Ancient document identification processing system

Abstract

The invention relates to the technical field of optical character recognition, in particular to an ancient document recognition processing system. The processing system comprises a login system, a management system, a task distribution system and an ancient book processing system, all the systems work cooperatively, ancient book images are segmented, recognized and corrected through the ancient book processing system, the ancient book images are segmented with vertical columns as units in the process, the segmented units are recognized, and the ancient book images are corrected. On the basis of recognition, through key prompt proofreading, column-to-column proofreading and clustering proofreading, the proofreading efficiency is greatly improved, and the proofreading recognition result is subjected to format distribution, so that the sorted data file can be output, and the whole ancient document recognition processing process is completed. According to the processing system, the proofreading efficiency is effectively improved, so that the ancient document identification processing efficiency is greatly improved, the data file conforming to the ancient book reading style can be obtained, and the processing system has good application and popularization prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of optical character recognition, and in particular to an ancient book and document recognition and processing system. Background Art

[0002] Optical Character Recognition (OCR) is a method of automatic text input. It refers to the process of scanning text materials, recognizing text through optical technology, obtaining image information on paper through photoelectric conversion, analyzing the morphological characteristics of text using various pattern recognition algorithms, determining the standard encoding of text, and storing it in a text file in a universal format to obtain text and layout information.

[0003] The OCR system is a key component of digital processing, primarily encompassing layout analysis and Chinese character recognition. Professional OCR, supplemented by manual proofreading, can produce highly accurate copies of ancient texts, significantly outperforming manual input and facilitating the use of digitized ancient books. Key metrics for measuring OCR system performance include rejection rate, false positive rate, recognition speed, user interface friendliness, product stability, ease of use, and feasibility. Debugging or leveraging auxiliary information to improve recognition accuracy is a crucial OCR issue.

[0004] When processing ancient books and documents, the existing OCR system basically includes the steps of segmentation, recognition and proofreading. In order to ensure the accuracy, multiple proofreadings are required during the process. For example, the existing segmentation basically segments the ancient book images into single words. Due to the adhesion between characters, single word segmentation is prone to misjudgment, so manual proofreading is required. In the process of character recognition, the recognition results also need to be corrected. Multiple corrections and checks have greatly reduced work efficiency, increased workload and caused waste of human resources. Therefore, an ancient book and document recognition and processing system that optimizes the above processing process to effectively improve processing efficiency is a technical problem that needs to be solved. Summary of the Invention

[0005] The purpose of the present invention is to provide an ancient book document recognition and processing system to solve the problem that in the existing ancient book processing process, proofreading is required after segmentation, which is not conducive to efficient ancient book processing.

[0006] To achieve the above-mentioned purpose, the present invention adopts the following technical solutions: an ancient book document recognition and processing system, comprising an ancient book processing system;

[0007] The ancient book processing system includes segmentation, recognition and proofreading;

[0008] The segmentation is to segment the scanned image of the ancient book into a plurality of units in columns or rows;

[0009] The recognition process recognizes each of the divided units, and arranges the recognition result of a unit in parallel with the unit, highlighting the characters that are difficult to recognize.

[0010] The proofreading is to compare and calibrate the recognition results placed side by side with the segmentation unit, and cluster and proofread the characters with the same recognition; the corrected recognition results are output as a data file.

[0011] Furthermore, it also includes a task allocation system, which allocates different volumes of ancient books to different groups or individuals to perform tasks in the ancient book processing system.

[0012] Furthermore, it also includes a login system and a management system;

[0013] The login system includes a login module for logging into the processing system;

[0014] The management system includes a role management system and a task statistics system. The role management system assigns roles and permissions to login personnel; the task statistics system counts the completion of tasks assigned by the task assignment system.

[0015] Furthermore, the segmentation is performed in units of vertical columns, and position information is assigned to each segmented unit.

[0016] Furthermore, the proofreading also includes layout allocation, which is used to assign different attributes to different areas of the corrected recognition result, including but not limited to title, author, text and comments.

[0017] Furthermore, the ancient book processing system also includes sorting and quality inspection;

[0018] The sorting is based on the position information and layout distribution during segmentation, and the recognition results are arranged in the data file;

[0019] The quality inspection performs a final check and calibration on the sorted recognition results.

[0020] Furthermore, the data files include json files, xml files, txt files and pdf files.

[0021] Beneficial effects of the present invention:

[0022] 1. The present invention eliminates the manual proofreading step after the existing single word segmentation and adopts the method of segmenting the entire line, which avoids the situation that the existing single word cannot be segmented. It improves the segmentation efficiency and does not need to perform manual proofreading of the segmentation, which can effectively improve the processing efficiency.

[0023] 2. The present invention eliminates the need for manual proofreading after segmentation and performs manual proofreading after recognition. In the process of manual proofreading after recognition, the method of highlighting difficult characters, parallel proofreading and cluster proofreading is adopted to greatly improve the efficiency of proofreading;

[0024] 3. The present invention distributes the format of the proofread recognition results, thereby being able to output data files that conform to the reading habits of ancient books, and has good application and promotion prospects. DETAILED DESCRIPTION

[0025] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention.

[0026] In the processing system of the present invention, a user logs into the processing system of the present invention through a login system, wherein the login system includes a login module, and the user inputs a user name and a login password through the login module to log into the processing system of the present invention.

[0027] In the system of the present invention, a super administrator account is set up. In the management system, the super administrator can assign roles and permissions to different logged-in users.

