A logo detail analysis method and system based on big data

Abstract

The present invention relates to a logo detail analysis method and system based on big data, the method comprising: outputting the original feature points of a logo image according to the logo feature point recognition model; matching and replacing feature points according to the type of the original feature points, using the replacement feature points to replace the original feature points in the logo image to be analyzed, and generating a reconstructed image; generating an online distinguishing question and answer instruction according to the logo image to be analyzed and the reconstructed image, receiving a user's receipt information for the online distinguishing question and answer instruction; determining a first attention degree and a second attention degree of each original feature point according to the receipt information; verifying the first attention degree and the second attention degree of each original feature point until the verification is qualified and a final attention degree is generated. The present invention can accurately obtain the user's attention degree to each original feature point of the logo, and the final attention degree of each original feature point obtained by the present invention provides a basis for determining the importance of each detail point of the logo.

Description

Technical Field

[0001] The present invention relates to the technical field of logistics management, and in particular to a logo detail analysis method and system based on big data. Background Art

[0002] Logo is an important part of brand identification, which helps consumers quickly identify a specific brand among many brands. Logos usually convey the core values, characteristics or vision of the brand. The existing logo detail analysis and evaluation systems currently include reasonable analysis of color matching, reasonable evaluation of graphic design, text layout detection, copyright and legal compliance detection, etc. Analyzing a logo to determine the importance of each detail point is a very critical and important process because it helps to understand and evaluate the information conveyed by the logo and its role in brand identification. However, the main function of the existing logo detail analysis and evaluation system is rationality detection, and there is a lack of analysis of the importance of logo details. Therefore, the present invention proposes a logo detail analysis method and system based on big data. Summary of the invention

[0003] The purpose of the present invention is to provide a logo detail analysis method and system based on big data to solve the problems raised in the above background technology.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A first aspect of the present invention provides a logo detail analysis method based on big data, the method comprising:

[0006] The sample data is used to train the neural network model to obtain a logo feature point recognition model, and the original feature points of the logo image are output according to the logo feature point recognition model and the input logo image to be analyzed, and each original feature point is marked with its corresponding type;

[0007] Matching replacement feature points according to the types of the original feature points, using the replacement feature points to replace the original feature points in the logo image to be analyzed, to generate a first reconstructed image and a second reconstructed image;

[0008] Generate an online distinguishing question and answer instruction according to the logo image to be analyzed and the first reconstructed image and the second reconstructed image respectively, and receive a user's receipt information for the online distinguishing question and answer instruction, wherein the receipt information includes a first receipt information set and a second receipt information set;

[0009] Determine a first attention level and a second attention level of each original feature point according to the first receipt information set and the second receipt information set;

[0010] The first attention degree and the second attention degree of each original feature point are verified until the verification is qualified to generate a final attention degree.

[0011] Furthermore, the step of matching and replacing feature points according to the types of the original feature points, and using the replacement feature points to replace the original feature points in the logo image to be analyzed, to generate the first reconstructed image and the second reconstructed image includes:

[0012] Acquire the type of the original feature point, and extract other feature points of the same type from a feature point storage library according to the type of the original feature point as target feature points;

[0013] Calculating the similarity between each target feature point and its original feature point of the same type based on a preset similarity calculation method, wherein the preset similarity calculation method includes geometric figure similarity, color similarity and text similarity;

[0014] Obtaining a preset similarity table, and obtaining a first similarity threshold in the preset similarity table according to a preset order, wherein the preset similarity table includes a plurality of similarity thresholds arranged in order;

[0015] Acquire target feature points whose similarity satisfies a first similarity threshold to form a first replacement feature point set, and randomly extract first replacement feature points corresponding to the original feature points from the first replacement feature point set;

[0016] Acquire target feature points whose similarity meets a second similarity threshold to form a second replacement feature point set, and randomly extract second replacement feature points corresponding to the original feature points from the second replacement feature point set;

[0017] Using the first replacement feature points to replace the original feature points in the logo image to be analyzed, to generate a first reconstructed image;

[0018] The second replacement feature points are used to replace the original feature points in the logo image to be analyzed to generate a second reconstructed image.

