A data analysis and processing method and system based on business license data content

Abstract

The present invention relates to the field of data analysis technology, and in particular to a data analysis and processing method and system based on the data content of a business license. The steps include the following steps: obtaining image data of a business license, performing image preprocessing on the image data, and generating standard image data; performing image block core and edge area division processing on the standard image data to generate a divided image data set; performing data noise reduction processing on the divided image data set to generate noise-reduced image data; performing feature information extraction processing on the noise-reduced image data to generate image text data; obtaining registration text data, performing text comparison detection processing on the registration text data and the image text data, and generating a mall registration error message when the registration text data and the image text data are inconsistent. The present invention recognizes, detects and processes the text of the business license image to achieve rapid inspection of the business license data content and rapid registration of merchants.

Description

Technical Field

[0001] The present invention relates to the field of data analysis technology, and in particular to a data analysis and processing method and system based on business license data content. Background Art

[0002] Data mining is a technology that analyzes each data and searches for its rules from a large amount of data. It mainly includes three steps: data preparation, rule finding and rule expression. Data preparation is to select the required data from relevant data sources and integrate them into a data set for data mining; rule finding is to find the rules contained in the data set by some method; rule expression is to express the found rules in a way that users can understand as much as possible (such as visualization). The tasks of data mining include association analysis, cluster analysis, classification analysis, anomaly analysis, specific group analysis and evolution analysis. With the rapid development of Internet and mobile Internet technology, more and more people are choosing online shopping and services. However, from the perspective of consumers, how to ensure the legitimacy and credibility of online merchants has become an important issue. Therefore, before conducting online transactions, it is necessary to verify and verify the identity of the merchant. The business license is an important certificate for the legal operation of the merchant, which contains the merchant's basic information data and business scope and other key contents. In the past, merchants usually needed to submit paper business licenses for registration, but this method had many disadvantages, such as time-consuming, labor-intensive, and easy to forge. Therefore, a method and system for checking the content of business license data and rapid registration based on data analysis came into being. However, traditional business license recognition content checking and merchant quick registration methods cannot clearly identify text in dynamic blur and static blur in the photographed business license, cannot clearly identify business licenses photographed at an angle and the tilted fonts on the business license, and merchants need to manually fill in the information on the business license when registering. Summary of the invention

[0003] Based on this, the present invention provides a data analysis and processing method and system based on the business license data content to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a data analysis and processing method based on business license data content is provided, characterized in that the method comprises the following steps:

[0005] Step S1: acquiring image data of a business license, and performing image preprocessing on the image data to generate standard image data;

[0006] Step S2: performing image block core and edge area division processing on the standard image data to generate a divided image data set; performing data noise reduction processing on the divided image data set to generate noise-reduced image data;

[0007] Step S3: extracting feature information from the noise-reduced image data to generate image text data;

[0008] Step S4: Obtain the registration text data, perform text comparison detection on the registration text data and the image text data, and generate a mall registration error message when the registration text data and the image text data are inconsistent, and generate merchant quick registration information when the registration text data and the image text data are consistent.

[0009] The present invention obtains the image data of the business license and pre-processes the image, so that the image data can be made clearer and standardized, thereby improving the accuracy of image recognition, removing interference in the image, making the image more concise and clear, accelerating the speed of image processing, automatically processing image data, reducing the cost and time of manual intervention, eliminating uncertain factors in the image, improving the stability and reliability of the system, making the image data more standardized, and reducing the cost and time of subsequent processing; by dividing the image block core and edge area, the image data is divided into multiple blocks to improve the efficiency of image processing, and the image is divided according to the edge and core parts of the image to make image recognition more accurate. By performing noise reduction processing on the edge image data and the core image data, the dynamic and static noise and interference in the image data are removed, the quality and accuracy of image processing are improved, and the image data is made clearer and concise. , speed up the image processing speed and reduce the influence of uncertain factors; extract the text features in the image by performing feature information extraction processing after tilt correction on the image data, improve the accuracy of image recognition, enhance the ability of image processing, improve the efficiency of image processing, automatically process the image data through feature information extraction processing, reduce the cost and time of manual intervention, eliminate the uncertain factors in the image, thereby improving the stability and reliability of the system; by obtaining the registration text data transmitted by the user, through text comparison detection processing, it can accurately and quickly detect whether the registration information is correct, improve the accuracy and reliability of registration, and improve the efficiency and speed of merchant registration, automatically process the registration information, reduce the cost and time of manual intervention, eliminate the uncertain factors in the registration information, reduce the cumbersome steps when the user registers, and improve the user experience and satisfaction. Therefore, the content inspection of business license recognition and the merchant rapid registration method of the present invention can clearly identify the text in dynamic blur and static blur of the photographed business license, can clearly identify the business license photographed by tilt and the tilted font on the business license, and automatically fill in the information on the business license when the merchant registers, saving manual cumbersome steps.