[0028] The user's permissions include establishing groups and assigning group members. The task allocation system can assign ancient book processing tasks to different groups, such as assigning the segmentation task to Group A, the recognition task to Group B, and the correction task to Group C; or assigning the segmentation and recognition tasks to Group A, and the correction task to Group B, etc., and can further subdivide the tasks within the group, such as assigning some volumes to individual A, and some volumes to individual B, etc.

[0029] After the task is assigned, the processing operation of the ancient books can be carried out. For the segmentation operation, in the system of the present invention, the ancient book image is segmented according to column or line units. For example, common ancient books are written and read from right to left, and are segmented column by column, and each column is given a position parameter; although common ancient books are written in vertical columns, it is not ruled out that some content is written in rows from right to left. Therefore, during the segmentation stage, the system of the present invention requires staff to proofread. It should be noted that this proofreading is a different concept from the traditional single-word segmentation and proofreading. The proofreading purpose of the system of the present invention here is two. One is the above-mentioned, for the content written in rows, the specific part is adjusted to horizontal unit segmentation, and the second is to avoid the situation where some characters are not segmented out, resulting in omissions.

[0030] The tasks after segmentation are identified. In the system of the present invention, identification is performed based on the units after segmentation. The recognition results after identification are placed side by side with the units before identification, and difficult characters and uncommon characters are highlighted during the recognition process to facilitate the subsequent proofreading process. The parallel placement and highlighting of difficult-to-identify positions in the system of the present invention can effectively improve the efficiency of proofreading. In addition, cluster proofreading is also provided in the system of the present invention, that is, the same characters in the recognition process are aggregated, and then the staff selects the correct characters and corrects them in batches, thereby greatly improving the efficiency of proofreading.

[0031] In the management system of the present invention, statistics can be collected on the task completion status of each group to facilitate the understanding of the processing progress of the work.

[0032] The proofread recognition results are assigned to a format in the system of the present invention. Through the format assignment, specific parts of the recognition results are assigned specific names, such as title, author, text and comments, etc., and then a final quality inspection is carried out. The data can be output as data files in various formats including json files, xml files, txt files and pdf files.

[0033] Example 1

[0034] In this embodiment, the system of the present invention is used to process "Three Commentaries on Treatise on Febrile Diseases".

[0035] During the process, a book image is obtained by a scanner, and the obtained image is processed by the processing system of the present invention, wherein the vertical column segmentation of the present invention is adopted, and then the difficult problems are highlighted, parallel proofreading and cluster proofreading are used to process the book.

[0036] Specifically, for the "Treatise on Febrile Diseases" with three annotations, seventeen volumes and one volume of "Treatise on Febrile Diseases Medical Prescriptions and Rhymes" (Qianlong period of the Qing Dynasty) - a total of 6 volumes, with a total of 155,132 words, there were 4 members in the team, one was responsible for segmentation, and three were responsible for identification and proofreading. The sorting work was completed in about 4 hours and passed the quality inspection.

[0037] Before the processing system of the present invention was adopted, the method of single-word segmentation, primary proofreading, recognition, and secondary proofreading was adopted. During the process, twelve volumes of famous doctors' cases - a total of 12 volumes, with a total of 304,013 words, were processed. During the process, there were also four members in the team, one was responsible for segmentation and primary proofreading, and three were responsible for recognition and secondary proofreading. The sorting work was completed after about 12 hours and passed the quality inspection.

[0038] In contrast, before adopting the system of the present invention, the processing capacity was 6333 words / person*h; after adopting the system of the present invention, the processing capacity was 9695 words / person*h; the efficiency was improved by about 50% compared with the past, and the effect was obvious.

[0039] The present invention is not limited to the above-mentioned optimal implementation mode. Anyone can derive other forms of products under the inspiration of the present invention. However, no matter what changes are made in the shape or structure, any technical solution that is the same or similar to that of the present application falls within the scope of protection of the present invention.

Claims

1. A system for identifying and processing ancient books and documents, characterized by: Including ancient book processing system; The ancient book processing system includes segmentation, O recognition and proofreading; The segmentation is to segment the scanned image of the ancient book into a plurality of units in columns or rows; The recognition process recognizes each of the divided units, and arranges the recognition result of a unit in parallel with the unit, highlighting the characters that are difficult to recognize. The proofreading is to compare and calibrate the recognition results placed side by side with the segmentation unit, and cluster and proofread the characters with the same recognition; the corrected recognition results are output as a data file.

2. The ancient book and document recognition and processing system according to claim 1, characterized in that: It also includes a task allocation system, which allocates different volumes of ancient books to different groups or individuals to perform tasks in the ancient book processing system.

3. The ancient book and document recognition and processing system according to claim 2, characterized in that: It also includes login system and management system; The login system includes a login module for logging into the processing system; The management system includes a role management system and a task statistics system. The role management system assigns roles and permissions to login personnel; the task statistics system counts the completion of tasks assigned by the task assignment system.

4. The ancient book and document recognition and processing system according to claim 1, characterized in that: The segmentation is performed using vertical columns as units, and position information is assigned to each segmented unit.

5. The ancient book and document recognition and processing system according to claim 4, characterized in that: The proofreading also includes layout allocation, which is used to assign different attributes to different areas of the corrected recognition result, including but not limited to title, author, text and comments.

6. The ancient book and document recognition and processing system according to claim 5, characterized in that: The ancient book processing system also includes sorting and quality inspection; The sorting is based on the position information and layout distribution during segmentation, and the recognition results are arranged in the data file; The quality inspection performs a final check and calibration on the sorted recognition results.

7. The ancient book and document recognition and processing system according to claim 1, characterized in that: The data files include json files, xml files, txt files and pdf files.