Comprehensive archive management system and management method

Abstract

The invention discloses a comprehensive archive management system and management method, and relates to the technical field of archive management, the comprehensive archive management system comprises a data entry module, an analysis adjustment module, a picture processing module, a data processing module and a data storage module, and the matched comprehensive archive management method is provided. According to the method, analysis processing and angle adjustment processing are carried out on the obtained archive picture, the angle of the picture is corrected, follow-up checking by related technicians and computer recognition are facilitated, then stain removal processing is carried out, so that the content on the picture is clearer, character filling processing is further carried out, the obtained archive content is complete, and the efficiency is improved. In general, clear and complete archive picture data can be obtained, and the method has a good use prospect.

Description

Technical Field

[0001] The present invention relates to the technical field of archive management, and in particular to a comprehensive archive management system and a management method. Background Art

[0002] Archives refer to original records of various forms with preservation value directly formed by people in various social activities. There are many types of archives. For example, comprehensive archives include project archives. Project archives are historical records of various forms with preservation value formed during the project construction and management process. A project generates a large amount of documents and materials from the process of project establishment, design, construction, supervision, and acceptance. In order to make the project archives complete, accurate, and accurate, a set of effective archive management methods is needed to ensure it.

[0003] At present, archives informatization is gradually being developed, and the archives management model has shifted from focusing on the preservation and utilization of archive entities to focusing on the digital storage of archives information and the provision of services, thus making archives work further move towards standardization, digitization, networking and socialization.

[0004] In the prior art, for comprehensive archive management, some studies have adopted the establishment of an image evaluation model, through which the scanned paper archive images are quality tested to judge whether the uploaded data is standard. However, in the above studies, there is no reference to the subsequent processing of archives. However, in the process of archive management, the photographed archive pictures are clearly an important aspect of archive informatization, but it also includes other important aspects, such as how to achieve convenient and accurate archive search, such as enterprise project archives, which have a wide range of content; and most of the uploaded materials are uploaded by taking photos, which are in the format of pictures, so it is impossible to quickly search directly through keywords; at the same time, when searching, it is very troublesome to quickly search for the archive and the required content information, which requires a lot of resources and energy. Summary of the invention

[0005] 1. Technical issues to be resolved In view of the deficiencies in the prior art, the present invention provides a comprehensive archive management system and management method, which analyzes and processes the acquired archive images and adjusts the angles to facilitate subsequent viewing by relevant technical personnel and computer recognition, and then performs stain removal processing to make the content on the image clearer, and further performs text filling processing to make the acquired archive content complete and convenient for subsequent viewing.

[0006] (II) Technical solution To achieve the above objectives, the present invention is implemented through the following technical solutions: A comprehensive archive management system includes a data entry module, an analysis and adjustment module, an image processing module, a data processing module and a data storage module: The data entry module determines whether the data source is electronic data or paper documents. If it is a paper document, a camera is used to shoot and enter the data. The analysis and adjustment module obtains the parameter data of the captured image, calculates the image resolution PPI and the number of paper document corners S in the recorded image, and compares the image resolution PPI with the set minimum resolution PPIb, and compares the number of paper document corners S with the standard number of corners Sb, to determine whether the image resolution and integrity meet the standards, and if they do not meet the standards, feeds back to the data entry module to re-enter the data; thereby completing the analysis and adjustment of the image; The image processing module obtains the analyzed and adjusted image, and sequentially performs angle adjustment, stain removal and text filling on the image to obtain a standard image; The data processing module obtains the standard image and performs text recognition on the standard image, obtains the text information stored in the image and the text information in the document in the electronic data and extracts the key words, generates the folder name, and compares the key words with the folder name stored in the data storage module for duplicate checking, and adds the keyword information to regenerate the folder name when duplication occurs; The data storage module arranges the entered standard images and document data (i.e., electronic data) in the order of entry, aggregates them into folders, and names the folders with the generated folder names.

[0007] Preferably, the image data that meets the standard is PPI≥PPIb, and S=Sb, wherein the calculation formula of the image resolution is: ; Where PPI is the number of pixels per inch in the image, X is the number of pixels in the length of the photo, Y is the number of pixels in the width of the photo, and Z is the diagonal length of the photo.

