Text recognition method and system applied to enterprise image data

Abstract

The present application relates to the technical field of character recognition, in particular to a character recognition method and system applied to enterprise image data. According to the character information features in the key information matching area, the key information degree of the corresponding area is obtained; according to the position distribution of the key information matching area of the enterprise data image and the key information area of each standard image, and the key information degree distribution of the key information matching area, the overall matching degree between the enterprise data image and each standard image is obtained; combined with the key information degree distribution of the key information matching area in the enterprise data image, the standard reference image of the enterprise data image is screened out; according to the text information features of the key information matching area in the relative standard reference image of different enterprise data images, the enterprise data template image is screened out for character recognition. The present application accurately screens out the enterprise data template image, optimizes the recognition template library, and improves the accuracy of character recognition.

Description

Technical Field

[0001] This invention relates to the field of text recognition technology, specifically to a text recognition method and system for enterprise image data. Background Technology

[0002] Text recognition can be used to process documents, forms, etc. containing a large amount of text information. In the existing technology, images taken at a fixed angle are used to recognize text in corporate image materials through OCR recognition technology. However, the existing template databases are mostly statically stored databases, which cannot optimize the recognition results according to the actual use scenario. The recognition accuracy is low for tilted and translated images, resulting in poor text recognition accuracy in corporate image materials. Summary of the Invention

[0003] To address the technical problem that existing template databases are mostly statically stored, resulting in poor accuracy in text recognition for tilted or translated images, the present invention aims to provide a text recognition method and system for enterprise image data. The specific technical solution adopted is as follows:

[0004] This invention proposes a text recognition method for enterprise image data, the method comprising:

[0005] Acquire multiple enterprise data images and corresponding standard images containing key information areas from different angles;

[0006] For any enterprise data image, obtain the key information matching area of ​​the enterprise data image relative to the key information area in the standard image; based on the text information features in the key information matching area, obtain the key information degree of the key information matching area; based on the positional distribution of the key information matching area of ​​the enterprise data image and the key information area of ​​each standard image, as well as the key information degree distribution of the key information matching area, obtain the overall matching degree between the enterprise data image and each standard image.

[0007] Based on the distribution of key information degree in the key information matching area in the enterprise data image, and the overall matching degree between the enterprise data image and different standard images, the selection degree between the enterprise data image and different standard images is obtained, and the standard reference image of the enterprise data image is selected.

[0008] Based on the text information features of key information matching areas in the relative standard reference images of different enterprise data images, enterprise data template images are selected for text recognition.

[0009] Furthermore, the method for obtaining the degree of key information includes:

[0010] Obtain fixed descriptive words for each key information region in the standard image; if a fixed descriptive word exists in the corresponding key information matching region in the enterprise data image, set the key information level in the key information matching region to a preset first level value; otherwise, set it to a preset second level value, wherein the preset first level value is greater than the preset second level value.

[0011] Furthermore, the method for obtaining the overall matching degree includes:

[0012] Based on the distribution of key information degree in the key information matching area of ​​the enterprise data image, the initial matching degree between the enterprise data image and each standard image is obtained;

[0013] Based on the location distribution of the key information matching area of ​​the enterprise data image and the key information area of ​​each standard image, as well as the distribution of the key information degree of the key information matching area, the degree of deviation between the enterprise data image and each standard image is obtained.

[0014] The ratio of the initial matching degree to the deviation degree is obtained and normalized to represent the overall matching degree between the enterprise data image and each standard image.

[0015] Furthermore, the method for obtaining the degree of deviation includes:

[0016] The key information matching area in the statistical enterprise data image where the key information level value is the preset first level value is used as the target area;

[0017] Using the image center as the origin of the coordinate axis, the mean relative distance between the fixed descriptive word segmentation centers of different target areas and the corresponding key information areas in each standard image is obtained and normalized to represent the degree of deviation between the enterprise data image and each standard image.

[0018] Furthermore, the method for obtaining the initial matching degree includes:

[0019] The number of valid matching regions in enterprise data images when the key information level value is the preset first level value;

[0020] The ratio of the number of valid matching regions to the total number of matching regions is used as the initial degree of matching between the enterprise data image and each standard image.

