodiment, since the physical examination report samples respectively corresponding to the plurality of physical examination report issuing mechanisms are stored in the underwriting system of the server, the sample report source information and the sample physical examination data distribution area corresponding to each physical examination report sample are known. Thus, the report source information corresponding to the physical examination report image picture set is compared with the sample report source information of the physical examination report sample set one by one, and the distribution area of the examination data corresponding to the physical examination report image picture set can be rapidly judged and obtained.
The to-be-identified region picture set positioning unit 130 is configured to, if sample report source information corresponding to a physical examination report sample in the physical examination report sample set is the same as the report source information of the physical examination report picture set, obtain a sample physical examination data distribution region of the physical examination report sample, so as to position a to-be-identified region in the physical examination report picture set to form a to-be-identified region picture set.
In this embodiment, for example, when the report source information of the physical examination report picture set uploaded by the user terminal corresponds to a1 with the mechanism B1 physical examination report type, and the sample report source information corresponding to the physical examination report sample in the physical examination report sample set also corresponds to a1 with the mechanism B1 physical examination report type, sample physical examination data distribution regions of the physical examination report sample (specifically, physical examination data is distributed in the lower half of the second page and the upper half of the third page of the physical examination report) can be obtained, the same region in the physical examination report picture set is directly located according to the sample physical examination data distribution regions as the to-be-identified region, and the to-be-identified region picture sets are composed of to-be-identified region pictures corresponding to the to-be-identified regions. The areas to be identified are in sequence, that is, each area to be identified corresponds to a page attribute value, for example, if physical examination data is distributed on the lower half of the second page of the physical examination report, the page attribute value corresponding to the area is equal to 2, and if physical examination data is distributed on the upper half of the third page of the physical examination report, the page attribute value corresponding to the area is equal to 3.
For example, if the physical examination report image picture set corresponds to a1 physical examination report type with a mechanism B1, the lower half area of the second page in the physical examination report image picture set is located as an area to be identified 1 (the picture corresponding to the area to be identified 1 is referred to as an area picture to be identified 1), the upper half area of the third page in the physical examination report image picture set is located as an area to be identified 2 (the picture corresponding to the area to be identified 2 is referred to as an area picture to be identified 2), and the area picture set to be identified is composed of the area picture to be identified 1 and the area picture to be identified 2. By the mode of quickly positioning the image set of the physical examination report according to the physical examination report issuing mechanism and the physical examination report type, the accurate positioning area for extracting the physical examination report data is quickly screened out, so that the area for subsequently performing text recognition is reduced, and the efficiency of text recognition and data extraction is improved.
And the text recognition unit 140 is configured to sequentially remove perspective deformation and character recognition from the to-be-recognized region picture set to obtain a recognition text corresponding to the to-be-recognized region picture set.
In this embodiment, after the picture set of the region to be identified is obtained through the report source information of the image set of the physical examination report, in order to perform text identification more accurately, preprocessing for removing perspective distortion needs to be performed on the picture set before text identification, so that the text in the picture can be extracted more accurately and quickly.
In one embodiment, as shown in fig. 5, the text recognition unit 140 includes:
the perspective distortion removing processing unit 141 is configured to remove perspective distortion from the image set of the region to be identified by using a Warping algorithm to obtain a first processed image set;
a character cutting unit 142, configured to perform character cutting on each first processed picture in the first processed picture set to obtain a plurality of character cut sub-pictures to form a character cut sub-picture set;
the character recognition unit 143 is configured to invoke a pre-trained CRNN-CTC character recognition model, perform character recognition on each character cut sub-picture in the character cut sub-picture set through the CRNN-CTC character recognition model, and obtain character recognition results corresponding to each character cut sub-picture;
and an identification result combining unit 144, configured to sequentially perform serial combination on the character identification results respectively corresponding to the character cutting sub-pictures to obtain an identification text corresponding to the picture set of the region to be identified.
In this embodiment, since the to-be-identified region picture set corresponding to the physical examination report image picture set is already acquired, and the to-be-identified region pictures in the to-be-identified region picture set are arranged in an ascending order according to the size of the page attribute value, at this time, the perspective deformation processing is sequentially removed from each to-be-identified region picture in the to-be-identified region picture set through a Warping algorithm (i.e., an image deformation algorithm), so as to obtain first processing pictures respectively corresponding to each to-be-identified region picture, so as to form a first processing picture set.
In an embodiment, the removing perspective deformation processing unit 141 is further configured to:
and converting all intersected line segments of the pictures of the areas to be identified in the picture set of the areas to be identified into parallel line segments through a homography matrix in a Warping algorithm to remove perspective deformation, thereby obtaining a first processing picture set.
In the present embodiment, the removal of perspective distortion can be achieved by a homography matrix (homography). In computer vision, the homography of a plane is defined as the projection mapping from one plane to another. Through a function findHomography () provided by OpenCV (which is a cross-platform computer vision and machine learning software library based on BSD licensing), corresponding point sequences are used as input, homography matrixes which best describe the corresponding points are returned, and therefore the solution of the homography matrixes is achieved. All intersecting line segments of the pictures of the areas to be identified in the picture set of the areas to be identified can be converted into parallel line segments through the homography matrix in the Warping algorithm, and then the processing of removing geometric deformation can be added, so that a first processed picture set is obtained.
