Image generation method, system and storage medium based on accurate matching of text features

Abstract

The present invention discloses an image generation method, system and storage medium based on accurate matching of text features, which relates to the field of image generation technology, including: establishing a text database; obtaining input text information; performing text splitting on the input text information according to a text splitting algorithm to obtain a plurality of text feature sets to be matched; respectively calculating the matching degree of each text feature set to be matched with the text database; screening out the text feature set to be matched corresponding to the maximum matching degree; obtaining an image feature group; and combining a plurality of image features in the image feature group to generate an image. The advantages of the present invention are: it can effectively ensure the overall matching of the input text information, thereby ensuring the high-precision matching between the produced image and the text features, and the feature information with mutations in the image greatly improves the image generation quality.

Description

Technical Field

[0001] The present invention relates to the field of image generation technology, and in particular to an image generation method, system and storage medium based on precise matching of text features. Background Art

[0002] The text-to-image system is an AI-based technology that processes and understands natural language text to generate image content corresponding to the text description. This system combines natural language processing (NLP) and computer vision (CV) technologies to simulate human creativity and achieve automatic image generation.

[0003] In existing text-to-image systems, the analysis depth of the input text information is insufficient, it is difficult to identify and calculate the intrinsic correlation of the input text information, and it is difficult to fully understand the overall meaning of the input text information. When generating images, it is extremely easy to generate sudden changes in feature information, causing the quality of the generated image to be distorted, affecting the quality of image generation. Summary of the invention

[0004] In order to solve the above technical problems, a method for image generation based on precise matching of text features is provided. This technical solution solves the problems in the above-mentioned existing text-generated image system, such as insufficient analysis depth of input text information, difficulty in identifying and calculating the intrinsic correlation of the input text information, difficulty in fully understanding the overall meaning of the input text information, and extremely easy generation of mutant feature information when generating images, resulting in distortion of the generated image quality and affecting the image generation quality.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0006] A method for generating an image based on accurate matching of text features, comprising:

[0007] Establishing a text database, wherein the text database includes a plurality of text feature classes, and each text feature class includes a plurality of text features of the same type;

[0008] Get the input text information;

[0009] The input text information is split according to the text splitting algorithm to obtain a plurality of text feature sets to be matched;

[0010] According to the matching algorithm, the matching degree between each text feature set to be matched and the text database is calculated respectively;

[0011] Filter out a to-be-matched text feature set corresponding to the maximum matching degree, and use the to-be-matched text feature set as an image generation feature group;

[0012] Based on the image feature corresponding to each text feature in the image generating feature group, an image feature group is obtained;

[0013] An image is generated by combining several image features in an image feature group.

[0014] Preferably, the step of performing text splitting on the input text information according to a text splitting algorithm to obtain a plurality of text feature sets to be matched specifically includes:

[0015] Get the total number of words M of the input text information;

[0016] Sequentially traverse the positive integers in the range of 0-M as the number of text splits n;

[0017] Generate text splitting restriction conditions based on the total number of words M of the input text information and the number of text splits n;

[0018] Under the constraints of text splitting conditions, generate several text splitting arrays;

[0019] According to the numbers in the text split array, extract the text features to be matched with the corresponding number of words from the input text information in turn;

[0020] All the to-be-matched text features are grouped into a to-be-matched text feature set;

[0021] The expression of the text splitting restriction condition is:

[0022]

[0023] In the formula, z i is the number of characters in the i-th text feature to be matched, and N+ is a set of positive integers.

[0024] Preferably, the step of calculating the matching degree between each to-be-matched text feature set and the text database according to the matching degree algorithm specifically includes:

[0025] Calculate the overlap between each text feature set to be matched and the text database;

[0026] Calculate the feature relevance of the text feature set to be matched, and obtain the feature relevance of the text feature set to be matched;

[0027] Based on the feature relevance of the to-be-matched text feature set and the overlap between the to-be-matched text feature set and the text database, the matching degree between the to-be-matched text feature set and the text database is calculated using a matching degree calculation formula;

[0028] The matching degree calculation formula is specifically:

[0029] P=C h ×G l

[0030] In the formula, P is the matching degree between the feature set of the text to be matched and the text database, C h G is the overlap between the feature set of the text to be matched and the text database, l is the feature correlation of the text feature set to be matched.