[0021] Furthermore, the method for obtaining the degree of selection includes:

[0022] Based on the distribution of key information levels in the key information matching areas of enterprise data images, the logical correlation degree of the regions is obtained;

[0023] The overall matching degree between the enterprise data image and each standard image is obtained, as well as the sum of the regional logical correlation degree when the key information degree is a preset second degree value. Then, normalization mapping is performed to determine the selection degree of the enterprise data image relative to each standard image.

[0024] Furthermore, the method for obtaining the logical correlation degree of the regions includes:

[0025] If a key information matching area in the enterprise data image has a key information level of a preset second level value and a key information matching area with a key information level of a preset first level value has a preset logical relationship, the key information matching area with a key information level of a preset second level value is taken as the logical relationship area.

[0026] The ratio of the number of logically related regions to the number of key information matching regions when the key information level is a preset second level value is used as the region logical correlation degree in the enterprise data image when the key information level is a preset second level value.

[0027] Furthermore, the method for acquiring the standard reference image includes:

[0028] Select the maximum value of the selected degree among all standard images from the enterprise data imagery, and use the corresponding standard imagery as the standard reference image.

[0029] Furthermore, the method for obtaining the enterprise data template image includes:

[0030] Based on the OCR recognition algorithm, text data in the matching area of ​​key information corresponding to enterprise data images is obtained, and the text validity of the text data is obtained.

[0031] If the text validity of any key information matching area is less than the preset validity threshold, the corresponding location area will be regarded as an abnormal area.

[0032] A negative correlation mapping is performed on the number of abnormal regions to obtain the product of the average text validity of different abnormal regions and the negative correlation mapping, and then the mapping is normalized to serve as a template possibility for enterprise data images.

[0033] If the template probability of any enterprise data image is greater than the preset probability threshold, the corresponding enterprise data image will be used as the enterprise data template image.

[0034] The present invention also proposes a text recognition system for enterprise image data, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of any one of the text recognition methods for enterprise image data.

[0035] The present invention has the following beneficial effects:

[0036] This invention, for any enterprise data image, obtains the key information matching region of the enterprise data image relative to the key information area in the standard image, accurately locating the position corresponding to each key area in the standard image; based on the text information features within the key information matching region, it obtains the key information degree of the key information matching region, quantifying the performance characteristics of the text information in the key information matching region; based on the positional distribution of the key information matching region of the enterprise data image and the key information region of each standard image, as well as the key information degree distribution of the key information matching region, it obtains the overall matching degree between the enterprise data image and each standard image, more comprehensively quantifying the matching degree between images; based on the key information degree distribution of the key information matching region in the enterprise data image, and the overall matching degree between the enterprise data image and different standard images, it obtains the selection degree between the enterprise data image and different standard images, selecting standard reference images for the enterprise data image, and selecting the most similar standard reference image, which is beneficial for matching analysis of enterprise data images; based on the text information features of the key information matching region of different enterprise data images relative to the standard reference image, it selects enterprise data template images for text recognition. This invention improves the accuracy of text recognition by accurately selecting enterprise data template images and optimizing the recognition template library. Attached Figure Description

[0037] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 A flowchart illustrating a text recognition method for enterprise image data provided in one embodiment of the present invention;

[0039] Figure 2 This is a flowchart illustrating a method for obtaining the overall matching degree according to an embodiment of the present invention. Detailed Implementation

[0040] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a text recognition method and system for enterprise image data proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0041] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0042] The following description, in conjunction with the accompanying drawings, details a specific solution for a text recognition method and system for enterprise image data provided by the present invention.

[0043] Please see Figure 1 The diagram illustrates a flowchart of a text recognition method for enterprise image data according to an embodiment of the present invention, specifically including:

[0044] Step S1: Acquire multiple enterprise data images and corresponding standard enterprise images containing key information areas from different angles.