[0019] Furthermore, the step of generating an online distinguishing question and answer instruction based on the logo image to be analyzed and the first reconstructed image and the second reconstructed image, and receiving a user's receipt information for the online distinguishing question and answer instruction comprises:

[0020] Generating a first online difference question and answer instruction according to the logo image to be analyzed and the first reconstructed image;

[0021] Receiving first receipt information from the user in response to the first online distinguishing question and answer instruction, and obtaining a first receipt information set;

[0022] generating a second online difference question and answer instruction according to the logo image to be analyzed and the second reconstructed image;

[0023] A second receipt information of the user in response to the second online distinguishing question and answer instruction is received to obtain a second receipt information set, wherein the receipt information includes the identified distinguishing feature points and the identification time of each distinguishing feature point.

[0024] Furthermore, the step of determining the first attention level and the second attention level of each original feature point according to the first receipt information set and the second receipt information set respectively includes:

[0025] Sort the distinguishing feature points in each receipt information in the first receipt information set and the second receipt information set according to the recognition time to obtain a plurality of feature point sorting tables;

[0026] Obtain the ranking of each distinguishing feature point in each feature point sorting table, and accumulate the rankings of the same distinguishing feature points in the first receipt information set and the second receipt information set, respectively, to obtain a first total ranking and a second total ranking of each distinguishing feature point;

[0027] The first total ranking and the second total ranking of each distinguishing feature point are sorted respectively to obtain a first total ranking table and a second total ranking table, and the first attention degree of each original feature point is determined according to the first total ranking table, and the second attention degree of each original feature point is determined according to the second total ranking table.

[0028] Furthermore, the step of verifying the first attention level and the second attention level of each original feature point until the verification is qualified and the final attention level is generated includes:

[0029] Determining whether a difference between the first attention level and the second attention level of each original feature point satisfies a preset difference threshold;

[0030] The final attention level is determined based on whether the difference meets the difference threshold.

[0031] Furthermore, the step of determining the final attention level according to the result of the difference and the difference threshold being satisfied includes:

[0032] When a preset difference threshold is met, a final attention level is generated according to the first attention level and the second attention level;

[0033] When the preset difference threshold is not met, the third preset threshold in the similarity preset table is obtained, the third attention is determined according to the third preset threshold, and it is determined whether the difference between the second attention and the third attention meets the preset difference threshold; when the preset difference threshold is met, the final attention is generated according to the second attention and the third attention; when the preset difference threshold is not met, the similarity thresholds sorted later in the similarity preset table are obtained in sequence until the final attention is generated.

[0034] The technical solution of the present invention also provides a logo detail analysis system based on big data, the system comprising:

[0035] A feature point acquisition module is used to train a neural network model using sample data to obtain a logo feature point recognition model, and output original feature points of the logo image according to the logo feature point recognition model and the input logo image to be analyzed, wherein each original feature point is marked with its corresponding type;

[0036] An image reconstruction module, used to match and replace feature points according to the types of the original feature points, and use the replacement feature points to replace the original feature points in the logo image to be analyzed, so as to generate a first reconstructed image and a second reconstructed image;

[0037] A receipt information acquisition module, used to generate an online distinguishing question and answer instruction according to the logo image to be analyzed and the first reconstructed image and the second reconstructed image, and receive a user's receipt information for the online distinguishing question and answer instruction;

[0038] A module for initially determining the degree of attention, used to determine the first degree of attention and the second degree of attention of each original feature point according to the first receipt information set and the second receipt information set;

[0039] The attention degree final determination module is used to verify the first attention degree and the second attention degree of each original feature point until the verification is qualified to generate the final attention degree.

[0040] Furthermore, the image reconstruction module includes:

[0041] a target data acquisition unit, configured to acquire the type of the original feature point, and extract other feature points of the same type from a feature point storage library according to the type of the original feature point as target feature points;

[0042] A similarity calculation unit, used to calculate the similarity between each target feature point and its original feature point of the same type based on a preset similarity calculation method, wherein the preset similarity calculation method includes geometric figure similarity, color similarity and text similarity;

[0043] A threshold acquisition unit, used to acquire a similarity preset table, and acquire a first similarity threshold in the similarity preset table according to a preset order, wherein the similarity preset table includes a plurality of similarity thresholds arranged in order;

[0044] A first feature point extraction unit is used to obtain target feature points whose similarity meets a first similarity threshold to form a first replacement feature point set, and randomly extract first replacement feature points corresponding to the original feature points from the first replacement feature point set;

[0045] A second feature point extraction unit is used to obtain target feature points whose similarity meets a second similarity threshold to form a second replacement feature point set, and randomly extract second replacement feature points corresponding to the original feature points from the second replacement feature point set;

[0046] a first image reconstruction unit, configured to replace original feature points in the logo image to be analyzed with the first replacement feature points to generate a first reconstructed image;

[0047] The second image reconstruction unit is used to replace the original feature points in the logo image to be analyzed with the second replacement feature points to generate a second reconstructed image.