[0031] Preferably, the calculation of the overlap between each to-be-matched text feature set and the text database specifically includes:

[0032] Traverse all text feature classes to determine whether the to-be-matched text feature in the to-be-matched text feature set is contained in the text feature class. If so, record the to-be-matched text feature as a coincident text feature; if not, record the to-be-matched text feature as a non-coincidence text feature.

[0033] Get the word count of all overlapping text features;

[0034] Based on the number of characters in the overlapping text features and the total number of characters in the input text information, the overlap between the to-be-matched text feature set and the text database is calculated using an overlap calculation formula.

[0035] Preferably, the calculation formula for the overlap degree is:

[0036]

[0037] Where N j is the number of characters in the jth overlapping text feature, and m is the total number of overlapping text features.

[0038] Preferably, the step of calculating the feature correlation of the to-be-matched text feature set to obtain the feature correlation of the to-be-matched text feature set specifically includes:

[0039] Sort all the overlapping text features according to the order in which they appear in the input text information;

[0040] Based on the text feature class described in each overlapping text feature, respectively analyzing the correlation between each overlapping text feature and its adjacent previous overlapping text feature and next overlapping text feature;

[0041] The maximum value of the correlation between the coincident text feature and its adjacent previous coincident text feature and the next coincident text feature is taken as the correlation of the coincident text feature;

[0042] According to the comprehensive formula of feature relevance, the feature relevance of the set of text features to be matched is calculated based on each overlapping text feature.

[0043] Preferably, the feature correlation comprehensive formula is specifically:

[0044]

[0045] In the formula, g j is the relevance of the jth overlapping text feature.

[0046] Furthermore, a system for generating an image based on accurate matching of text features is proposed, which is used to implement the above-mentioned method for generating an image based on accurate matching of text features, including:

[0047] A storage module, wherein the storage module is used to store a text database;

[0048] A text input module, wherein the text input module is used to obtain input text information;

[0049] A feature splitting module, the feature splitting module is electrically connected to the text input module, and the feature splitting module is used to perform text splitting on the input text information according to a text splitting algorithm to obtain a plurality of text feature sets to be matched;

[0050] A matching calculation module, the matching calculation module is electrically connected to the storage module and the feature separation module, and the matching calculation module is used to calculate the matching degree of each to-be-matched text feature set and the text database according to a matching degree algorithm;

[0051] A screening module, the screening module is electrically connected to the matching calculation module, and the screening module is used to screen out a to-be-matched text feature set corresponding to a maximum matching degree, and use the to-be-matched text feature set as an image generation feature group;

[0052] An image generation module is electrically connected to the screening module, and is used to obtain an image feature group based on the image feature corresponding to each text feature in the image feature group, and to combine several image features in the image feature group to generate an image.

[0053] Optionally, the feature splitting module includes:

[0054] A splitting condition unit, wherein the splitting condition unit is used to generate text splitting restriction conditions;

[0055] A text splitting unit, wherein the text splitting unit is used to generate a plurality of text splitting arrays under the text splitting restriction condition and extract the to-be-matched text features of the corresponding number of words from the input text information in turn according to the numbers in the text splitting arrays;

[0056] The matching calculation module includes:

[0057] An overlap calculation unit, the overlap calculation unit is used to calculate the overlap degree between each to-be-matched text feature set and the text database;

[0058] An association calculation unit, the association calculation unit is used to calculate the feature association degree of the to-be-matched text feature set, and obtain the feature association degree of the to-be-matched text feature set;

[0059] The matching calculation unit is used to calculate the matching degree between the to-be-matched text feature set and the text database through a matching degree calculation formula based on the feature association degree of the to-be-matched text feature set and the overlap degree between the to-be-matched text feature set and the text database.

[0060] Furthermore, a computer-readable storage medium is proposed, on which a computer-readable program is stored. When the computer-readable program is called, the image generation method based on precise matching of text features as described above is executed.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] The present invention proposes an image generation scheme based on precise matching of text features. The scheme greatly improves the analysis depth of the input text information by carefully splitting the input text information into a plurality of text feature sets to be matched, and combines the overlap between the text feature sets to be matched and the text database and the correlation between the text features in the text feature sets to be matched. The scheme can effectively ensure the overall matching of the input text information, thereby ensuring a high-precision match between the produced image and the text features, and the feature information with mutations in the image, which greatly improves the image generation quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a flow chart of the image generation method based on accurate matching of text features proposed by the present invention;

[0064] Figure 2 A flow chart of a method for obtaining a set of text features to be matched according to the present invention;

[0065] Figure 3 A flow chart of a method for calculating the matching degree between a text feature set to be matched and a text database according to the present invention;

[0066] Figure 4 A flow chart of a method for calculating the degree of overlap between a text feature set to be matched and a text database according to the present invention;

[0067] Figure 5 The present invention is a flow chart of a method for calculating the feature relevance of a set of text features to be matched. DETAILED DESCRIPTION

[0068] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.