[0045] In the embodiments of the present invention, considering the possibility of tilting or translation of the image to ensure the accuracy of text recognition, the company data image and the corresponding standard images at different angles are matched and analyzed. First, the format and file size that meet the preset conditions are processed, and the image size is adjusted to be the same as the size of each image template in the database to obtain the company data image.

[0046] It should be noted that, in order to facilitate subsequent data processing, existing noise reduction algorithms are used to preprocess the image, such as deblurring and grayscale conversion, and the size is adjusted to be the same as that of each reference template in the database for subsequent processing; the specific methods are well known to those skilled in the art and will not be described in detail here.

[0047] Based on this, the implementers, using their relevant professional knowledge and experience, obtained a predefined standard image of the enterprise data that was free from angular tilt, and marked the key information areas of the standard image; they then rotated the standard image to obtain enterprise standard images containing the key information areas at different angles.

[0048] It should be noted that, in the embodiments of the present invention, the interval between angles can be set according to specific circumstances, such as setting the interval to 5°; the key information area is obtained by the implementer by framing each key information according to relevant professional knowledge.

[0049] Step S2: For any enterprise data image, obtain the key information matching area of ​​the enterprise data image relative to the key information area in the standard image; based on the text information features in the key information matching area, obtain the key information degree of the key information matching area; based on the positional distribution of the key information matching area of ​​the enterprise data image and the key information area of ​​each standard image, as well as the key information degree distribution of the key information matching area, obtain the overall matching degree between the enterprise data image and each standard image.

[0050] To accurately locate the specific position of the text to be extracted for subsequent text recognition, for any enterprise data image, a key information matching region is obtained relative to the key information region in the standard image. It should be noted that key points and feature descriptors are extracted from the enterprise data image and the standard image using feature extraction algorithms SIFT or ORB for matching and alignment, obtaining the corresponding position region on the enterprise data image corresponding to the key information region in the standard image, which is used as the key information matching region. The specific methods are well known to those skilled in the art and will not be elaborated here.

[0051] The key information area reflects the text features displayed. Based on the distribution of key information in the enterprise data image, the degree of key information in the enterprise data image is obtained. Based on the text information features in the key information matching area, the degree of key information in the key information matching area is obtained.

[0052] Preferably, in one embodiment of the present invention, the method for obtaining the level of key information includes:

[0053] Obtain fixed descriptive words for each key information region in the standard image; if fixed descriptive words exist in the corresponding key information matching region in the enterprise data image, set the key information level in the key information matching region to a preset first level value; otherwise, set it to a preset second level value, where the preset first level value is greater than the preset second level value.

[0054] It should be noted that, in the embodiments of the present invention, the more fixed descriptive words exist, the greater the degree of key information. Therefore, the preset first degree value is greater than the preset second degree value, and can be set according to specific circumstances. For example, the preset first degree value is a positive integer 1, and the preset second degree value is 0.

[0055] It should be noted that, in the embodiments of the present invention, the fixed descriptive words can be obtained in advance by the implementers based on relevant professional knowledge. These refer to the unchanging text in the standard image of enterprise data. Each key information area has its own fixed descriptive words, such as the enterprise name, organization code, registered address, and other descriptive words in the business license.

[0056] It should be noted that in the embodiments of the present invention, the text in the key information area is obtained by OCR recognition. The specific means are well known to those skilled in the art and will not be described in detail here.

[0057] The key information level reflects the degree to which key information regions contain fixed descriptive words; the higher the key information level, the more fixed descriptive words are contained. Based on the distribution of key information levels in different key information regions on the enterprise data image, the initial matching degree between the enterprise data image and each standard image is obtained.

[0058] Preferably, in one embodiment of the present invention, the method for obtaining the overall matching degree is described in [reference needed]. Figure 2 It illustrates a flowchart of a method for obtaining the overall matching degree, including:

[0059] Step S201: Based on the key information degree distribution in the key information matching area of ​​the enterprise data image, obtain the initial matching degree between the enterprise data image and each standard image.

