Image data and structured data fusion analysis method based on artificial intelligence algorithm

Abstract

The present invention discloses an image data and structured data fusion analysis method based on an artificial intelligence algorithm, which relates to the field of data fusion technology. The method includes calculating the first image feature weight and the second image feature weight of each image feature vector, calculating the comprehensive weight of the image feature, and obtaining the first structural feature and the second structural feature weight of each structured data feature vector, calculating the comprehensive weight of the structural feature, deleting redundant image features or structural features sorted at the end after alignment processing, starting from the image feature with the largest comprehensive weight of the image feature, matching the image feature with the structural feature corresponding to the largest cosine similarity among the cosine similarities of all structured data features, repeating the above steps until all image features are matched. The method can promote cross-domain fusion between image data and structured data, and provide support for comprehensive data analysis and decision-making.

Description

Technical Field

[0001] The present invention relates to the technical field of data fusion, and in particular to a method for fusion analysis of image data and structured data based on an artificial intelligence algorithm. Background Art

[0002] With the rapid development of information technology and artificial intelligence, data processing and analysis capabilities have been significantly improved, especially in the field of multimedia data processing. Image data and structured data, as two important data types, play a vital role in various fields. Image data, with its rich visual information and intuitive presentation, is widely used in medical imaging, autonomous driving, intelligent security and other fields; while structured data, with its precise and standardized characteristics, shows great potential in data analysis and decision support. However, a single type of data is often difficult to meet the needs of complex application scenarios. Therefore, how to achieve the effective fusion and analysis of image data and structured data has become an important research topic.

[0003] In the Chinese invention application with application publication number CN117437513A, a method for image data fusion analysis based on big data is disclosed. To overcome the shortcomings of the prior art in multi-image processing, such as high performance requirements, limited single image information, high complexity and low efficiency of multi-image processing tasks, slow data retrieval, and inaccurate data, a method for image data fusion analysis based on big data is provided. Image analysis is carried out using a big data approach, multiple similar images are marked and stored in the same storage space, and the contents of multiple similar images are merged through image fusion. This improves the efficiency of data retrieval and the efficiency of the system in image analysis, making the image analysis results more universal.

[0004] In the above invention application, image analysis is carried out through big data, multiple similar images are marked and stored in the same storage space, and the contents of multiple similar pictures are merged through image fusion, but the basis is the fusion between data of the same type. However, a single type of data is often difficult to meet the needs of complex application scenarios, and often cannot fully and deeply reveal the laws and patterns behind the data.

[0005] To this end, the present invention provides an image data and structured data fusion analysis method based on artificial intelligence algorithm. Summary of the Invention

[0006] (1) Technical problems solved

[0007] In view of the shortcomings of the existing technology, the present invention provides an image data and structured data fusion analysis method based on artificial intelligence algorithm. The present invention calculates the first image feature weight of each image feature vector and the second image feature weight , calculate the comprehensive weight of image features , and obtain the first structural feature of each structured data feature vector and the second structural feature weight , calculate the comprehensive weight of structural features After alignment, delete the redundant image features or structural features that are ranked later, and calculate the comprehensive weight of the image features. Starting with the largest image feature, the image feature is matched with the structural feature corresponding to the largest cosine similarity among the cosine similarities of all structured data features, and the above steps are repeated until all image features are matched. This not only takes into account the similarity between features of the same type, but also the similarity between features of different types. It can promote cross-domain integration between image data and structured data, provide support for comprehensive data analysis and decision-making, and thus solve the technical problems recorded in the background technology.

[0008] (2) Technical solution

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for fusion analysis of image data and structured data based on artificial intelligence algorithms, comprising the following steps:

[0010] Use various feature extraction algorithms to extract image features of image data, then use dimensionality reduction technology to reduce the high-dimensional feature vector to a low-dimensional space, and calculate the cosine similarity between each image feature vector in the low-dimensional space. , and calculate the first image feature weight of each image feature vector ;

[0011] Use the NLP library to analyze structured data, extract structured text features, and calculate the cosine similarity between each structured text feature vector in a low-dimensional space. , and calculate the first structural feature weight of the structured text feature vector ;

[0012] Obtain image features and structured data features, and analyze the cosine similarity between each structured data feature and all image features , calculate the second structural feature weight of each structured data feature , and analyze the cosine similarity between each image feature and all structured data features , calculate the second image feature weight of each image feature ;

[0013] Get the first image feature weight of each image feature vector and the second image feature weight , calculate the comprehensive weight of image features , and obtain the first structural feature of each structured data feature vector and the second structural feature weight , calculate the comprehensive weight of structural features After alignment, delete the redundant image features or structural features that are ranked later, and calculate the comprehensive weight of the image features. Starting from the largest image feature, the image feature is matched with the structural feature corresponding to the largest cosine similarity among the cosine similarities of all structured data features, and the above steps are repeated until all image features are matched.