[0069] Reference Figure 1 As shown, an image generation method based on accurate matching of text features includes:

[0070] Establishing a text database, wherein the text database includes a plurality of text feature classes, and each text feature class includes a plurality of text features of the same type;

[0071] Get the input text information;

[0072] The input text information is split according to the text splitting algorithm to obtain a plurality of text feature sets to be matched;

[0073] According to the matching algorithm, the matching degree between each text feature set to be matched and the text database is calculated respectively;

[0074] Filter out a to-be-matched text feature set corresponding to the maximum matching degree, and use the to-be-matched text feature set as an image generation feature group;

[0075] Based on the image feature corresponding to each text feature in the image generating feature group, an image feature group is obtained;

[0076] An image is generated by combining several image features in an image feature group.

[0077] This solution greatly improves the analysis depth of the input text information by carefully splitting the input text information into several text feature sets to be matched, and combines the dual recognition judgment of the overlap between the text feature sets to be matched and the text database and the correlation between the text features in the text feature sets to be matched, which can effectively ensure the overall matching of the input text information.

[0078] Reference Figure 2 As shown, the text information input is split according to the text splitting algorithm to obtain a plurality of text feature sets to be matched, specifically including:

[0079] Get the total number of words M of the input text information;

[0080] Sequentially traverse the positive integers in the range of 0-M as the number of text splits n;

[0081] Generate text splitting restriction conditions based on the total number of words M of the input text information and the number of text splits n;

[0082] Under the constraints of text splitting conditions, generate several text splitting arrays;

[0083] Split the numbers in the array according to the text, and sequentially extract the text feature to be matched with the corresponding number of characters from the input text information;

[0084] Combine all the text features to be matched into a set of text features to be matched;

[0085] Among them, the expression of the text splitting limit condition is:

[0086]

[0087] In the formula, z i is the number of characters of the i-th text feature to be matched, and N+ is the set of positive integers.

[0088] Illustrate the above text splitting process with a specific embodiment:

[0089] In one embodiment, the input text information is "Guan Yu", then the total number of characters of the input text information is 2. Sequentially traverse the positive integers in the range of 0-2, which are 1 and 2. Then when n = 1, the text splitting limit condition is:

[0090]

[0091] The generated text split array is {2}, and the set of text features to be matched split out is {Guan Yu};

[0092] When n = 2, the text splitting limit condition is:

[0093]

[0094] The generated text split array is {1, 1}, and the set of text features to be matched split out is {Guan, Yu}.

[0095] Refer to Figure 3 As shown, the specific steps of calculating the matching degree between each set of text features to be matched and the text database according to the matching degree algorithm include:

[0096] Calculate the coincidence degree between each set of text features to be matched and the text database;

[0097] Perform feature correlation degree calculation on the set of text features to be matched to obtain the feature correlation degree of the set of text features to be matched;

[0098] Based on the feature correlation degree of the set of text features to be matched and the coincidence degree between the set of text features to be matched and the text database, calculate the matching degree between the set of text features to be matched and the text database through the matching degree calculation formula;

[0099] The specific matching degree calculation formula is:

[0100] P=C h ×G l

[0101] In the formula, P is the matching degree between the feature set of the text to be matched and the text database, C h G is the overlap between the feature set of the text to be matched and the text database, l is the feature correlation of the text feature set to be matched.

[0102] Reference Figure 4 As shown, the calculation of the overlap between each to-be-matched text feature set and the text database specifically includes:

[0103] Traverse all text feature classes to determine whether the to-be-matched text feature in the to-be-matched text feature set is contained in the text feature class. If so, record the to-be-matched text feature as a coincident text feature; if not, record the to-be-matched text feature as a non-coincidence text feature.

[0104] Get the word count of all overlapping text features;

[0105] Based on the number of characters in the overlapping text features and the total number of characters in the input text information, the overlap between the to-be-matched text feature set and the text database is calculated using an overlap calculation formula.