[0010] Preferably, step S1 comprises the following steps:

[0011] Step S11: Acquire image data of the business license;

[0012] Step S12: performing data feasibility filtering processing on the image data to generate filtered image data;

[0013] Step S13: performing image data cleaning processing on the filtered image data to generate cleaned image data;

[0014] Step S14: performing image binarization processing on the cleaned image data to generate binary image data;

[0015] Step S15: Perform image edge detection processing on the binary image to generate standard image data.

[0016] The present invention can remove some unusable or non-compliant image data by performing data feasibility filtering processing on the original image data, thereby improving the efficiency and accuracy of subsequent processing; performing data cleaning processing on the filtered image data can remove some impurities, noise and other factors that interfere with the subsequent processing, thereby improving the robustness of the subsequent processing; performing binarization processing on the cleaned image data can convert the image into a black and white binary image, which is convenient for subsequent processing and recognition; performing edge detection processing on the binary image can detect the edges and contours in the image, thereby generating standard image data, which is convenient for subsequent processing and recognition. The image preprocessing process can improve the accuracy and robustness of image recognition, thereby more reliably performing automated processing and recognition.

[0017] Preferably, the step S2 of dividing the image data set into core pixel block fuzziness data and edge pixel block fuzziness data comprises the following steps:

[0018] Step S21: using Fourier transform to perform image spectrum conversion processing on the standard image data to generate an image block spectrum;

[0019] Step S22: performing image block core and edge area division processing on the standard image data to generate core area pixel block data and edge area pixel block data;

[0020] Step S23: performing image blur calculation processing on the core area pixel block data and the edge area pixel block data based on the image block spectrum diagram, and generating core pixel block blur data and edge pixel block blur data respectively;

[0021] Step S24: performing image core area data noise reduction processing on the core pixel block blur data to generate core noise reduction pixel block data;

[0022] Step S25: performing image edge area data noise reduction processing on the edge pixel block blur data to generate edge noise reduction pixel block data;

[0023] Step S26: performing image pixel block integration on the core noise reduction pixel block data and the edge noise reduction pixel block data, thereby generating noise reduction image data.

[0024] The present invention can obtain the image block spectrum corresponding to the image data by performing Fourier transform and spectrum conversion processing on the standard image data. The image block spectrum can be used to calculate the blur of the image, which is convenient for calculating the noise data of the image and subsequent processing; the standard image data is divided into core and edge areas, which can extract the image feature information more finely; the image blur calculation processing is performed on the core area pixel block data and the edge area pixel block data, which can more accurately judge the clarity and quality of the pixel block, so as to select a suitable noise reduction formula for data noise reduction processing; the core pixel block blur data and the edge pixel block blur data are respectively Noise reduction processing, since the core area pixel block denoising formula has a better dynamic blur noise reduction effect and a more accurate result, and the edge area pixel block denoising formula has a fast processing speed for static blur noise reduction, it saves computing power and time, and reduces the load pressure of the system. Taking into account the application of denoising formula to ensure data accuracy, a large amount of resources can be saved, which can ensure that the data processing effects in different areas are more precise and accurate, and save computing power; image pixel block integration of core denoising pixel block data and edge denoising pixel block data can generate more accurate and complete denoised image data, which is convenient for subsequent processing and recognition, and more reliable for automated processing and recognition.