[0014] Furthermore, various feature extraction algorithms (such as SIFT, SURF, ORB and other local feature descriptors) and convolutional layers in deep learning methods are used to extract color features, texture features, shape features, spatial relationship features, etc. of image data to obtain high-level image feature text representations. The high-dimensional feature vectors are then reduced to a low-dimensional space through the bag-of-words model. The similarity between each image feature vector is calculated in the low-dimensional space to obtain the cosine similarity between each image feature vector. .

[0015] The bag-of-words model is the simplest text vectorization method. It treats text as an unordered collection of words, ignoring the order and grammatical relationships between words. The specific steps are as follows: Vocabulary Construction: Iterates through all texts, counts all occurrences of words, and constructs a vocabulary containing all words. Text Vectorization: For each text, a vector is created based on the vocabulary, with each dimension of the vector corresponding to a word in the vocabulary. If the text contains the word, the value of the corresponding dimension is the number of times the word appears in the text (or other weight); otherwise, it is 0.

[0016] Furthermore, the cosine similarity between the feature vectors of each image is obtained , calculate the first image feature weight of each image feature vector :

[0017]

[0018] in, a and b Both represent the sequential numbers of image feature vectors, and a =1, 2, ..., n, b =1, 2,…, n.

[0019] Furthermore, we use NLP libraries (such as NLTK, spaCy, Stanford NLP in Python) to perform more complex text analysis tasks on structured data, extract structured text features, and then use the bag-of-words model to reduce the high-dimensional feature vectors to a low-dimensional space. In the low-dimensional space, we calculate the similarity between each structured text feature vector and obtain the cosine similarity between each structured text feature vector. .

[0020] Structured data has well-defined fields and attributes, each with a specific data type and value range. This type of data is easy to organize, store, and analyze, and can be effectively processed and mined using various algorithms and tools. Structured data is typically presented in tables, tree structures, graphs, or other formats, such as tables in relational databases, CSV files, XML documents, and JSON data.

[0021] Furthermore, the cosine similarity between each structured text feature vector is obtained , calculate the first structural feature of the structured text feature vector :

[0022]

[0023] in, c and d Both represent the sequential numbers of image feature vectors, and c =1, 2, ..., m, d =1, 2,…, m.

[0024] Furthermore, the image features and structured data features are obtained, and the cosine similarity between each structured data feature and all image features is analyzed. , calculate the second structural feature weight of each structured data feature :

[0025]

[0026] Cosine similarity is a mathematical method that measures the similarity between two vectors in a multidimensional space. The core idea is to treat the vectors as rays starting from the origin and evaluate their similarity by calculating the cosine of the angle between the two rays.

[0027] Furthermore, the image features and structured data features are obtained, and the cosine similarity between each image feature and all structured data features is analyzed. , calculate the second image feature weight of each image feature :

[0028]

[0029] The second structural feature weight of each corresponding image feature The calculation formula is as above.

[0030] Furthermore, the first image feature weight of each image feature vector is obtained and the second image feature weight , calculate the comprehensive weight of image features :

[0031]

[0032] The comprehensive weight of the corresponding image features The calculation formula is as above.

[0033] Furthermore, the first structural feature of each structured data feature vector is obtained and the second structural feature weight , calculate the comprehensive weight of structural features :

[0034]

[0035] The comprehensive weight of the corresponding structural features The calculation formula is as above.

[0036] Furthermore, the comprehensive weight of image features is obtained and the comprehensive weight of structural features , after sorting them from large to small, the numbers are aligned and the redundant image features or structural features sorted later are deleted.

[0037] Furthermore, after the number of image features and structural features is consistent, the comprehensive weight of image features is obtained. Starting from the largest image feature, the image feature is matched with the structural feature corresponding to the largest cosine similarity among the cosine similarities of all structured data features, and the above steps are repeated until all image features are matched.