[0048] Furthermore, the receipt information acquisition module includes:

[0049] A first instruction generating unit, configured to generate a first online distinguishing question and answer instruction according to the logo image to be analyzed and the first reconstructed image;

[0050] A first information acquisition unit, configured to receive first receipt information of a user in response to a first online distinguishing question and answer instruction, and obtain a first receipt information set;

[0051] A second instruction generating unit, configured to generate a second online distinguishing question and answer instruction according to the logo image to be analyzed and the second reconstructed image;

[0052] The second information acquisition unit is used to receive the second receipt information of the user in response to the second online distinguishing question and answer instruction, and obtain a second receipt information set, wherein the receipt information includes the identified distinguishing feature points and the identification time of each distinguishing feature point.

[0053] Furthermore, the attention degree preliminary determination module includes:

[0054] A feature point sorting unit, used to sort the distinguishing feature points in each receipt information in the first receipt information set and the second receipt information set according to the recognition time, to obtain a plurality of feature point sorting tables;

[0055] A feature point total ranking unit, used to obtain the ranking of each distinguishing feature point in each feature point ranking table, and to accumulate the rankings of the same distinguishing feature points in the first receipt information set and the second receipt information set, respectively, to obtain a first total ranking and a second total ranking of each distinguishing feature point;

[0056] The unit for initially determining the degree of attention is used to sort the first total ranking and the second total ranking of each distinguishing feature point respectively to obtain a first total ranking table and a second total ranking table, determine the first degree of attention of each original feature point according to the first total ranking table, and determine the second degree of attention of each original feature point according to the second total ranking table.

[0057] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention provides a logo detail analysis method and system based on big data, which can accurately obtain the user's attention to each original feature point of the logo. In the present invention, the logo image is identified and analyzed by a logo feature point recognition model to obtain the original feature points of the logo, and the original feature points are replaced to generate a first reconstructed image and a second reconstructed image, and the user's receipt information for the recognition of the first reconstructed image and the second reconstructed image with the logo image to be analyzed is received, and the attention of each original feature point is determined by analyzing the receipt information. The final attention of each original feature point obtained by the present invention provides a basis for determining the importance of each detail point of the logo. Determining the importance of each detail point helps to ensure that the key elements in the logo can effectively convey the core values ​​and characteristics of the brand, enhance the brand's recognition and memory points, and also helps to optimize the design of the logo. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention.

[0059] Figure 1 This is a flowchart of a logo detail analysis method based on big data.

[0060] Figure 2 This is a flowchart of the image reconstruction step in a logo detail analysis method based on big data.

[0061] Figure 3 The present invention is a flowchart of the steps of determining the first attention level and the second attention level in a logo detail analysis method based on big data.

[0062] Figure 4 This is a structural diagram of a logo detail analysis system based on big data. DETAILED DESCRIPTION

[0063] The technical solution of the present invention is further described in detail below in conjunction with specific implementation methods.

[0064] like Figure 1 As shown, an embodiment of the present invention provides a logo detail analysis method based on big data, and the method includes the following steps:

[0065] Step S100, using sample data to train a neural network model to obtain a logo feature point recognition model, and outputting original feature points of the logo image according to the logo feature point recognition model and the input logo image to be analyzed; the logo image to be analyzed is composed of a plurality of original feature points, the original feature points include colors, geometric figures and texts in the logo image, the geometric figures include circles, squares and ellipses, etc., and each original feature point is marked with its corresponding type, wherein the types of the original feature points include geometric figures, colors and texts; the logo image in the sample data is marked with feature point information, and after the sample data is used to train the neural network model, when the accuracy of the output result reaches a preset threshold, it can be determined that the training is completed and the logo feature point recognition model is obtained;