After the first processed picture set is obtained, in order to perform text recognition more accurately, character cutting needs to be performed on each first processed picture in the first processed picture set, that is, each first processed picture is refined and cut into a plurality of character cutting sub-pictures again to form a character cutting sub-picture set. Through the character cutting mode with finer granularity, the text recognition of small areas can be performed more accurately in the follow-up process. In specific implementation, a statistical segmentation method or a horizontal/vertical projection character cutting method can be selected to perform character cutting on each first processed picture in the first processed picture set so as to obtain a character cutting sub-picture set.
And calling a locally stored and trained CRNN-CTC character recognition model in the server after the character cutting sub-picture set is obtained so as to perform text recognition on each character cutting sub-picture in the character cutting sub-picture set. The CRNN network in the CRNN-CTC character recognition model is a neural network formed by mixing CNN (convolutional neural network) and RNN (recurrent neural network), the convolutional layer of the CRNN network is a CNN network (convolutional characteristic matrix for extracting input images), the recurrent network layer of the CRNN network is a deep bidirectional LSTM network (used for continuously extracting character sequence characteristics on the basis of the convolutional characteristic matrix), and extracted texts corresponding to the input images are output after the output of the recurrent network layer of the CRNN network is softmax. CTC (the english name of CTC is connection Temporal Classification, which represents connection timing Classification), that is, CTC is used to solve the problem that input sequences and output sequences are difficult to correspond one to one, and can improve the robustness of single-line text recognition.
After the character recognition results are sequentially recognized according to the sequence of cutting the sub-pictures by the characters, the character recognition results and the character recognition results can be sequentially connected in series and combined to obtain a recognition text corresponding to the picture set of the region to be recognized, wherein the recognition text comprises a plurality of characters (possibly English characters, numeric characters, symbolic characters and/or Chinese characters). After the identification of the picture set of the area to be identified is completed, the key data required by the subsequent user can be extracted.
And a target text set obtaining unit 150, configured to obtain the target field and the value of the target field in the recognition text through semantic analysis positioning to form a target text set.
In this embodiment, sentence-level semantic analysis is used to extract a target field and a value of the target field from the recognition text, and more specifically, a semantic role labeling method based on complete syntactic analysis may be used to perform semantic analysis. After the recognition text is given, syntactic analysis, candidate argument pruning, argument recognition, argument labeling and post-processing (processing the arguments after labeling or adding some richer information) are sequentially performed to obtain a target text set, so that the extraction of the target field and the value of the target field is realized.
In one embodiment, the target text set obtaining unit 150 includes:
the sentence acquisition unit to be recognized is used for dividing the recognition text into a plurality of sentences to be recognized according to separators;
and the target text set acquisition unit is used for acquiring a target field and a target field value which are respectively included in each sentence to be recognized through semantic analysis so as to form a target text set.
In this embodiment, since sentence-level semantic analysis is used to extract the target field and the target field value of the recognition text, before the semantic analysis, the recognition text may be divided into a plurality of sentences to be recognized according to separators (for example, periods, twiddle symbols, or the last end of each character cut sub-picture is automatically regarded as a separator), and then sentence-level semantic analysis may be used to extract the target field and the target field value of the recognition text.
And the target text correction unit 160 is configured to call a pre-stored standard field set, obtain an approximate field corresponding to each target field in the standard field set in the target text set, and correct each target field in the target text set to obtain a corrected target text set.
In this embodiment, since the identified target text set may have target fields that are not standard physical examination terms, at this time, an approximate field to be corresponding may be searched in the standard field set for each target field in the target text set (this approximate field may be completely the same as the target field, or may have a difference of 1-2 characters), and each target field in the target text set is corrected by the approximate field of each target field, so as to obtain a corrected target text set. Through the field correction process, the obtained text recognition result is more accurate.
In one embodiment, the target-text correcting unit 160 includes:
an approximate field obtaining unit, configured to obtain a string edit distance between each target field in the target text set and each standard field in the standard field set, and use a standard field having a minimum string edit distance with each target field as an approximate field corresponding to each target field;
a similar field judging unit for judging whether each target field is the same as the corresponding approximate field;
and the corrected target text set acquisition unit is used for replacing the target field with the corresponding approximate field if the target field is different from the corresponding approximate field so as to obtain the corrected target text set.
In the present embodiment, the character string edit distance is used to determine the similarity between two characters. The string edit distance refers to the minimum number of operands required to convert string a to string B using character manipulation. Wherein the character operation includes: deleting a character; inserting a character; a character is modified. For example, for the character strings "if" and "iff", the purpose can be achieved by inserting one 'f' or deleting one 'f'.
And if all the target fields in the target text set are the same as the corresponding approximate fields, the target fields do not need to be corrected at the moment, and the target text set is directly used as a corrected target text set. If some target fields are different from the corresponding approximate fields, which indicates that the target fields are caused by non-standard expressions or recognition errors, the target fields can be replaced by the corresponding approximate fields to obtain the corrected target text set. By the approximate field replacement method based on the character string editing distance, the text recognition result can be corrected more quickly and accurately.
A structural information obtaining unit 170, configured to store the corrected target text set in a correspondingly created storage area, so as to obtain physical examination report structural information.