[0038] (3) Beneficial effects

[0039] The present invention provides an image data and structured data fusion analysis method based on artificial intelligence algorithm, which has the following beneficial effects:

[0040] 1. Use various feature extraction algorithms to extract image features of image data, and then use dimensionality reduction technology to reduce the high-dimensional feature vector to a low-dimensional space, and calculate the cosine similarity between each image feature vector in the low-dimensional space. , and calculate the first image feature weight of each image feature vector , we can evaluate the similarity between different image feature vectors, further understand which features play a key role in image classification, retrieval, fusion and other tasks, and provide a basis for subsequent classification, retrieval, fusion and other tasks.

[0041] 2. Use the NLP library to analyze structured data, extract structured text features, and calculate the cosine similarity between each structured text feature vector in a low-dimensional space. , and calculate the first structural feature of the structured text feature vector , can evaluate the similarity between different structured text feature vectors, which helps to gain a deeper understanding of the meaning and context of the data and provide strong support for subsequent analysis and processing.

[0042] 3. Obtain image features and structured data features, and analyze the cosine similarity between each structured data feature and all image features , calculate the second structural feature weight of each structured data feature , and analyze the cosine similarity between each image feature and all structured data features , calculate the second image feature weight of each image feature , we can analyze the impact of different types of data on the overall feature set, thereby improving the accuracy and effectiveness of feature representation, helping subsequent analysis and processing processes to focus more on key information and improve overall efficiency.

[0043] 4. Obtain the first image feature weight of each image feature vector and the second image feature weight , calculate the comprehensive weight of image features , and obtain the first structural feature of each structured data feature vector and the second structural feature weight , calculate the comprehensive weight of structural features After alignment, delete the redundant image features or structural features that are ranked later, and calculate the comprehensive weight of the image features. Starting from the largest image feature, the image feature is matched with the structural feature corresponding to the largest cosine similarity among the cosine similarities of all structured data features. The above steps are repeated until all image features are matched. This not only considers the similarity between features of the same type, but also the similarity between features of different types. It can promote the cross-domain integration between image data and structured data, and provide support for comprehensive data analysis and decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 The figure is a flow chart of the image data and structured data fusion analysis method based on artificial intelligence algorithm of the present invention. DETAILED DESCRIPTION

[0045] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0046] See also Figure 1 The present invention provides an image data and structured data fusion analysis method based on artificial intelligence algorithm, comprising the following steps:

[0047] Step 1: Use various feature extraction algorithms to extract image features of image data, and then use dimensionality reduction technology to reduce the high-dimensional feature vector to a low-dimensional space, and calculate the cosine similarity between each image feature vector in the low-dimensional space. , and calculate the first image feature weight of each image feature vector .

[0048] The step 1 includes the following:

[0049] Step 101: Use various feature extraction algorithms (such as SIFT, SURF, ORB and other local feature descriptors) and the convolution layer in the deep learning method to extract the color features, texture features, shape features, spatial relationship features, etc. of the image data to obtain a high-level image feature text representation. Then, use the bag-of-words model to reduce the high-dimensional feature vector to a low-dimensional space, calculate the similarity between each image feature vector in the low-dimensional space, and obtain the cosine similarity between each image feature vector. .

[0050] The bag-of-words model is the simplest text vectorization method. It treats text as an unordered collection of words, ignoring the order and grammatical relationships between words. The specific steps are as follows: Vocabulary Construction: Iterates through all texts, counts all occurrences of words, and constructs a vocabulary containing all words. Text Vectorization: For each text, a vector is created based on the vocabulary, with each dimension of the vector corresponding to a word in the vocabulary. If the text contains the word, the value of the corresponding dimension is the number of times the word appears in the text (or other weight); otherwise, it is 0.

[0051] Step 102: Obtain the cosine similarity between each image feature vector , calculate the first image feature weight of each image feature vector :

[0052]

[0053] in, a and b Both represent the sequential numbers of image feature vectors, and a =1, 2, ..., n, b =1, 2,…, n.