[0066] Step S200, matching replacement feature points according to the types of the original feature points, using the replacement feature points to replace the original feature points in the logo image to be analyzed, and generating a first reconstructed image and a second reconstructed image;

[0067] Step S300: Generate an online distinguishing question and answer instruction according to the logo image to be analyzed and the first reconstructed image and the second reconstructed image, and receive a user's receipt information for the online distinguishing question and answer instruction, wherein the receipt information includes a first receipt information set and a second receipt information set;

[0068] Among them, a first online distinguishing question and answer instruction is generated according to the logo image to be analyzed and the first reconstructed image; a first receipt information of the user for the first online distinguishing question and answer instruction is received to obtain a first receipt information set; a second online distinguishing question and answer instruction is generated according to the logo image to be analyzed and the second reconstructed image; a second receipt information of the user for the second online distinguishing question and answer instruction is received to obtain a second receipt information set, and the receipt information includes the identified distinguishing feature points and the recognition time of each distinguishing feature point.

[0069] Step S400, determining a first attention level and a second attention level of each original feature point according to the first receipt information set and the second receipt information set respectively;

[0070] Step S500: verify the first attention level and the second attention level of each original feature point until the verification is qualified and a final attention level is generated.

[0071] In this embodiment, the original feature points of the logo image to be analyzed are obtained based on the logo feature point recognition model analysis, and then the replacement feature points of each original feature point are obtained according to the similarity matching, and the replacement feature points are used to replace the original feature points in the logo image to be analyzed to generate a first reconstructed image and a second reconstructed image; an online distinguishing question and answer instruction is generated according to the logo image to be analyzed and the first reconstructed image and the second reconstructed image respectively, and the user's receipt information for the online distinguishing question and answer instruction is received, and the first attention and the second attention of each original feature point are obtained according to the receipt information analysis, and finally the first attention and the second attention are verified, and the final attention is obtained according to the verification result. The final attention of each original feature point obtained by the present invention provides a basis for determining the importance of each detail point of the logo. Determining the importance of each detail point helps to ensure that the key elements in the logo can effectively convey the core values ​​and characteristics of the brand, enhance the brand's recognition and memory points, and also helps to optimize the design of the logo.

[0072] See also Figure 2 As another preferred embodiment of the present invention, the step of matching and replacing feature points according to the types of the original feature points, replacing the original feature points in the logo image to be analyzed with the replacement feature points, and generating the first reconstructed image and the second reconstructed image comprises:

[0073] Step S201, obtaining the type of the original feature point, and extracting other feature points of the same type from a feature point repository according to the type of the original feature point as target feature points; wherein the feature point repository contains a large amount of feature point data, and each feature point is marked with its corresponding type;

[0074] Step S202: calculating the similarity between each target feature point and its original feature point of the same type based on a preset similarity calculation method, wherein the preset similarity calculation method includes geometric figure similarity, color similarity and text similarity;

[0075] Among them, the similarity of geometric figures can be calculated using Euclidean distance similarity (geometric figures can be represented as point sets or vectors, and the Euclidean distance is used to measure the similarity between them. The smaller the Euclidean distance, the more similar the two geometric figures are), the similarity of colors can be calculated using the distance algorithm in the HSV color space (the color is converted from the RGB space to the HSV space, where H represents hue, S represents saturation, and V represents lightness. Calculating the distance between the HSV values ​​of two colors can measure their similarity), and the similarity of text can be calculated using the edit distance (measure the minimum number of editing operations (such as insertion, deletion, or replacement) required to convert one string to another. The smaller the edit distance, the more similar the two strings are). The specific similarity calculation method is not described in detail here and is common knowledge in the art.

[0076] Step S203: obtaining a preset similarity table, and obtaining a first similarity threshold in the preset similarity table in a preset order, wherein the preset similarity table includes a plurality of similarity thresholds arranged in order;

[0077] Step S204: acquiring target feature points whose similarity satisfies a first similarity threshold to form a first replacement feature point set, and randomly extracting first replacement feature points corresponding to the original feature points from the first replacement feature point set;

[0078] Step S205: acquiring target feature points whose similarity satisfies a second similarity threshold to form a second replacement feature point set, and randomly extracting second replacement feature points corresponding to the original feature points from the second replacement feature point set;

[0079] Step S206: using the first replacement feature points to replace the original feature points in the logo image to be analyzed, to generate a first reconstructed image;

[0080] Step S207: using the second replacement feature points to replace the original feature points in the logo image to be analyzed, to generate a second reconstructed image;

[0081] When the first replacement feature point and the second replacement feature point replace the original feature point, it is ensured that the first replacement feature point and the second replacement feature point are consistent in position and direction with the original feature point.