In this embodiment, the corrected target text set obtained after the field correction may be stored locally in the server as the final physical examination report structured information. When the storage area is created in advance, the name of the storage area can be the same as the terminal unique identification code of the user side, and a plurality of subfolders are arranged in the storage area to respectively store the physical examination report structured information corresponding to the physical examination report image picture sets uploaded by the user side in different operation time periods. By the aid of the partitioned storage mode, partitioned retrieval of the physical examination report structured information of each user is facilitated, and retrieval efficiency is improved.
In one embodiment, the structural extraction device 100 for medical examination report information further comprises:
and the data uplink unit is used for uploading the physical examination report structural information to a block chain network.
In this embodiment, the server may serve as a blockchain link point device to upload the physical examination report structured information to a blockchain network, and make full use of the characteristic that blockchain data cannot be tampered with, so as to implement data evidence solidification.
The physical examination report structured information is obtained by obtaining corresponding summary information, specifically, the summary information is obtained by performing hash processing on the physical examination report structured information, for example, by using a sha256 algorithm. Uploading summary information to the blockchain can ensure the safety and the fair transparency of the user. The user equipment can download the summary information from the blockchain so as to verify whether the physical examination report structured information is tampered. The blockchain referred to in this example is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, encryption algorithm, and the like. A block chain (Blockchain), which is essentially a decentralized database, is a series of data blocks associated by using a cryptographic method, and each data block contains information of a batch of network transactions, so as to verify the validity (anti-counterfeiting) of the information and generate a next block. The blockchain may include a blockchain underlying platform, a platform product service layer, an application service layer, and the like.
The device realizes the quick positioning of the area to be identified, the accurate text identification of the area to be identified and the quick structured extraction of the core physical examination data of the physical examination report.
The physical examination report information structured extraction means can be implemented in the form of a computer program which can be run on a computer device as shown in fig. 6.
Referring to fig. 6, fig. 6 is a schematic block diagram of a computer device according to an embodiment of the present invention. The computer device 500 is a server, and the server may be an independent server or a server cluster composed of a plurality of servers.
Referring to fig. 6, the computer device 500 includes a processor 502, memory, and a network interface 505 connected by a system bus 501, where the memory may include a non-volatile storage medium 503 and an internal memory 504.
The non-volatile storage medium 503 may store an operating system 5031 and a computer program 5032. The computer program 5032, when executed, can cause the processor 502 to perform a physical examination report information structured extraction method.
The processor 502 is used to provide computing and control capabilities that support the operation of the overall computer device 500.
The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503, and when the computer program 5032 is executed by the processor 502, the processor 502 can be caused to execute the physical examination report information structured extraction method.
The network interface 505 is used for network communication, such as providing transmission of data information. Those skilled in the art will appreciate that the configuration shown in fig. 6 is a block diagram of only a portion of the configuration associated with aspects of the present invention and is not intended to limit the computing device 500 to which aspects of the present invention may be applied, and that a particular computing device 500 may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
The processor 502 is configured to run the computer program 5032 stored in the memory to implement the method for structured extraction of physical examination report information disclosed in the embodiment of the present invention.
Those skilled in the art will appreciate that the embodiment of a computer device illustrated in fig. 6 does not constitute a limitation on the specific construction of the computer device, and that in other embodiments a computer device may include more or fewer components than those illustrated, or some components may be combined, or a different arrangement of components. For example, in some embodiments, the computer device may only include a memory and a processor, and in such embodiments, the structures and functions of the memory and the processor are consistent with those of the embodiment shown in fig. 6, and are not described herein again.
It should be understood that, in the embodiment of the present invention, the Processor 502 may be a Central Processing Unit (CPU), and the Processor 502 may also be other general purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Wherein a general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
In another embodiment of the invention, a computer-readable storage medium is provided. The computer readable storage medium may be a non-volatile computer readable storage medium. The computer readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements the structured extraction method for physical examination report information disclosed by the embodiments of the present invention.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described apparatuses, devices and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again. Those of ordinary skill in the art will appreciate that the elements and algorithm steps of the examples described in connection with the embodiments disclosed herein may be embodied in electronic hardware, computer software, or combinations of both, and that the components and steps of the examples have been described in a functional general in the foregoing description for the purpose of illustrating clearly the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
In the embodiments provided by the present invention, it should be understood that the disclosed apparatus, device and method can be implemented in other ways. For example, the above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only a logical division, and there may be other divisions when the actual implementation is performed, or units having the same function may be grouped into one unit, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may also be an electric, mechanical or other form of connection.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment of the present invention.
In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a storage medium. Based on such understanding, the technical solution of the present invention essentially or partially contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a magnetic disk, or an optical disk.
While the invention has been described with reference to specific embodiments, the invention is not limited thereto, and various equivalent modifications and substitutions can be easily made by those skilled in the art within the technical scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims (10)