[0008] The value of Sb is 4, because the paper is rectangular with four corners. When there are three corners or five corners, that is, the picture is beyond the shooting range or the paper is folded, there is a high possibility that the shooting is incomplete, so a quick warning is needed.

[0009] When the present invention analyzes the archive, the analyzed data can be directly obtained from the picture, the analysis speed is very fast, and the results can be analyzed in real time, so as to timely warn the uploader, avoid the upload error and disordered upload sequence of the uploader, thereby improving the accuracy of archive entry and having a good use effect.

[0010] Preferably, the steps of adjusting the angle of the image are as follows: Calculate the image tilt angle; The calculation formula for the image tilt angle is as follows: ; Among them, θ is the tilt angle of the image, (X1, Y1) (X2, Y2) are the coordinates of the two corners at the bottom of the paper document; Determine whether the image tilt angle meets the set standard; Compare the calculated θ with the set threshold interval [0-δ, 0+δ], where δ is [0, 8], When θ is within the set threshold range, the image tilt angle meets the standard and no angle adjustment is required; When θ is outside the set threshold range, the image tilt angle does not meet the standard, and the image angle is adjusted; Adjust the angle of images that do not meet the set threshold range.

[0011] When shooting and uploading, the paper document may be tilted. At this time, the photographed paper document image will also be tilted, which will affect subsequent processing, so the angle needs to be adjusted.

[0012] Preferably, the calculation formula for adjusting the angle of the image that does not meet the set threshold range is as follows: ; Where d(x, y) is the output image, f(x, y) is a two-dimensional function representing the input image with integer pixel coordinates, processed using square root operations, and its convolution kernel is as follows; .

[0013] You can also rotate a fixed angle around a point.

[0014] Preferably, the steps of the stain removal process are as follows: obtaining text color data and stain color data, and calculating the distance between the text color data and the stain color data, determining whether the distance meets a set value range, and removing other color stains that are not within the value range.

[0015] Preferably, the calculation formula for calculating the difference between the text color data and the defaced color data is as follows: ; In the formula, (R w , G w , B w ) is the RGB value corresponding to the text color, (R s , G s , B s) is the RGB value corresponding to the contaminated color, and H is the distance between the two colors.

[0016] Preferably, when judging whether the distance meets the set value range, H is compared with the set value range [0, λ], the value range of λ is [5, 12], and the data whose difference is not within the value range is set as the paper color.

[0017] Preferably, the steps of text filling processing are as follows: Get the spatial distance of the missing part and calculate the number of missing characters. The calculation formula is: number of missing characters Qssl = (next character position - previous character position) / single character width; Remove the background, extract the missing text features, and predict text A based on the text features m And the possible proportion K of the text m ; For missing parts filling, the score is calculated according to the characters and the possible proportion of the characters, and a filling order is generated according to the order of the scores, and the missing parts are filled in according to the order. The score calculation formula is as follows; ; In the formula, is the Ath predicted for the i-th missing m The possible proportion of words; Semantic judgment: Get the filled text data and the preceding and following data to judge whether the sentence is fluent and score it according to the degree of fluency. If the score is not within the set range, fill in the next group of text according to the filling order until the score meets the set range. The scoring criteria include the word Cy, the sentence semantics Yy, and the correlation between the previous and next content Gd. The scoring formula is as follows: ; In the formula, Pf is the score, C´y is the weight corresponding to the word item, Y´y is the weight corresponding to the semantic item, and G´d is the weight corresponding to the relevance item of the previous and next contents.

[0018] Preferably, the steps of extracting keywords and generating folder names are as follows: Count the acquired data and sort the keywords according to the number of times they appear in the data; Keyword screening: remove keywords other than nouns and keep the top-ranked T group keywords; Generate the folder name by combining the task name or project name corresponding to the file with the first V items in the T group of keywords, where V<T.