[0082] See also Figure 3 As another preferred embodiment of the present invention, the step of determining the first attention level and the second attention level of each original feature point according to the first receipt information set and the second receipt information set respectively comprises:

[0083] Step S401, sorting the distinguishing feature points in each receipt information in the first receipt information set and the second receipt information set according to the recognition time to obtain a plurality of feature point sorting tables, where one receipt information corresponds to one feature point sorting table;

[0084] Step S402: Obtain the ranking of each distinguishing feature point in each feature point ranking table, and add up the rankings of the same distinguishing feature points in the first receipt information set and the second receipt information set to obtain a first total ranking and a second total ranking of each distinguishing feature point;

[0085] Step S403, sort the first total ranking and the second total ranking of each distinguishing feature point respectively to obtain a first total ranking table and a second total ranking table, determine the first attention degree of each original feature point according to the first total ranking table, and determine the second attention degree of each original feature point according to the second total ranking table.

[0086] In this embodiment, the user's attention to each original feature point is determined by the user's recognition speed of each distinguishing feature point.

[0087] As another preferred embodiment of the present invention, the step of verifying the first attention level and the second attention level of each original feature point until the verification is qualified and the final attention level is generated includes:

[0088] Determining whether a difference between the first attention level and the second attention level of each original feature point satisfies a preset difference threshold;

[0089] When a preset difference threshold is met, a final attention level is generated according to the first attention level and the second attention level;

[0090] Specifically, the following steps are performed: calculating the average value between the first attention degree and the second attention degree of each original feature point to obtain the final attention degree, or determining the second attention degree as the final attention degree;

[0091] When the preset difference threshold is not met, the third preset threshold in the similarity preset table is obtained, the third attention is determined according to the third preset threshold, and it is determined whether the difference between the second attention and the third attention meets the preset difference threshold; when the preset difference threshold is met, the final attention is generated according to the second attention and the third attention; when the preset difference threshold is not met, the similarity thresholds sorted later in the similarity preset table are obtained in sequence until the final attention is generated.

[0092] In this embodiment, the method used to determine the third attention level according to the third preset threshold is the same as the method used to determine the first attention level and the second attention level as mentioned above, wherein the final attention level is generated based on the second attention level and the third attention level, and the final attention level can be obtained by calculating the average value between the first attention level and the second attention level of each original feature point.

[0093] See also Figure 4 Another object of an embodiment of the present invention is to provide a logo detail analysis system based on big data, the system comprising:

[0094] The feature point acquisition module 100 is used to train a neural network model using sample data to obtain a logo feature point recognition model, and output the original feature points of the logo image according to the logo feature point recognition model and the input logo image to be analyzed, wherein the logo image to be analyzed is decomposed to obtain a plurality of original feature points, and the logo image to be analyzed is composed of a plurality of original feature points, wherein the types of the feature points include geometric figures, colors, and texts; the logo image in the sample data is annotated with feature point data;

[0095] An image reconstruction module 200, configured to match and replace feature points according to the types of the original feature points, and use the replacement feature points to replace the original feature points in the logo image to be analyzed, so as to generate a first reconstructed image and a second reconstructed image;

[0096] A receipt information acquisition module 300 is used to generate an online distinguishing question and answer instruction according to the logo image to be analyzed and the first reconstructed image and the second reconstructed image, and receive a user's receipt information for the online distinguishing question and answer instruction;

[0097] A preliminary attention level determination module 400 is used to determine the first attention level and the second attention level of each original feature point according to the first receipt information set and the second receipt information set respectively;

[0098] The attention level final determination module 500 is used to verify the first attention level and the second attention level of each original feature point until the verification is qualified to generate a final attention level.