[0025] Preferably, step S21 includes the following steps:

[0026] Step S211: performing image segmentation processing on the standard image data to generate image block data of the standard image data;

[0027] Step S213: performing a windowing operation on the image block data using a Hamming window function to generate windowed image block data;

[0028] Step S214: Performing image frequency domain mapping processing on the windowed image block data by Fourier transform to generate frequency domain mapping data of the windowed image block data;

[0029] Step S215: Calculate and process the amplitude spectrum and phase spectrum according to the frequency domain mapping data to generate an image block frequency spectrum diagram corresponding to the frequency domain mapping data.

[0030] The present invention can divide a large image into several small image blocks by performing image cutting processing on standard image data, and each image block can be processed separately, thereby reducing the complexity of image processing and improving processing efficiency; performing windowing operation on image block data with Hamming window function can further improve the accuracy and reliability of image processing, reduce spectrum leakage, and thus ensure the accuracy of signals; performing image frequency domain mapping processing on windowed image block data using Fourier transform can improve the accuracy and efficiency of image processing, and convert it into frequency domain signals for better frequency domain analysis and processing; performing amplitude spectrum and phase spectrum calculation processing on frequency domain mapping data can effectively analyze the frequency domain information of the image block, the amplitude spectrum represents the amplitude of each frequency component in the frequency domain signal, and the phase spectrum represents the phase relationship between each frequency component, and by calculating the amplitude spectrum and the phase spectrum, the characteristics of the image block in the frequency domain can be further understood, providing an important basis for subsequent image processing.

[0031] Preferably, step S3 comprises the following steps:

[0032] Step S33: performing image data tilt correction processing on the noise reduction image data to generate corrected image data;

[0033] Step S34: extracting feature information from the corrected image data to generate a character structure feature vector;

[0034] Step S35: Perform image text data generation processing according to the character structure feature vector, thereby generating image text data.

[0035] The present invention performs rotation correction processing on the image data through image data tilt correction, so that the objects in the image are aligned in the horizontal or vertical direction, thereby improving the clarity and readability of the image; performing feature information extraction processing on the corrected image data can help the recognition algorithm to better understand the image data and extract useful feature vectors, thereby improving the recognition accuracy; performing image text data generation processing based on the character structure feature vector can effectively convert the text data in the corrected image data, so that the text information in the image is easier to understand and process, thereby facilitating the extraction and application of information.

[0036] Preferably, the image text data generation process includes character tilt structure feature vector correction process, character sequence prediction process, and character separation and connection process, and step S35 includes the following steps:

[0037] Step S351: using a deep convolutional neural network to perform character tilt structural feature vector correction processing on the character structural feature vector to generate a corrected character feature vector;

[0038] Step S352: using a convolutional neural network model to perform character sequence prediction processing on the corrected character feature vector to generate a sorted character feature vector;

[0039] Step S353: Use a fully convolutional neural network to perform character separation and connection processing on the sorted character feature vectors to generate image text data.

[0040] The present invention uses a deep convolutional neural network model to correct the character tilt structure feature vector, which can effectively correct the recognition error caused by character tilt; the convolutional neural network model performs character sequence prediction processing on the corrected character feature vector, which can help identify and correct problems such as character misalignment and missing; the full convolutional neural network model performs character separation and connection processing on the sorted character feature vector, which can effectively deal with the connectivity problem between characters, thereby generating image text data more accurately.

[0041] Preferably, step S4 comprises the following steps:

[0042] Step S41: Obtaining registration text data;

[0043] Step S42: Perform text comparison detection processing using the registered text data and the image text data. When the registered text data and the image text data are inconsistent, a mall registration error message is generated. When the registered text data and the image text data are consistent, a merchant quick registration message is generated.