[0019] Preferably, a comprehensive archive management method comprises the following steps: Determine whether the data source is electronic data or a paper document. If it is a paper document, use a camera to capture and input it; Obtain the parameter data of the captured image, calculate the image resolution PPI and the number of paper document corners S in the recorded image, compare the image resolution PPI with the set minimum resolution PPIb, and compare the number of paper document corners S with the standard number of corners Sb, determine whether the image resolution and integrity meet the standards, and if they do not meet the standards, feedback to the data entry module to re-enter the data; Obtain the analyzed and adjusted image, and sequentially perform angle adjustment processing, stain removal processing, and text filling processing on the image to obtain a standard image; Obtain a standard image, perform text recognition on the standard image, obtain text information stored in the image and text information in the document in the electronic data and extract key words, generate a folder name, compare the key words with the folder name stored in the data storage module for duplicate checking, and add keyword information to regenerate the folder name when duplication occurs; Arrange the entered standard images and document data in the order in which they were entered, summarize them into folders, and name the folders with the generated folder names.

[0020] (III) Beneficial effects The present invention provides a comprehensive archive management system and management method, which has the following beneficial effects: 1. The present invention can analyze and process the acquired archival images and adjust the angles, so as to facilitate subsequent viewing by relevant technical personnel and computer recognition, and then perform stain removal processing to make the content on the image clearer, and further perform text filling processing to make the acquired archival content complete and convenient for subsequent viewing.

[0021] 2. The present invention performs text recognition on the input image after adjusting the angle, extracts the text contained in the image, performs keyword judgment in combination with the text content in the document, extracts the key content of the project, and then generates a folder name by combining the keyword with the name of the project or the person. Subsequent personnel can realize rapid retrieval of the file by combining the keyword, project or person's name during the retrieval process, and the retrieval speed is greatly improved, which effectively solves the technical problem of inconvenient search of archives and archive contents in the prior art, and has good application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A modular structure diagram of a comprehensive archive management system in an embodiment of the present invention; Figure 2 4 is a flowchart of the steps of the comprehensive archive management method in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The technical solutions in the embodiments of the present invention are clearly and completely described below in conjunction with the accompanying drawings.

[0024] Example 1: Figure 1 As shown, a comprehensive archive management system includes a data entry module, an analysis and adjustment module, an image processing module, a data processing module, and a data storage module. In this implementation, it also includes an identity authentication module and a retrieval module. The identity authentication module and the retrieval module are optional modules. Adding the identity authentication module to the comprehensive archive management system can improve security. The retrieval module provides a specific retrieval implementation method, and other methods can also be used to implement retrieval.

[0025] File Entry Before admission, the user needs to identify the user's identity to determine whether he is the staff member who enters the file. Because the content of the file is confidential, only people with relevant permissions are qualified to enter the file; Identifying user identities is based on the identity authentication module. Common methods of the identity authentication module include face recognition, fingerprint recognition, voiceprint recognition, vein recognition, etc. through mobile devices.

[0026] After identity recognition, select the corresponding file entry option provided by the data entry module. At this time, the file data is entered. The data entry module determines whether the data source is electronic data or a paper document. When it is a paper document, a camera device is used to capture and enter the data.

[0027] When entering data, new and old file data will appear.

[0028] Most of the new archival data are electronic data (document data created by computers and other devices), which can be directly retrieved.

[0029] Most of the old archive data are paper files (in order to build an information system, it is necessary to process all previous data). When uploading paper files, the most common method is to take photos and then enter them into the system.

[0030] When it is a paper document, a camera is used to record the document or a scanner is used to scan the document (this recording is standard, so there will be no angle tilt, which can reduce the steps of angle adjustment).

[0031] Analysis Adjustment When the material is a paper document, the paper document may be tilted or unclear when photographed manually. In this case, it is necessary to analyze the image to determine whether the image meets the requirements; The analysis and adjustment of the image is based on the analysis and adjustment module. The analysis and adjustment module obtains the parameter data of the captured image, calculates the resolution PPI of the image and the number of corners S of the paper document in the recorded image, and compares the resolution PPI of the image with the set minimum resolution PPIb, and compares the number of corners S of the paper document with the standard number of corners Sb, to determine whether the resolution and integrity of the image meet the standards, and if not, feeds back to the data entry module to re-enter the data; ; Where PPI is the number of pixels per inch in the picture, X is the number of pixels in the length of the photo, Y is the number of pixels in the width of the photo, and Z is the diagonal length of the photo. Existing equipment and shooting instruments all have functions such as denoising, sharpening, motion compensation and fill light. Therefore, this item generally does not need to be adjusted.