[0099] As another preferred embodiment of the present invention, the image reconstruction module 200 includes:

[0100] a target data acquisition unit, configured to acquire the type of the original feature point, and extract other feature points of the same type from a feature point storage library according to the type of the original feature point as target feature points;

[0101] A similarity calculation unit, used to calculate the similarity between each target feature point and its original feature point of the same type based on a preset similarity calculation method, wherein the preset similarity calculation method includes geometric figure similarity, color similarity and text similarity;

[0102] A threshold acquisition unit, used to acquire a similarity preset table, and acquire a first similarity threshold in the similarity preset table according to a preset order, wherein the similarity preset table includes a plurality of similarity thresholds arranged in order;

[0103] A first feature point extraction unit is used to obtain target feature points whose similarity meets a first similarity threshold to form a first replacement feature point set, and randomly extract first replacement feature points corresponding to the original feature points from the first replacement feature point set;

[0104] A second feature point extraction unit is used to obtain target feature points whose similarity meets a second similarity threshold to form a second replacement feature point set, and randomly extract second replacement feature points corresponding to the original feature points from the second replacement feature point set;

[0105] a first image reconstruction unit, configured to replace original feature points in the logo image to be analyzed with the first replacement feature points to generate a first reconstructed image;

[0106] a second image reconstruction unit, configured to replace the original feature points in the logo image to be analyzed with the second replacement feature points to generate a second reconstructed image;

[0107] When the first replacement feature point and the second replacement feature point replace the original feature point, it is ensured that the first replacement feature point and the second replacement feature point are consistent in position and direction with the original feature point.

[0108] As another preferred embodiment of the present invention, the receipt information acquisition module 300 includes:

[0109] A first instruction generating unit, configured to generate a first online distinguishing question and answer instruction according to the logo image to be analyzed and the first reconstructed image;

[0110] A first information acquisition unit, configured to receive first receipt information of a user in response to a first online distinguishing question and answer instruction, and obtain a first receipt information set;

[0111] A second instruction generating unit, configured to generate a second online distinguishing question and answer instruction according to the logo image to be analyzed and the second reconstructed image;

[0112] The second information acquisition unit is used to receive the second receipt information of the user in response to the second online distinguishing question and answer instruction, and obtain a second receipt information set, wherein the receipt information includes the identified distinguishing feature points and the identification time of each distinguishing feature point.

[0113] As another preferred embodiment of the present invention, the attention level preliminary determination module 400 includes:

[0114] A feature point sorting unit, used to sort the distinguishing feature points in each receipt information in the first receipt information set and the second receipt information set according to the recognition time, to obtain a plurality of feature point sorting tables, one receipt information corresponds to one feature point sorting table;

[0115] A feature point total ranking unit, used to obtain the ranking of each distinguishing feature point in each feature point ranking table, and to accumulate the rankings of the same distinguishing feature points in the first receipt information set and the second receipt information set, respectively, to obtain a first total ranking and a second total ranking of each distinguishing feature point;

[0116] The unit for initially determining the degree of attention is used to sort the first total ranking and the second total ranking of each distinguishing feature point respectively to obtain a first total ranking table and a second total ranking table, determine the first degree of attention of each original feature point according to the first total ranking table, and determine the second degree of attention of each original feature point according to the second total ranking table.

[0117] As another preferred embodiment of the present invention, the attention degree final determination module 500 includes:

[0118] A difference determination unit, configured to determine whether a difference between the first attention degree and the second attention degree of each original feature point satisfies a preset difference threshold;

[0119] A final attention level first determination unit, configured to generate a final attention level according to the first attention level and the second attention level when a preset difference threshold is met;

[0120] The second final attention determination unit is used to obtain the third preset threshold in the similarity preset table when the preset difference threshold is not met, determine the third attention according to the third preset threshold, and judge whether the difference between the second attention and the third attention meets the preset difference threshold; when the preset difference threshold is met, generate the final attention according to the second attention and the third attention; when the preset difference threshold is not met, obtain the similarity thresholds sorted later in the similarity preset table in sequence until the final attention is generated.