[0106] The calculation formula of the overlap degree is:

[0107]

[0108] Where N j is the number of characters in the jth overlapping text feature, and m is the total number of overlapping text features.

[0109] Specifically, for the input text information "Guan Yu", the text feature sets to be matched generated are {Guan Yu} and {Guan, Yu}, among which the text to be matched "Guan Yu" is overlapped in the historical figure text feature class, the text to be matched "Guan" is overlapped in the action text feature class, and the text to be matched "Yu" is overlapped in the object text feature class. Then, the overlap degree of the text feature sets to be matched {Guan Yu} and {Guan, Yu} with the text database is 1.

[0110] Reference Figure 5 As shown, the feature correlation degree calculation is performed on the to-be-matched text feature set to obtain the feature correlation degree of the to-be-matched text feature set, and specifically includes:

[0111] Sort all the overlapping text features according to the order in which they appear in the input text information;

[0112] Based on the text feature class described in each overlapping text feature, respectively analyzing the correlation between each overlapping text feature and its adjacent previous overlapping text feature and next overlapping text feature;

[0113] The maximum value of the correlation between the coincident text feature and its adjacent previous coincident text feature and the next coincident text feature is taken as the correlation of the coincident text feature;

[0114] According to the comprehensive formula of feature relevance, the feature relevance of the set of text features to be matched is calculated based on each overlapping text feature.

[0115] The feature correlation comprehensive formula is specifically:

[0116]

[0117] In the formula, g j is the relevance of the jth overlapping text feature.

[0118] Specifically, for the input text information "Guan Yu", the generated text feature sets to be matched are {Guan Yu} and {Guan, Yu}. According to the above feature correlation comprehensive formula, the feature correlation of the text feature set to be matched {Guan Yu} is calculated to be 1. Since there is no effective correlation between the text to be matched "Guan" and the text to be matched "Yu", the correlation of the text feature to be matched "Guan" and the text feature to be matched "Yu" is set to 0.1. Then, according to the above feature correlation comprehensive formula, the feature correlation of the text feature set to be matched {Guan, Yu} is calculated to be 0.1;

[0119] Combined with the overlap between the text feature sets to be matched {Guan Yu} and {Guan, Yu} and the text database, as well as the feature correlation between the text feature sets to be matched {Guan Yu} and {Guan, Yu}, it is calculated that the matching degree of the text feature set to be matched {Guan Yu} and the text database is 1, and the matching degree of the text feature set to be matched {Guan, Yu} and the text database is 0.1. Therefore, it is the text feature set that best matches the text feature set to be matched {Guan Yu} and the input text information. Generating image features according to the text feature set to be matched {Guan Yu} and combining them to generate an image can maximize the degree of fit between the image and the input text.

[0120] It should be noted that the text "Guan Yu" input above is only an exemplary embodiment of the present solution. Inputting other text information and realizing image generation according to the technical features disclosed in the present solution are all different embodiments of the present solution.

[0121] Further, based on the same inventive concept as the above-mentioned image generation method based on precise matching of text features, the present solution also proposes an image generation system based on precise matching of text features, comprising:

[0122] A storage module, wherein the storage module is used to store a text database;

[0123] A text input module, wherein the text input module is used to obtain input text information;

[0124] A feature splitting module, the feature splitting module is electrically connected to the text input module, and the feature splitting module is used to perform text splitting on the input text information according to a text splitting algorithm to obtain a plurality of text feature sets to be matched;

[0125] A matching calculation module, the matching calculation module is electrically connected to the storage module and the feature separation module, and the matching calculation module is used to calculate the matching degree of each to-be-matched text feature set and the text database according to a matching degree algorithm;

[0126] A screening module, the screening module is electrically connected to the matching calculation module, and the screening module is used to screen out a to-be-matched text feature set corresponding to a maximum matching degree, and use the to-be-matched text feature set as an image generation feature group;

[0127] An image generation module is electrically connected to the screening module, and is used to obtain an image feature group based on the image feature corresponding to each text feature in the image feature group, and to combine several image features in the image feature group to generate an image.