[0054] When using, combine the contents in steps 101 and 102:

[0055] Use various feature extraction algorithms to extract image features of image data, then use dimensionality reduction technology to reduce the high-dimensional feature vector to a low-dimensional space, and calculate the cosine similarity between each image feature vector in the low-dimensional space. , and calculate the first image feature weight of each image feature vector , we can evaluate the similarity between different image feature vectors, further understand which features play a key role in image classification, retrieval, fusion and other tasks, and provide a basis for subsequent classification, retrieval, fusion and other tasks.

[0056] Step 2: Use the NLP library to analyze the structured data, extract the structured text features, and calculate the cosine similarity between each structured text feature vector in the low-dimensional space. , and calculate the first structural feature weight of the structured text feature vector .

[0057] The second step includes the following:

[0058] Step 201: Use NLP libraries (such as NLTK, spaCy, Stanford NLP in Python) to perform more complex text analysis tasks on structured data, extract structured text features, and then use the bag-of-words model to reduce the high-dimensional feature vectors to a low-dimensional space. Calculate the similarity between each structured text feature vector in the low-dimensional space to obtain the cosine similarity between each structured text feature vector. .

[0059] Structured data has well-defined fields and attributes, each with a specific data type and value range. This type of data is easy to organize, store, and analyze, and can be effectively processed and mined using various algorithms and tools. Structured data is typically presented in tables, tree structures, graphs, or other formats, such as tables in relational databases, CSV files, XML documents, and JSON data.

[0060] Step 202: Obtain the cosine similarity between each structured text feature vector , calculate the first structural feature of the structured text feature vector :

[0061]

[0062] in, c and d Both represent the sequential numbers of image feature vectors, and c =1, 2, ..., m, d =1, 2,…, m.

[0063] When using, combine the contents in steps 201 and 202:

[0064] Use the NLP library to analyze structured data, extract structured text features, and calculate the cosine similarity between each structured text feature vector in a low-dimensional space. , and calculate the first structural feature of the structured text feature vector , can evaluate the similarity between different structured text feature vectors, which helps to gain a deeper understanding of the meaning and context of the data and provide strong support for subsequent analysis and processing.

[0065] Step 3: Obtain image features and structured data features, and analyze the cosine similarity between each structured data feature and all image features , calculate the second structural feature weight of each structured data feature , and analyze the cosine similarity between each image feature and all structured data features , calculate the second image feature weight of each image feature .

[0066] The step three includes the following:

[0067] Step 301: Obtain image features and structured data features, and analyze the cosine similarity between each structured data feature and all image features. , calculate the second structural feature weight of each structured data feature :

[0068]

[0069] Cosine similarity is a mathematical method that measures the similarity between two vectors in a multidimensional space. The core idea is to treat the vectors as rays starting from the origin and evaluate their similarity by calculating the cosine of the angle between the two rays.

[0070] Step 302: Obtain image features and structured data features, and analyze the cosine similarity between each image feature and all structured data features. , calculate the second image feature weight of each image feature :

[0071]

[0072] The second structural feature weight of each corresponding image feature The calculation formula is as above.

[0073] When using, combine the contents in steps 301 and 302:

[0074] Obtain image features and structured data features, and analyze the cosine similarity between each structured data feature and all image features , calculate the second structural feature weight of each structured data feature , and analyze the cosine similarity between each image feature and all structured data features , calculate the second image feature weight of each image feature , we can analyze the impact of different types of data on the overall feature set, thereby improving the accuracy and effectiveness of feature representation, helping subsequent analysis and processing processes to focus more on key information and improve overall efficiency.

[0075] Step 4: Obtain the first image feature weight of each image feature vector and the second image feature weight , calculate the comprehensive weight of image features , and obtain the first structural feature of each structured data feature vector and the second structural feature weight , calculate the comprehensive weight of structural features After alignment, delete the redundant image features or structural features that are ranked later, and calculate the comprehensive weight of the image features. Starting from the largest image feature, the image feature is matched with the structural feature corresponding to the largest cosine similarity among the cosine similarities of all structured data features, and the above steps are repeated until all image features are matched.

[0076] The step 4 includes the following contents:

[0077] Step 401: Obtain the first image feature weight of each image feature vector and the second image feature weight , calculate the comprehensive weight of image features :

[0078]

[0079] The comprehensive weight of the corresponding image features The calculation formula is as above.

[0080] Step 402: Obtain the first structural feature of each structured data feature vector and the second structural feature weight , calculate the comprehensive weight of structural features :

[0081]

[0082] The comprehensive weight of the corresponding structural features The calculation formula is as above.