1. A physical examination report information structured extraction method is characterized by comprising the following steps:
receiving a physical examination report image picture set uploaded by a user side, and acquiring report source information corresponding to the physical examination report image picture set through OCR recognition; wherein the report source information comprises a physical examination report issuing organization name and a physical examination report type;
calling a pre-stored physical examination report sample set to acquire sample report source information and a sample physical examination data distribution area corresponding to each physical examination report sample;
if sample report source information corresponding to a physical examination report sample in the physical examination report sample set is the same as the report source information of the physical examination report image picture set, acquiring a sample physical examination data distribution area of the physical examination report sample so as to position an area to be identified in the physical examination report image picture set to form an area to be identified picture set;
sequentially removing perspective deformation and character recognition on the picture set of the area to be recognized to obtain a recognition text corresponding to the picture set of the area to be recognized;
obtaining a target field and a target field value in the recognition text through semantic analysis positioning to form a target text set;
calling a pre-stored standard field set, acquiring an approximate field corresponding to each target field in the target text set in the standard field set, and correcting each target field in the target text set to obtain a corrected target text set; and
and storing the corrected target text set to a correspondingly created storage area to obtain the structural information of the physical examination report.
2. The structured extraction method of physical examination report information according to claim 1, wherein the obtaining of report source information corresponding to the set of physical examination report image pictures through OCR recognition comprises:
and acquiring the upper half area of the top page of the physical examination report image picture set, and acquiring the first three lines of characters corresponding to the upper half area of the top page through OCR recognition to acquire report source information corresponding to the physical examination report image picture set.
3. The physical examination report information structured extraction method of claim 1, wherein the step of sequentially removing perspective deformation and character recognition from the picture set of the region to be recognized to obtain the recognition text corresponding to the picture set of the region to be recognized comprises:
removing perspective deformation from the picture set of the region to be identified through a Warping algorithm to obtain a first processed picture set;
performing character cutting on each first processed picture in the first processed picture set to obtain a plurality of character cutting sub-pictures so as to form a character cutting sub-picture set;
calling a pre-trained CRNN-CTC character recognition model, and performing character recognition on each character cutting sub-picture in the character cutting sub-picture set through the CRNN-CTC character recognition model to obtain character recognition results corresponding to each character cutting sub-picture;
and sequentially carrying out serial combination on the character recognition results respectively corresponding to the character cutting sub-pictures to obtain recognition texts corresponding to the picture set of the area to be recognized.
4. The physical examination report information structured extraction method of claim 3, wherein the removing perspective distortion from the picture set of the region to be identified by a Warping algorithm to obtain a first processed picture set comprises:
and converting all intersected line segments of the pictures of the areas to be identified in the picture set of the areas to be identified into parallel line segments through a homography matrix in a Warping algorithm to remove perspective deformation, thereby obtaining a first processing picture set.
5. The physical examination report information structured extraction method of claim 1, wherein the obtaining of the target field and the value of the target field in the recognition text through semantic analysis and positioning to form a target text set comprises:
dividing the recognition text into a plurality of sentences to be recognized according to separators;
and obtaining a target field and a target field value respectively included by each sentence to be recognized through semantic analysis to form a target text set.
6. The method of claim 1, wherein the obtaining of the approximate field corresponding to each target field in the target text set in the standard field set and the correcting of each target field in the target text set to obtain the corrected target text set comprises:
acquiring the character string editing distance between each target field in the target text set and each standard field in the standard field set, and taking the standard field with the minimum character string editing distance with each target field as the approximate field corresponding to each target field;
judging whether each target field is the same as the corresponding approximate field;
and if the target field is different from the corresponding approximate field, replacing the target field with the corresponding approximate field to obtain the corrected target text set.
7. The structured extraction method for physical examination report information according to claim 1, wherein after storing the corrected target text set in a correspondingly created storage area to obtain the structured information for physical examination report, the method further comprises:
and uploading the physical examination report structured information to a blockchain network.
8. A physical examination report information structured extraction device is characterized by comprising:
the physical examination report picture receiving unit is used for receiving a physical examination report image picture set uploaded by a user terminal and acquiring report source information corresponding to the physical examination report image picture set through OCR recognition; wherein the report source information comprises a physical examination report issuing organization name and a physical examination report type;
the system comprises a sample set acquisition unit, a sample analysis unit and a sample analysis unit, wherein the sample set acquisition unit is used for calling a pre-stored physical examination report sample set and acquiring sample report source information and a sample physical examination data distribution area corresponding to each physical examination report sample;
the picture set positioning unit of the area to be identified is used for acquiring a sample physical examination data distribution area of the physical examination report sample if sample report source information corresponding to the physical examination report sample in the physical examination report sample set is the same as the report source information of the physical examination report picture set so as to position the area to be identified in the physical examination report picture set to form a picture set of the area to be identified;
the text recognition unit is used for sequentially removing perspective deformation and character recognition on the picture set of the area to be recognized to obtain a recognition text corresponding to the picture set of the area to be recognized;
the target text set acquisition unit is used for acquiring a target field and a target field value in the recognition text through semantic analysis positioning so as to form a target text set;
the target text correction unit is used for calling a pre-stored standard field set, acquiring approximate fields corresponding to all target fields in the target text set in the standard field set, and correcting all target fields in the target text set to obtain a corrected target text set; and
and the structural information acquisition unit is used for storing the corrected target text set into a correspondingly created storage area so as to obtain the structural information of the physical examination report.
9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the physical examination report information structured extraction method according to any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, characterized in that it stores a computer program which, when executed by a processor, causes the processor to carry out the structured extraction method of physical examination report information according to any one of claims 1 to 7.
CN202011149638.3A 2020-10-23 2020-10-23 Physical examination report information structured extraction method and device and computer equipment Active CN112257613B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011149638.3A CN112257613B (en) 2020-10-23 2020-10-23 Physical examination report information structured extraction method and device and computer equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011149638.3A CN112257613B (en) 2020-10-23 2020-10-23 Physical examination report information structured extraction method and device and computer equipment