[0121] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0122] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A logo detail analysis method based on big data, characterized in that: The method comprises: The sample data is used to train the neural network model to obtain a logo feature point recognition model, and the original feature points of the logo image are output according to the logo feature point recognition model and the input logo image to be analyzed, and each original feature point is marked with its corresponding type; Matching replacement feature points according to the types of the original feature points, using the replacement feature points to replace the original feature points in the logo image to be analyzed, to generate a first reconstructed image and a second reconstructed image; Generate an online distinguishing question and answer instruction according to the logo image to be analyzed and the first reconstructed image and the second reconstructed image respectively, and receive a user's receipt information for the online distinguishing question and answer instruction, wherein the receipt information includes a first receipt information set and a second receipt information set; Determine a first attention level and a second attention level of each original feature point according to the first receipt information set and the second receipt information set; The first attention degree and the second attention degree of each original feature point are verified until the verification is qualified to generate a final attention degree.

2. The logo detail analysis method based on big data according to claim 1 is characterized in that: The step of matching and replacing feature points according to the types of the original feature points, using the replacement feature points to replace the original feature points in the logo image to be analyzed, and generating the first reconstructed image and the second reconstructed image comprises: Acquire the type of the original feature point, and extract other feature points of the same type from a feature point storage library according to the type of the original feature point as target feature points; Calculating the similarity between each target feature point and its original feature point of the same type based on a preset similarity calculation method, wherein the preset similarity calculation method includes geometric figure similarity, color similarity and text similarity; Obtaining a preset similarity table, and obtaining a first similarity threshold in the preset similarity table according to a preset order, wherein the preset similarity table includes a plurality of similarity thresholds arranged in order; Acquire target feature points whose similarity satisfies a first similarity threshold to form a first replacement feature point set, and randomly extract first replacement feature points corresponding to the original feature points from the first replacement feature point set; Acquire target feature points whose similarity meets a second similarity threshold to form a second replacement feature point set, and randomly extract second replacement feature points corresponding to the original feature points from the second replacement feature point set; Using the first replacement feature points to replace the original feature points in the logo image to be analyzed, to generate a first reconstructed image; The second replacement feature points are used to replace the original feature points in the logo image to be analyzed to generate a second reconstructed image.

3. The logo detail analysis method based on big data according to claim 2 is characterized in that: The step of generating an online distinguishing question and answer instruction according to the logo image to be analyzed and the first reconstructed image and the second reconstructed image, and receiving a user's receipt information for the online distinguishing question and answer instruction comprises: Generating a first online difference question and answer instruction according to the logo image to be analyzed and the first reconstructed image; Receiving first receipt information from the user in response to the first online distinguishing question and answer instruction, and obtaining a first receipt information set; generating a second online difference question and answer instruction according to the logo image to be analyzed and the second reconstructed image; A second receipt information of the user in response to the second online distinguishing question and answer instruction is received to obtain a second receipt information set, wherein the receipt information includes the identified distinguishing feature points and the identification time of each distinguishing feature point.

4. The logo detail analysis method based on big data according to claim 3 is characterized in that: The step of determining the first attention level and the second attention level of each original feature point according to the first receipt information set and the second receipt information set respectively comprises: Sort the distinguishing feature points in each receipt information in the first receipt information set and the second receipt information set according to the recognition time to obtain a plurality of feature point sorting tables; Obtain the ranking of each distinguishing feature point in each feature point sorting table, and respectively accumulate the rankings of the same distinguishing feature points in the first receipt information set and the second receipt information set to obtain a first total ranking and a second total ranking of each distinguishing feature point; The first total ranking and the second total ranking of each distinguishing feature point are sorted respectively to obtain a first total ranking table and a second total ranking table, and the first attention degree of each original feature point is determined according to the first total ranking table, and the second attention degree of each original feature point is determined according to the second total ranking table.

5. The logo detail analysis method based on big data according to claim 2 is characterized in that: The step of verifying the first attention level and the second attention level of each original feature point until the verification is qualified and the final attention level is generated comprises: Determining whether a difference between the first attention level and the second attention level of each original feature point satisfies a preset difference threshold; The final attention level is determined based on whether the difference meets the difference threshold.

6. The logo detail analysis method based on big data according to claim 5 is characterized in that: The step of determining the final attention degree according to the result of the difference and the difference threshold being satisfied comprises: When a preset difference threshold is met, a final attention level is generated according to the first attention level and the second attention level; When the preset difference threshold is not met, the third preset threshold in the similarity preset table is obtained, the third attention is determined according to the third preset threshold, and it is determined whether the difference between the second attention and the third attention meets the preset difference threshold; when the preset difference threshold is met, the final attention is generated according to the second attention and the third attention; when the preset difference threshold is not met, the similarity thresholds sorted later in the similarity preset table are obtained in sequence until the final attention is generated.