[0032] The value of Sb is 4, because the paper is rectangular with four corners. When there are three or more corners, that is, the picture is beyond the shooting range or the paper is folded, there is a high possibility that the shooting is incomplete, so a quick warning is needed.

[0033] This step mainly checks whether the shooting is comprehensive to avoid missing shots.

[0034] Normal shooting is done on the desktop. The desktop and the paper have different colors. Therefore, the boundary shape of the paper can be clearly judged through the distribution of RGB values, so that the number of corners can be quickly determined, thereby realizing the determination of the number of corners.

[0035] You can also add a ratio setting. Since archival documents are all printed on A4 paper, you can also set the aspect ratio, and further determine whether the document is completely captured by the aspect ratio.

[0036] The present invention analyzes the archives, and the analyzed data can be directly obtained from the pictures. The analysis speed is very fast, and the results can be analyzed in real time, so as to timely warn the uploader, avoid the upload error and disordered upload sequence of the uploader, thereby improving the accuracy of archive entry and having a good use effect.

[0037] When shooting and uploading, the paper document may be tilted. At this time, the photographed paper document image will also be tilted, which will affect subsequent processing, so the angle needs to be adjusted.

[0038] The steps to adjust the angle of the image are as follows: Calculate the image tilt angle; The calculation formula for the image tilt angle is as follows: ; Among them, θ is the tilt angle of the image, (X1, Y1) (X2, Y2) are the coordinates of the two corners at the bottom of the paper document; Determine whether the image tilt angle meets the set standard; Compare the calculated θ with the set threshold interval [0-δ, 0+δ], where δ is [0, 8], When θ is within the set threshold range, the image tilt angle meets the standard and no angle adjustment is required; When θ is outside the set threshold range, the image tilt angle does not meet the standard, and the image angle is adjusted; Adjust the angle of images that do not meet the set threshold range.

[0039] For example, in the captured image, the coordinates of the two corners of the lower end of the paper document are (0, 0) and (2.0, 0.8), and the calculated angle is about 21.8°. At this time, the calculated angle is not within the set threshold range. At this time, the adjusted angle can be directly rotated -21.8°. In this case, the shooting plane is parallel to the paper placement plane, for example, the shooting is performed using a high-definition camera. The following methods can also be used.

[0040] ; Where d(x, y) is the output image, f(x, y) is a two-dimensional function representing the input image with integer pixel coordinates, processed using square root operations, and its convolution kernel is as follows; .

[0041] This situation mainly applies to the situation that the shooting plane and the plane on which the paper is located are not parallel.

[0042] Some materials are photographed with mobile phones. Since the plane of the mobile phone intersects with the paper, the angle adjustment is more complicated. The commonly used methods include Hough transform, projection, cross-correlation, nearest neighbor and Fourier transform to calculate the inclination angle, and then use the rotation transformation algorithm to rotate the pattern. This processing is mainly aimed at the situation when the shooting is more casual and the pictures taken are tilted in multiple directions.

[0043] The image processing module obtains the analyzed and adjusted image, and performs angle adjustment, stain removal and text filling processing on the image in sequence to obtain a standard image.

[0044] The steps of the stain removal process are as follows: obtain text color data and stain color data, and calculate the distance between the text color data and the stain color data, determine whether the distance meets the set value range, and remove other color stains that are not within the value range.

[0045] This removal is mainly aimed at seal-like contents in the archives, which block the text and make it unclear.

[0046] Preferably, the calculation formula for calculating the difference between the text color data and the defaced color data is as follows: ; In the formula, (R w , G w , B w ) is the RGB value corresponding to the text color, (R s , G s , B s ) is the RGB value corresponding to the contaminated color, and H is the distance between the two colors.

[0047] For example, if the text color is black, the corresponding RGB value (R w , G w , B w ) is (0, 0, 0). When the stain color is red, its corresponding RGB value (R s , G s , B s ) is (255, 0, 0).