[0083] Step 403: Obtain comprehensive weights of image features and the comprehensive weight of structural features , after sorting them from large to small, the numbers are aligned and the redundant image features or structural features sorted later are deleted.

[0084] Step 404: After the number of image features and structural features is consistent, the comprehensive weight of the image features is calculated. Starting from the largest image feature, the image feature is matched with the structural feature corresponding to the largest cosine similarity among the cosine similarities of all structured data features, and the above steps are repeated until all image features are matched.

[0085] When using, combine the contents in steps 401 to 404:

[0086] Get the first image feature weight of each image feature vector and the second image feature weight , calculate the comprehensive weight of image features , and obtain the first structural feature of each structured data feature vector and the second structural feature weight , calculate the comprehensive weight of structural features After alignment, delete the redundant image features or structural features that are ranked later, and calculate the comprehensive weight of the image features. Starting from the largest image feature, the image feature is matched with the structural feature corresponding to the largest cosine similarity among the cosine similarities of all structured data features. The above steps are repeated until all image features are matched. This not only considers the similarity between features of the same type, but also the similarity between features of different types. It can promote the cross-domain integration between image data and structured data, and provide support for comprehensive data analysis and decision-making.

[0087] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will appreciate that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution.

[0088] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0089] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A method for fusion analysis of image data and structured data based on an artificial intelligence algorithm, characterized by: The steps include: Use various feature extraction algorithms to extract image features of image data, then use dimensionality reduction technology to reduce the high-dimensional feature vector to a low-dimensional space, and calculate the cosine similarity between each image feature vector in the low-dimensional space. , and calculate the first image feature weight of each image feature vector ; Use the NLP library to analyze structured data, extract structured text features, and calculate the cosine similarity between each structured text feature vector in a low-dimensional space. , and calculate the first structural feature weight of the structured text feature vector ; Obtain image features and structured data features, and analyze the cosine similarity between each structured data feature and all image features , calculate the second structural feature weight of each structured data feature ; Obtain image features and structured data features, and analyze the cosine similarity between each image feature and all structured data features , calculate the second image feature weight of each image feature : Get the first image feature weight of each image feature vector and the second image feature weight , calculate the comprehensive weight of image features : Get the first structural feature of each structured data feature vector and the second structural feature weight , calculate the comprehensive weight of structural features : After alignment, the redundant image features or structural features that are ranked later are deleted, and the comprehensive weight of the image features is obtained. Starting from the largest image feature, the image feature is matched with the structural feature corresponding to the largest cosine similarity among the cosine similarities of all structured data features, and the above steps are repeated until all image features are matched.

2. The method for fusion analysis of image data and structured data based on an artificial intelligence algorithm according to claim 1, characterized in that: Get the cosine similarity between each image feature vector , calculate the first image feature weight of each image feature vector : in, a and b Both represent the sequential numbers of image feature vectors, and a =1, 2, ..., n, b =1, 2,…, n.

3. The method for fusion analysis of image data and structured data based on an artificial intelligence algorithm according to claim 1, characterized in that: Get the cosine similarity between each structured text feature vector , calculate the first structural feature weight of the structured text feature vector : in, c and d Both represent the sequential numbers of image feature vectors, and c =1, 2, …, m, d =1, 2,…, m.

4. The method for fusion analysis of image data and structured data based on an artificial intelligence algorithm according to claim 1, characterized in that: Obtain image features and structured data features, and analyze the cosine similarity between each structured data feature and all image features , calculate the second structural feature weight of each structured data feature : The second structural feature weight of each corresponding structured data feature The calculation formula is as above.

5. The method for fusion analysis of image data and structured data based on artificial intelligence algorithm according to claim 1, characterized in that: Get the comprehensive weight of image features and the comprehensive weight of structural features , after sorting them from large to small, the numbers are aligned and the redundant image features or structural features sorted later are deleted.

6. The method for fusion analysis of image data and structured data based on artificial intelligence algorithm according to claim 5, characterized in that: After the number of image features and structural features is consistent, the comprehensive weight of image features is obtained. Starting from the largest image feature, the image feature is matched with the structural feature corresponding to the largest cosine similarity among the cosine similarities of all structured data features, and the above steps are repeated until all image features are matched.

Images