7. A logo detail analysis system based on big data, characterized in that: The system comprises: A feature point acquisition module is used to train a neural network model using sample data to obtain a logo feature point recognition model, and output original feature points of the logo image according to the logo feature point recognition model and the input logo image to be analyzed, wherein each original feature point is marked with its corresponding type; An image reconstruction module, used to match and replace feature points according to the types of the original feature points, and use the replacement feature points to replace the original feature points in the logo image to be analyzed, so as to generate a first reconstructed image and a second reconstructed image; A receipt information acquisition module, used to generate an online distinguishing question and answer instruction according to the logo image to be analyzed and the first reconstructed image and the second reconstructed image, and receive a user's receipt information for the online distinguishing question and answer instruction; A module for initially determining the degree of attention, used to determine the first degree of attention and the second degree of attention of each original feature point according to the first receipt information set and the second receipt information set; The attention degree final determination module is used to verify the first attention degree and the second attention degree of each original feature point until the verification is qualified to generate the final attention degree.

8. The logo detail analysis system based on big data according to claim 7 is characterized in that: The image reconstruction module comprises: a target data acquisition unit, configured to acquire the type of the original feature point, and extract other feature points of the same type from a feature point storage library according to the type of the original feature point as target feature points; A similarity calculation unit, used to calculate the similarity between each target feature point and its original feature point of the same type based on a preset similarity calculation method, wherein the preset similarity calculation method includes geometric figure similarity, color similarity and text similarity; A threshold acquisition unit, used to acquire a similarity preset table, and acquire a first similarity threshold in the similarity preset table according to a preset order, wherein the similarity preset table includes a plurality of similarity thresholds arranged in order; A first feature point extraction unit is used to obtain target feature points whose similarity meets a first similarity threshold to form a first replacement feature point set, and randomly extract first replacement feature points corresponding to the original feature points from the first replacement feature point set; A second feature point extraction unit is used to obtain target feature points whose similarity meets a second similarity threshold to form a second replacement feature point set, and randomly extract second replacement feature points corresponding to the original feature points from the second replacement feature point set; a first image reconstruction unit, configured to replace original feature points in the logo image to be analyzed with the first replacement feature points to generate a first reconstructed image; The second image reconstruction unit is used to replace the original feature points in the logo image to be analyzed with the second replacement feature points to generate a second reconstructed image.

9. The logo detail analysis system based on big data according to claim 8, characterized in that: The receipt information acquisition module includes: A first instruction generating unit, configured to generate a first online distinguishing question and answer instruction according to the logo image to be analyzed and the first reconstructed image; A first information acquisition unit, configured to receive first receipt information of a user in response to a first online distinguishing question and answer instruction, and obtain a first receipt information set; A second instruction generating unit, configured to generate a second online distinguishing question and answer instruction according to the logo image to be analyzed and the second reconstructed image; The second information acquisition unit is used to receive the second receipt information of the user in response to the second online distinguishing question and answer instruction, and obtain a second receipt information set, wherein the receipt information includes the identified distinguishing feature points and the identification time of each distinguishing feature point.

10. The logo detail analysis system based on big data according to claim 9, characterized in that: The attention degree preliminary determination module comprises: A feature point sorting unit, used to sort the distinguishing feature points in each receipt information in the first receipt information set and the second receipt information set according to the recognition time, to obtain a plurality of feature point sorting tables; A feature point total ranking unit, used to obtain the ranking of each distinguishing feature point in each feature point ranking table, and to accumulate the rankings of the same distinguishing feature points in the first receipt information set and the second receipt information set, respectively, to obtain a first total ranking and a second total ranking of each distinguishing feature point; The unit for initially determining the degree of attention is used to sort the first total ranking and the second total ranking of each distinguishing feature point respectively to obtain a first total ranking table and a second total ranking table, determine the first degree of attention of each original feature point according to the first total ranking table, and determine the second degree of attention of each original feature point according to the second total ranking table.

Images