[0048] When judging whether the distance is within the set value range, H is compared with the set value range [0, λ], the value range of λ is [5, 12], and the data whose difference is not within the value range is set as the paper color, which is equivalent to achieving the removal of contaminated color.

[0049] For example, when there is a seal, the R s The part is very high, keep the text part (R w , G w , B w ) text color value, and other values ​​(R s , G s , B s ) is set as the background color, which is the color of the paper. s , G s , B s ) to (255, 255, 255) to remove the dirty color data.

[0050] When paper is stored for a long time, the ink will fade, and the value of λ will also fade. This value range is for paper documents within 3 years. When the time is longer, the value of λ can be increased.

[0051] The steps of text filling processing are as follows: Get the spatial distance of the missing part and calculate the number of missing characters. The calculation formula is: number of missing characters Qssl = (next character position - previous character position) / single character width; Calculate how many Chinese characters are missing. Since the content format of the archives is unified, the number of missing characters can be directly calculated by distance.

[0052] Remove the background, extract the missing text features (no text filling is performed when the text features are missing), and predict the text A based on the text features m And the possible proportion K of the text m ; The text on the archives is normally written in Song font, which has regular strokes. When text features exist, the difficulty of filling in missing positions can be greatly reduced.

[0053] Predict text A based on text features m When the text features are compared with the text library, the text library stores commonly used texts, which are set according to the industry.

[0054] A m The value range is 0-5. If the data volume exceeds this range, it is too large. When the number of consecutive missing characters exceeds 5, A m The value range is 0-3.

[0055] K m The value range is 10-100. There are more characters below 10, the amount of data to be processed is relatively large, and the accuracy is not high, so they are not considered.

[0056] For missing parts filling, the score is calculated according to the characters and the possible proportion of the characters, and a filling order is generated according to the order of the scores, and the missing parts are filled in according to the order. The score calculation formula is as follows; ; In the formula, is the Ath predicted for the i-th missing m The possible proportion of words; For example, a row is missing two characters. Based on the character features, the predicted characters and possible proportions are shown in Table 1: Table 1 Prediction text and prediction ratio

[0057] The calculated scores are shown in Table 2: Table 2 Calculated scores

[0058] At this time, the filling order is A1B1, A1B2, A2B1, A1B3, A2B2, A2B3.

[0059] Fill in the first group in the order above, and then make semantic judgments; During semantic judgment, the filled text data and the preceding and following data are obtained to determine whether the sentence is fluent and to score it according to the degree of fluentness. If the score is not within the set range, the next group of text is filled in according to the filling order until the score meets the set range. The scoring criteria include the word Cy, the sentence semantics Yy, and the correlation between the previous and next content Gd. The scoring formula is as follows: ; Where Pf is the score, C´y is the weight corresponding to the word item, Y´y is the weight corresponding to the semantic item, and G´d is the weight corresponding to the relevance item of the previous and next contents, where Y´y>G´d>C´y, and Y´y+G´d+C´y=1.

[0060] The score range is [70,100].

[0061] For example, if the A1B1 score calculated above is 72, the added A1B1 will be directly used as the result; For example, if the score of A1B1 calculated above is 65, A1B2 is added for recalculation. If the score of A1B2 is 59, the next group is added until the score meets the set range. For example, if the score of A1B3 is 84, the added A1B3 is directly used as the result.

[0062] For example, the scores of the six groups of data calculated above are all lower than 70, that is, they do not meet the requirements. In this case, the group of text with the highest score is used as the supplementary result.

[0063] After all the images are processed, standard images are obtained. The purpose of obtaining standard images is to facilitate the acquisition of information contained in the images. This step is based on the data processing module.

[0064] The data processing module obtains the standard image, performs text recognition on the standard image, obtains the text information stored in the image and the text information in the document in the electronic data and extracts the key words, generates the folder name, and compares the key words with the folder name stored in the data storage module for duplicate checking, and adds the keyword information to regenerate the folder name when duplication occurs; When performing text recognition on a standard image, the recognized text is generated into a document corresponding to the format and style of the shooting, that is, the effect of converting the image to a document is achieved, and the content can be easily found.