[0060] Preferably, the key information level reflects the state of fixed descriptive words within the region. The higher the key information level, the more fixed descriptive words are contained, and the better it matches the standard image. In one embodiment of the present invention, the method for obtaining the initial matching level includes:

[0061] The number of valid matching regions in enterprise data images when the key information level value is the preset first level value;

[0062] The ratio of the number of valid matching regions to the total number of matching regions is used as the initial degree of matching between the enterprise data image and each standard image.

[0063] Step S202: Based on the location distribution of the key information matching area of ​​the enterprise data image and the key information area of ​​each standard image, as well as the key information degree distribution of the key information matching area, obtain the degree of deviation between the enterprise data image and each standard image.

[0064] Preferably, the positional distribution of the key information matching area of ​​the enterprise data image and the key information area of ​​the labeled image reflects the correlation between the positions of the corresponding key information matching areas. The closer the positions, the smaller the deviation. In one embodiment of the present invention, the method for obtaining the degree of deviation includes:

[0065] The key information matching area in the statistical enterprise data image where the key information level value is the preset first level value is used as the target area;

[0066] Using the image center as the origin of the coordinate axis, the mean relative distance between the fixed descriptive word segmentation centers of different target areas and the corresponding key information areas in each standard image is obtained and normalized to represent the degree of deviation between the enterprise data image and each standard image.

[0067] It should be noted that in the embodiments of the present invention, the relative distance is calculated by Euclidean distance or Manhattan distance; and the normalization process is performed by linear normalization or normalization function. The specific means are well known to those skilled in the art and will not be described in detail here.

[0068] Step S203: Obtain the first sum of the deviation degree and the preset adjustment coefficient, obtain the ratio of the initial matching degree to the first sum, and normalize it as the overall matching degree between the enterprise data image and each standard image.

[0069] It should be noted that, in the embodiments of the present invention, in order to avoid the deviation degree being 0 and the formula being meaningless when calculating the ratio, a first sum of the deviation degree and the preset adjustment coefficient is obtained. The preset adjustment coefficient can be set according to specific circumstances to avoid the formula ratio being 0, such as setting it to 0.01. This is not limited or elaborated here.

[0070] Step S3: Based on the distribution of key information degree in the key information matching area in the enterprise data image, and the overall matching degree between the enterprise data image and different standard images, obtain the selection degree between the enterprise data image and different standard images, and filter out the standard reference image of the enterprise data image.

[0071] The overall matching degree assesses the similarity between enterprise data images and standard images from a global perspective. The greater the overall matching degree, the better the enterprise data images and standard images match. The key information degree reflects the distribution of matching confidence in each key information region when matching images. The greater the key information degree, the greater the matching confidence. Therefore, based on the key information degree distribution in the key information matching regions of enterprise data images and the overall matching degree between enterprise data images and different standard images, the selectivity between enterprise data images and different standard images can be obtained.

[0072] Preferably, in one embodiment of the present invention, the method for obtaining the degree of selection includes:

[0073] Based on the distribution of key information levels in the key information matching areas of enterprise data images, the logical correlation degree of the regions is obtained;

[0074] Preferably, in one embodiment of the present invention, the method for obtaining the region logical correlation degree includes:

[0075] If a key information matching area in the enterprise data image has a key information level of a preset second level value and a key information matching area with a key information level of a preset first level value has a preset logical relationship, the key information matching area with a key information level of a preset second level value is taken as the logical relationship area.

[0076] It should be noted that, in the embodiments of the present invention, the preset logical association can be predefined by the implementer based on relevant business knowledge and domain logic, such as the composition of the organization code in the business license being related to the administrative division code of the registration authority, and the ID number of the legal person's ID card being related to the date of birth.

[0077] The ratio of the number of logically related regions to the number of key information matching regions when the key information level is a preset second level value is used as the region logical correlation degree in the enterprise data image when the key information level is a preset second level value.

[0078] The overall matching degree between the enterprise data image and each standard image is obtained, as well as the sum of the regional logical correlation degree when the key information degree is a preset second degree value. Then, normalization mapping is performed to determine the selection degree of the enterprise data image relative to each standard image.