[0128] The feature splitting module comprises:

[0129] A splitting condition unit, wherein the splitting condition unit is used to generate text splitting restriction conditions;

[0130] A text splitting unit, wherein the text splitting unit is used to generate a plurality of text splitting arrays under the text splitting restriction condition and extract the to-be-matched text features of the corresponding number of words from the input text information in turn according to the numbers in the text splitting arrays;

[0131] The matching calculation module includes:

[0132] An overlap calculation unit, the overlap calculation unit is used to calculate the overlap degree between each to-be-matched text feature set and the text database;

[0133] An association calculation unit, the association calculation unit is used to calculate the feature association degree of the to-be-matched text feature set, and obtain the feature association degree of the to-be-matched text feature set;

[0134] A matching calculation unit is used to calculate the matching degree between the feature set of the to-be-matched text and the text database through a matching degree calculation formula based on the feature association degree of the feature set of the to-be-matched text and the overlap degree between the feature set of the to-be-matched text and the text database.

[0135] The usage process of the above-mentioned image generation system based on precise matching of text features is as follows:

[0136] Step 1: Input text information through the input module;

[0137] Step 2: The splitting condition unit obtains the total number of characters M of the input text information; sequentially traverses the positive integers in the range of 0-M as the text split number n; based on the total number of characters M of the input text information and the text split number n, generates a text splitting restriction condition;

[0138] Step 3: Under the text splitting restriction condition, generate several text splitting arrays and extract the to-be-matched text features of the corresponding number of words from the input text information in turn according to the numbers in the text splitting arrays;

[0139] Step 4: The overlap calculation unit calculates the overlap between each text feature set to be matched and the text database;

[0140] Step 5: The association calculation unit calculates the feature association degree of the to-be-matched text feature set to obtain the feature association degree of the to-be-matched text feature set;

[0141] Step 6: The matching calculation unit is used to calculate the matching degree between the feature set of the to-be-matched text and the text database through a matching degree calculation formula based on the feature association degree of the feature set of the to-be-matched text and the overlap degree between the feature set of the to-be-matched text and the text database;

[0142] Step 7: The screening module screens out the to-be-matched text feature set corresponding to the maximum matching degree, and uses the to-be-matched text feature set as the image generation feature group;

[0143] Step 8: The image generation module obtains the image feature group based on the image feature corresponding to each text feature in the image feature group and combines several image features in the image feature group to generate an image.

[0144] Furthermore, the present solution also proposes a storage medium on which a computer-readable program is stored. When the computer-readable program is called, the above-mentioned image generation method based on precise matching of text features is executed.

[0145] It is understandable that the storage medium may be a magnetic medium, such as a floppy disk, a hard disk, or a magnetic tape; an optical medium, such as a DVD; or a semiconductor medium, such as a solid state drive (SSD).

[0146] In summary, the advantages of the present invention are that it can effectively ensure the overall matching of the input text information, thereby ensuring a high-precision match between the produced image and the text features, and the sudden feature information in the image greatly improves the image generation quality.

[0147] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for generating an image based on accurate matching of text features, characterized in that: include: Establishing a text database, wherein the text database includes a plurality of text feature classes, and each text feature class includes a plurality of text features of the same type; Get the input text information; The input text information is split according to the text splitting algorithm to obtain a plurality of text feature sets to be matched; According to the matching algorithm, the matching degree between each text feature set to be matched and the text database is calculated respectively; Filter out a to-be-matched text feature set corresponding to the maximum matching degree, and use the to-be-matched text feature set as an image generation feature group; Based on the image feature corresponding to each text feature in the image generating feature group, an image feature group is obtained; Combining several image features in the image feature group to generate an image; The step of calculating the matching degree of each to-be-matched text feature set and the text database according to the matching degree algorithm specifically includes: Calculate the overlap between each text feature set to be matched and the text database; Calculate the feature relevance of the text feature set to be matched, and obtain the feature relevance of the text feature set to be matched; Based on the feature relevance of the to-be-matched text feature set and the overlap between the to-be-matched text feature set and the text database, the matching degree between the to-be-matched text feature set and the text database is calculated using a matching degree calculation formula; The matching degree calculation formula is specifically: P=C h ×G l In the formula, P is the matching degree between the feature set of the text to be matched and the text database, C h G is the overlap between the feature set of the text to be matched and the text database, l is the feature correlation of the text feature set to be matched; The calculation of the overlap between each to-be-matched text feature set and the text database specifically includes: Traverse all text feature classes to determine whether the to-be-matched text feature in the to-be-matched text feature set is contained in the text feature class. If so, record the to-be-matched text feature as a coincident text feature; if not, record the to-be-matched text feature as a non-coincidence text feature. Get the word count of all overlapping text features; Based on the number of characters of the overlapping text features and the total number of characters of the input text information, the overlap between the to-be-matched text feature set and the text database is calculated using an overlap calculation formula; The calculation formula of the overlap degree is: Where N j is the number of characters in the jth overlapping text feature, and m is the total number of overlapping text features.