Publications (2)

Publication Number Publication Date
CN112257613A true CN112257613A (en) 2021-01-22
CN112257613B CN112257613B (en) 2023-08-08

Family

ID=74261766

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011149638.3A Active CN112257613B (en) 2020-10-23 2020-10-23 Physical examination report information structured extraction method and device and computer equipment

Country Status (1)

Country Link
CN (1) CN112257613B (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112819003A (en) * 2021-04-19 2021-05-18 北京妙医佳健康科技集团有限公司 Method and device for improving OCR recognition accuracy of physical examination report
CN112836632A (en) * 2021-02-02 2021-05-25 浪潮云信息技术股份公司 Method and system for realizing user-defined template character recognition
CN113138916A (en) * 2021-04-06 2021-07-20 青岛以萨数据技术有限公司 Automatic testing method and system for picture structuring algorithm based on labeled sample
CN113342883A (en) * 2021-05-25 2021-09-03 国网上海市电力公司 Power equipment detection data structuring method, device, medium and equipment
CN114241487A (en) * 2021-12-20 2022-03-25 北京妙医佳健康科技集团有限公司 OCR recognition method
CN114360679A (en) * 2022-01-07 2022-04-15 启康保(北京)健康科技有限公司 Physical examination report digital management method and system
CN115019925A (en) * 2022-06-01 2022-09-06 科亚医疗科技股份有限公司 Automatic classification method and automatic classification device for health examination reports