[0079] Therefore, the greater the selection range, the stronger the matching relationship with the enterprise data images, which is more helpful for subsequent screening and the selection of standard reference images for enterprise data images.

[0080] Preferably, in one embodiment of the present invention, the method for acquiring the standard reference image includes:

[0081] Select the maximum value of the selected degree among all standard images from the enterprise data imagery, and use the corresponding standard imagery as the standard reference image.

[0082] Step S4: Based on the text information features of the key information matching area in the relative standard reference image of different enterprise data images, screen out the enterprise data template image for text recognition.

[0083] Textual information features help analyze the visual quality, structure, and credibility of text in the key information matching areas of enterprise document images, and analyze the likelihood of involvement. Based on the textual information features of key information matching areas in different enterprise document images relative to standard reference images, enterprise document template images are selected for text recognition.

[0084] Preferably, in one embodiment of the present invention, the method for obtaining the enterprise data template image includes:

[0085] Based on the OCR recognition algorithm, text data in the matching area of ​​key information corresponding to enterprise data images is obtained, and the text validity of the text data is obtained.

[0086] If the text validity of any key information matching area is less than the preset validity threshold, the corresponding location area will be regarded as an abnormal area.

[0087] A negative correlation mapping is performed on the number of abnormal regions to obtain the product of the average text validity of different abnormal regions and the negative correlation mapping, and then the mapping is normalized to serve as a template possibility for enterprise data images.

[0088] If the template probability of any enterprise data image is greater than the preset probability threshold, the corresponding enterprise data image will be used as the enterprise data template image.

[0089] It should be noted that, in one embodiment of the present invention, the preset probability threshold is 0.6; in other embodiments of the present invention, the size of the preset probability threshold can be set according to specific circumstances, and will not be limited or elaborated here.

[0090] It should be noted that, in the embodiments of the present invention, the validity of text data in each key information matching area is obtained based on the text verification mechanism. The validity value is 0-1. When the text validity is 1, the text data is more normal. When the text validity is less than 1, the text data in the area is more abnormal. Therefore, the preset validity threshold is 1. The OCR recognition algorithm and text verification mechanism are technical means well known to those skilled in the art, and will not be described in detail here.

[0091] It should be noted that, in the embodiments of the present invention, negative correlation mapping is performed by taking the reciprocal or by using an exponential function with the natural constant as the base. When taking the reciprocal, in order to avoid the formula being meaningless with a denominator of 0, an artificially set threshold, such as 0.01, is added to the denominator. The specific means are well known to those skilled in the art and will not be described in detail here.

[0092] Based on this, after obtaining the enterprise data template image, it is added to the template library to enrich the template library, improve the correction speed and accuracy of subsequent similar pose images, and perform text recognition on the image.

[0093] In summary, this invention obtains the key information degree of the key information matching area based on the text information features within the key information matching area; it obtains the overall matching degree between the enterprise data image and each standard image based on the positional distribution of the key information matching area and the key information area of ​​each standard image, as well as the key information degree distribution of the key information matching area; it selects standard reference images for the enterprise data images by combining the key information degree distribution of the key information matching areas in the enterprise data images; and it selects enterprise data template images for text recognition based on the text information features of the key information matching areas of different enterprise data images relative to the standard reference images. This invention improves the accuracy of text recognition by accurately selecting enterprise data template images and optimizing the recognition template library.

[0094] The present invention also proposes a text recognition system for enterprise image data, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the steps of a text recognition method for enterprise image data.