[0065] The steps to extract keywords and generate folder names are as follows: Count the acquired data and sort the keywords according to the number of times they appear in the data; Keyword screening: remove keywords except nouns and keep the top T groups of keywords, where T is 5-8; Generate folder name by combining the task name or project name corresponding to the file with the first V items in the T group of keywords, where V<T and V is 3-5.

[0066] This step realizes the extraction of keywords, which is mainly used to solve the problems of duplication of names and inability to accurately find files when there are many files. After the text is extracted in this step, subsequent personnel can directly find and locate the required content through the search function after opening the corresponding file.

[0067] After the data is processed, the data is stored, and the data storage is based on a data storage module.

[0068] The data storage module arranges the input standard image and document data in the order of input, aggregates them into folders, and names the folders with the generated folder names.

[0069] File Reading Before reading the files, the user needs to identify the user's identity to determine whether he is the staff member who reads the files. Since the contents of the files are confidential, only those with relevant permissions are qualified to read the files. Identifying user identities is based on the identity authentication module. Common methods of the identity authentication module include face recognition, fingerprint recognition, voiceprint recognition, vein recognition, etc. through mobile devices.

[0070] After identity recognition, enter the search module, which is connected to the data storage module. Directly search for keyword time archives. After entering the folder where the archive is located, click on the file to directly retrieve the required information. It is convenient to use.

[0071] Example 2: Figure 2 As shown, a comprehensive archive management method comprises the following steps: Determine whether the data source is electronic data or a paper document. If it is a paper document, use a camera to capture and input it; Obtain the parameter data of the captured image, calculate the image resolution PPI and the number of paper document corners S in the recorded image, compare the image resolution PPI with the set minimum resolution PPIb, and compare the number of paper document corners S with the standard number of corners Sb, determine whether the image resolution and integrity meet the standards, and if they do not meet the standards, feedback to the data entry module to re-enter the data; Obtain the analyzed and adjusted image, and sequentially perform angle adjustment processing, stain removal processing, and text filling processing on the image to obtain a standard image; Obtain a standard image, perform text recognition on the standard image, obtain text information stored in the image and text information in the document in the electronic data and extract key words, generate a folder name, compare the key words with the folder name stored in the data storage module for duplicate checking, and add keyword information to regenerate the folder name when duplication occurs; Arrange the entered standard images and document data in the order in which they were entered, summarize them into folders, and name the folders with the generated folder names.

[0072] The above formulas are all dimensionless and numerical calculations. The formula is a formula obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formula are set by technicians in this field according to actual conditions. The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product.

Claims

1. A comprehensive archive management system, characterized in that: include: The data entry module determines whether the data source is electronic data or paper documents. If it is a paper document, a camera is used to shoot and enter the data. The analysis and adjustment module obtains the parameter data of the captured image, calculates the image resolution PPI and the number of paper document corners S in the captured image, compares the image resolution PPI with the set minimum resolution PPIb, and compares the number of paper document corners S with the standard number of corners Sb, determines whether the image resolution and integrity meet the standards, and feeds back to the data entry module to re-enter the data if they do not meet the standards; The image processing module obtains the analyzed and adjusted image, and sequentially performs angle adjustment, stain removal and text filling on the image to obtain a standard image; The data processing module obtains the standard image and performs text recognition on the standard image, obtains the text information stored in the image and the text information in the document in the electronic data and extracts the key words, generates the folder name, and compares the key words with the folder name stored in the data storage module for duplicate checking, and adds the keyword information to regenerate the folder name when duplication occurs; The data storage module arranges the entered standard images and document data in the order in which they were entered, aggregates them into folders, and names the folders with the generated folder names.

2. A comprehensive archive management system according to claim 1, characterized in that: The image data that meets the standard is PPI≥PPIb, and S=Sb, where the calculation formula of the image resolution is: ； Where PPI is the number of pixels per inch in the image, X is the number of pixels in the length of the photo, Y is the number of pixels in the width of the photo, and Z is the diagonal length of the photo.