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2019174130A1 (en) * 2018-03-14 2019-09-19 平安科技（深圳）有限公司 Bill recognition method, server, and computer readable storage medium
CN110781726A (en) * 2019-09-11 2020-02-11 深圳壹账通智能科技有限公司 Image data identification method and device based on OCR (optical character recognition), and computer equipment
CN111275038A (en) * 2020-01-17 2020-06-12 平安医疗健康管理股份有限公司 Image text recognition method and device, computer equipment and computer storage medium

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2019174130A1 (en) * 2018-03-14 2019-09-19 平安科技（深圳）有限公司 Bill recognition method, server, and computer readable storage medium
CN110781726A (en) * 2019-09-11 2020-02-11 深圳壹账通智能科技有限公司 Image data identification method and device based on OCR (optical character recognition), and computer equipment
CN111275038A (en) * 2020-01-17 2020-06-12 平安医疗健康管理股份有限公司 Image text recognition method and device, computer equipment and computer storage medium

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112836632A (en) * 2021-02-02 2021-05-25 浪潮云信息技术股份公司 Method and system for realizing user-defined template character recognition
CN113138916A (en) * 2021-04-06 2021-07-20 青岛以萨数据技术有限公司 Automatic testing method and system for picture structuring algorithm based on labeled sample
CN113138916B (en) * 2021-04-06 2024-04-30 青岛以萨数据技术有限公司 Automatic testing method and system for picture structuring algorithm based on labeling sample
CN112819003A (en) * 2021-04-19 2021-05-18 北京妙医佳健康科技集团有限公司 Method and device for improving OCR recognition accuracy of physical examination report
CN113342883A (en) * 2021-05-25 2021-09-03 国网上海市电力公司 Power equipment detection data structuring method, device, medium and equipment
CN114241487A (en) * 2021-12-20 2022-03-25 北京妙医佳健康科技集团有限公司 OCR recognition method
CN114360679A (en) * 2022-01-07 2022-04-15 启康保(北京)健康科技有限公司 Physical examination report digital management method and system
CN115019925A (en) * 2022-06-01 2022-09-06 科亚医疗科技股份有限公司 Automatic classification method and automatic classification device for health examination reports

Also Published As

Publication number Publication date
CN112257613B (en) 2023-08-08

Similar Documents

Publication Publication Date Title
CN112257613B (en) Physical examination report information structured extraction method and device and computer equipment
US11232300B2 (en) System and method for automatic detection and verification of optical character recognition data
CN111680634B (en) Document file processing method, device, computer equipment and storage medium
US10489645B2 (en) System and method for automatic detection and verification of optical character recognition data
CN111695439A (en) Image structured data extraction method, electronic device and storage medium
US20210366055A1 (en) Systems and methods for generating accurate transaction data and manipulation
US10740940B2 (en) Automatic generation of fundus drawings
CN109872784A (en) Intelligent case history generation method, device, computer equipment and storage medium
CN112036145A (en) Financial statement identification method and device, computer equipment and readable storage medium
CN112036295B (en) Bill image processing method and device, storage medium and electronic equipment
CN117523593B (en) Patient medical record data processing method and system
CN113536771B (en) Element information extraction method, device, equipment and medium based on text recognition
CN114913942A (en) Intelligent matching method and device for patient recruitment projects
US11527102B2 (en) Systems and methods of automated biometric identification reporting
CN113918686A (en) Intelligent question-answering model construction method and device, computer equipment and storage medium
CN111930976A (en) Presentation generation method, device, equipment and storage medium
RU2597163C2 (en) Comparing documents using reliable source
CN111144416A (en) Information processing method and device
CN116226391A (en) Specific field name error correction method and system, storage medium and terminal
US20220044048A1 (en) System and method to recognise characters from an image
TWM593008U (en) Form content identification system
JP2023081564A (en) Information representation pattern generation device and method
CN117542487A (en) Medical data acquisition and intelligent processing system
CN117351508A (en) Identity information verification method, device, equipment and medium for insurance scene
CN116778495A (en) Medical document image analysis method, device, equipment and medium

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