[0095] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0096] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A text recognition method applied to enterprise image data, characterized in that, The method includes: Acquire multiple enterprise data images and corresponding standard images containing key information areas from different angles; For any enterprise data image, obtain the key information matching area of ​​the enterprise data image relative to the key information area in the standard image; based on the text information features in the key information matching area, obtain the key information degree of the key information matching area; based on the positional distribution of the key information matching area of ​​the enterprise data image and the key information area of ​​each standard image, as well as the key information degree distribution of the key information matching area, obtain the overall matching degree between the enterprise data image and each standard image. Based on the distribution of key information degree in the key information matching area in the enterprise data image, and the overall matching degree between the enterprise data image and different standard images, the selection degree between the enterprise data image and different standard images is obtained, and the standard reference image of the enterprise data image is selected. Based on the text information features of key information matching areas in the relative standard reference images of different enterprise data images, enterprise data template images are selected for text recognition. The method for obtaining the degree of key information includes: obtaining fixed descriptive words for each key information region in the standard image; if a fixed descriptive word exists in the corresponding key information matching region in the enterprise data image, setting the degree of key information in the key information matching region to a preset first degree value; otherwise, setting it to a preset second degree value, wherein the preset first degree value is greater than the preset second degree value; The method for obtaining the selection degree includes: obtaining the region logical correlation degree based on the key information degree distribution of the key information matching region in the enterprise data image; obtaining the overall matching degree between the enterprise data image and each standard image, and the sum of the region logical correlation degree when the key information degree is a preset second degree value, and performing normalization mapping as the selection degree of the enterprise data image relative to each standard image; The method for obtaining the overall matching degree includes: obtaining the initial matching degree between the enterprise data image and each standard image based on the key information degree distribution in the key information matching area of ​​the enterprise data image; obtaining the deviation degree between the enterprise data image and each standard image based on the positional distribution of the key information area in the key information matching area of ​​the enterprise data image and the key information area in each standard image, as well as the key information degree distribution in the key information matching area; obtaining the ratio of the initial matching degree to the deviation degree, and normalizing it, as the overall matching degree between the enterprise data image and each standard image; The location distribution of the key information matching area between the enterprise data image and the key information area of ​​the standard image reflects the correlation between the locations of the corresponding key information matching areas. The closer the locations are, the smaller the deviation.

2. The text recognition method for enterprise image data according to claim 1, characterized in that, The method for obtaining the degree of deviation includes: The key information matching area in the statistical enterprise data image where the key information level value is the preset first level value is used as the target area; Using the image center as the origin of the coordinate axis, the mean relative distance between the fixed descriptive word segmentation centers of different target areas and the corresponding key information areas in each standard image is obtained and normalized to represent the degree of deviation between the enterprise data image and each standard image.

3. The text recognition method for enterprise image data according to claim 1, characterized in that, The method for obtaining the initial matching degree includes: The number of valid matching regions in enterprise data images when the key information level value is the preset first level value; The ratio of the number of valid matching regions to the total number of matching regions is used as the initial degree of matching between the enterprise data image and each standard image.

4. The text recognition method for enterprise image data according to claim 1, characterized in that, The method for obtaining the logical correlation degree of the region includes: If a key information matching area in the enterprise data image has a key information level of a preset second level value and a key information matching area with a key information level of a preset first level value has a preset logical relationship, the key information matching area with a key information level of a preset second level value is taken as the logical relationship area. The ratio of the number of logically related regions to the number of key information matching regions when the key information level is a preset second level value is used as the region logical correlation degree in the enterprise data image when the key information level is a preset second level value.

5. The text recognition method for enterprise image data according to claim 1, characterized in that, The method for acquiring the standard reference image includes: Select the maximum value of the selected level from all standard images in the enterprise data imagery, and use the corresponding standard image as the standard reference image.

6. The text recognition method for enterprise image data according to claim 1, characterized in that, The method for obtaining the enterprise data template image includes: Based on the OCR recognition algorithm, text data in the matching area of ​​key information corresponding to enterprise data images is obtained, and the text validity of the text data is obtained. If the text validity of any key information matching area is less than the preset validity threshold, the corresponding location area will be regarded as an abnormal area. A negative correlation mapping is performed on the number of abnormal regions to obtain the product of the average text validity of different abnormal regions and the negative correlation mapping, and then the mapping is normalized to serve as a template possibility for enterprise data images. If the template probability of any enterprise data image is greater than the preset probability threshold, the corresponding enterprise data image will be used as the enterprise data template image.

7. A text recognition system for enterprise image data, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the text recognition method for enterprise image data as described in any one of claims 1 to 6.

Images