3. A comprehensive archive management system according to claim 2, characterized in that: The steps to adjust the angle of the image are as follows: Calculate the image tilt angle; The calculation formula for the image tilt angle is as follows: ； Among them, θ is the tilt angle of the image, (X1, Y1) (X2, Y2) are the coordinates of the two corners at the bottom of the paper document; Determine whether the image tilt angle meets the set standard; Compare the calculated θ with the set threshold interval [0-δ, 0+δ], where δ is [0, 8], When θ is within the set threshold range, the image tilt angle meets the standard and no angle adjustment is required; When θ is outside the set threshold range, the image tilt angle does not meet the standard, and the image angle is adjusted; Adjust the angle of images that do not meet the set threshold range.

4. A comprehensive archive management system according to claim 3, characterized in that: The calculation formula for adjusting the angle of images that do not meet the set threshold range is as follows: ； Where d(x, y) is the output image, f(x, y) is a two-dimensional function representing the input image with integer pixel coordinates, processed using square root operations, and its convolution kernel is as follows; 。 5. A comprehensive archive management system according to claim 4, characterized in that: The steps of the stain removal process are as follows: obtain text color data and stain color data, and calculate the distance between the text color data and the stain color data, determine whether the distance meets the set value range, and remove other color stains that are not within the value range.

6. A comprehensive archive management system according to claim 5, characterized in that: The calculation formula for calculating the difference between the text color data and the defaced color data is as follows: ； In the formula, (R w , G w , B w ) is the RGB value corresponding to the text color, (R s , G s , B s ) is the RGB value corresponding to the contaminated color, and H is the distance between the two colors.

7. A comprehensive archive management system according to claim 6, characterized in that: When judging whether the distance is within the set value range, H is compared with the set value range [0, λ], the value range of λ is [5, 12], and the data whose difference is not within the value range is set as the paper color.

8. A comprehensive archive management system according to claim 7, characterized in that: The steps of text filling processing are as follows: Get the spatial distance of the missing part and calculate the number of missing characters. The calculation formula is: number of missing characters Qssl = (next character position - previous character position) / single character width; Remove the background, extract the missing text features, and predict text A based on the text features m And the possible proportion K of the text m ; For missing parts filling, the score is calculated according to the characters and the possible proportion of the characters, and a filling order is generated according to the order of the scores, and the missing parts are filled in according to the order. The score calculation formula is as follows; ； In the formula, is the Ath predicted for the i-th missing m The possible proportion of words; Semantic judgment: Get the filled text data and the preceding and following data to judge whether the sentence is fluent and score it according to the degree of fluency. If the score is not within the set range, fill in the next group of text according to the filling order until the score meets the set range. The scoring criteria include the word Cy, the sentence semantics Yy, and the correlation between the previous and next content Gd. The scoring formula is as follows: ； In the formula, Pf is the score, C´y is the weight corresponding to the word item, Y´y is the weight corresponding to the semantic item, and G´d is the weight corresponding to the relevance item of the previous and next contents.

9. A comprehensive archive management system according to claim 8, characterized in that: The steps to extract keywords and generate folder names are as follows: Count the acquired data and sort the keywords according to the number of times they appear in the data; Keyword screening: remove keywords other than nouns and keep the top T group keywords; Generate the folder name by combining the task name or project name corresponding to the file with the first V items in the T group of keywords, where V<T.

10. A comprehensive archive management method, using any one of the systems in claims 1 to 9, characterized in that: The steps include: Determine whether the data source is electronic data or a paper document. If it is a paper document, use a camera to capture and input it; Obtain the parameter data of the captured image, calculate the image resolution PPI and the number of paper document corners S in the recorded image, compare the image resolution PPI with the set minimum resolution PPIb, and compare the number of paper document corners S with the standard number of corners Sb, determine whether the image resolution and integrity meet the standards, and if they do not meet the standards, feedback to the data entry module to re-enter the data; Obtain the analyzed and adjusted image, and sequentially perform angle adjustment processing, stain removal processing, and text filling processing on the image to obtain a standard image; Obtain a standard image, perform text recognition on the standard image, obtain text information stored in the image and text information in the document in the electronic data and extract key words, generate a folder name, compare the key words with the folder name stored in the data storage module for duplicate checking, and add keyword information to regenerate the folder name when duplication occurs; Arrange the entered standard images and document data in the order in which they were entered, summarize them into folders, and name the folders with the generated folder names.

Images