2. The image generation method based on accurate text feature matching according to claim 1, characterized in that: The step of performing text splitting on the input text information according to the text splitting algorithm to obtain a plurality of text feature sets to be matched specifically includes: Get the total number of words M of the input text information; Sequentially traverse the positive integers in the range of 0-M as the number of text splits n; Generate text splitting restriction conditions based on the total number of words M of the input text information and the number of text splits n; Under the constraints of text splitting conditions, generate several text splitting arrays; According to the numbers in the text split array, extract the text features to be matched with the corresponding number of words from the input text information in turn; All the to-be-matched text features are grouped into a to-be-matched text feature set; The expression of the text splitting restriction condition is: In the formula, z i is the number of characters in the i-th text feature to be matched, and N+ is a set of positive integers.

3. The image generation method based on accurate text feature matching according to claim 2 is characterized in that: The step of calculating the feature relevance of the to-be-matched text feature set and obtaining the feature relevance of the to-be-matched text feature set specifically includes: Sort all the overlapping text features according to the order in which they appear in the input text information; Based on the text feature class described in each overlapping text feature, respectively analyzing the correlation between each overlapping text feature and its adjacent previous overlapping text feature and next overlapping text feature; The maximum value of the correlation between the coincident text feature and its adjacent previous coincident text feature and the next coincident text feature is taken as the correlation of the coincident text feature; According to the comprehensive formula of feature relevance, the feature relevance of the set of text features to be matched is calculated based on each overlapping text feature.

4. The image generation method based on accurate text feature matching according to claim 3 is characterized in that: The feature correlation comprehensive formula is specifically: In the formula, g j is the relevance of the jth overlapping text feature.

5. An image generation system based on accurate matching of text features, characterized in that: The method for generating an image based on accurate matching of text features as claimed in any one of claims 1 to 4 comprises: A storage module, wherein the storage module is used to store a text database; A text input module, wherein the text input module is used to obtain input text information; A feature splitting module, the feature splitting module is electrically connected to the text input module, and the feature splitting module is used to perform text splitting on the input text information according to a text splitting algorithm to obtain a plurality of text feature sets to be matched; A matching calculation module, the matching calculation module is electrically connected to the storage module and the feature separation module, and the matching calculation module is used to calculate the matching degree of each to-be-matched text feature set and the text database according to a matching degree algorithm; A screening module, the screening module is electrically connected to the matching calculation module, and the screening module is used to screen out a to-be-matched text feature set corresponding to a maximum matching degree, and use the to-be-matched text feature set as an image generation feature group; An image generation module is electrically connected to the screening module, and is used to obtain an image feature group based on the image feature corresponding to each text feature in the image feature group, and to combine several image features in the image feature group to generate an image.

6. The image generation system based on accurate text feature matching according to claim 5, characterized in that: The feature splitting module comprises: A splitting condition unit, wherein the splitting condition unit is used to generate text splitting restriction conditions; A text splitting unit, wherein the text splitting unit is used to generate a plurality of text splitting arrays under the text splitting restriction condition and extract the to-be-matched text features of the corresponding number of words from the input text information in turn according to the numbers in the text splitting arrays; The matching calculation module includes: An overlap calculation unit, the overlap calculation unit is used to calculate the overlap degree between each to-be-matched text feature set and the text database; An association calculation unit, the association calculation unit is used to calculate the feature association degree of the to-be-matched text feature set, and obtain the feature association degree of the to-be-matched text feature set; The matching calculation unit is used to calculate the matching degree between the to-be-matched text feature set and the text database through a matching degree calculation formula based on the feature association degree of the to-be-matched text feature set and the overlap degree between the to-be-matched text feature set and the text database.

7. A computer-readable storage medium having a computer-readable program stored thereon, characterized in that: When the computer-readable program is called, the image generation method based on precise matching of text features as described in any one of claims 1 to 4 is executed.

Images