[0044] The present invention obtains registration text data transmitted by the user, and uses the registration text data to perform text comparison detection processing with the image text data. When the registration text data and the image text data are inconsistent, a mall registration error message is generated. When the registration text data and the image text data are consistent, a merchant quick registration information is generated. It can accurately and quickly detect whether the registration information is correct, improve the accuracy and reliability of registration, and improve the efficiency and speed of merchant registration. It automatically processes registration information, reduces the cost and time of manual intervention, eliminates uncertain factors in registration information, reduces cumbersome steps when users register, and improves user experience and satisfaction.

[0045] In this specification, a data analysis and processing system based on business license data content is provided, including:

[0046] at least one processor; and,

[0047] a memory communicatively connected to the at least one processor; wherein,

[0048] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the data analysis and processing method based on the business license data content as described in any of the above items.

[0049] The present invention obtains the image data of the business license to check the content. Since the obtained business license may be blurred or direct recognition may cause the input of erroneous information, we need to process the image data. First, the image data is pre-processed to reduce the impact of useless, redundant and abnormal data, and the image data is divided into a core part and an edge part, and different noise reduction processing methods are applied to blur noise points in different areas. Since the core area noise reduction processing has a better dynamic blur noise reduction effect and a more accurate result, the edge area noise reduction processing has a fast processing speed for the static blur noise reduction effect, which saves computing power and time and reduces the load pressure of the system. Comprehensively considering the application of the noise reduction formula to ensure data accuracy, a large amount of resources is saved, and the image data is tilted according to the historical image data, and then the tilted fonts in the image are tilted, so that the content on the business license can be more accurately identified, and the errors and misjudgments that may occur in manual recognition are avoided, the time and workload of manual processing are greatly reduced, and the work efficiency is improved. The relevant laws, regulations and industry standards are more effectively complied with to ensure the compliance of subsequent operations. Therefore, the content inspection of business license identification and the quick merchant registration method of the present invention can clearly identify the text in dynamic blur and static blur in the photographed business license, can clearly identify the business license photographed at an angle and the tilted fonts on the business license, and automatically fill in the information on the business license when the merchant registers, saving tedious manual steps. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A schematic diagram of the steps of a data analysis and processing method based on business license data content of the present invention;

[0051] Figure 2 for Figure 1 Detailed implementation steps of step S1 in FIG.

[0052] Figure 3 for Figure 1 Detailed implementation steps of step S2 in the flowchart;

[0053] Figure 4 for Figure 1 Detailed implementation steps of step S3 in FIG. DETAILED DESCRIPTION

[0054] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are 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 technicians in this field without creative work are within the scope of protection of the present invention.

[0055] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different network and / or processor methods and / or microcontroller methods.

[0056] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0057] The embodiment of the present application provides a data analysis and processing method and system based on the business license data content, wherein the registration text data includes but is not limited to at least one of: merchant registration information text data transmitted by the merchant.

[0058] To achieve this, please refer to Figures 1 to 4 , a data analysis and processing method based on business license data content, the method comprising the following steps:

[0059] Step S1: acquiring image data of a business license, and performing image preprocessing on the image data to generate standard image data;

[0060] Step S2: performing image block core and edge area division processing on the standard image data to generate a divided image data set; performing data noise reduction processing on the divided image data set to generate noise-reduced image data;

[0061] Step S3: extracting feature information from the noise-reduced image data to generate image text data;

[0062] Step S4: Obtain the registration text data, perform text comparison detection on the registration text data and the image text data, and generate a mall registration error message when the registration text data and the image text data are inconsistent, and generate merchant quick registration information when the registration text data and the image text data are consistent.

[0063] The present invention obtains the image data of the business license and pre-processes the image, so that the image data can be made clearer and standardized, thereby improving the accuracy of image recognition, removing interference in the image, making the image more concise and clear, accelerating the speed of image processing, automatically processing image data, reducing the cost and time of manual intervention, eliminating uncertain factors in the image, improving the stability and reliability of the system, making the image data more standardized, and reducing the cost and time of subsequent processing; by dividing the image block core and edge area, the image data is divided into multiple blocks to improve the efficiency of image processing, and the image is divided according to the edge and core parts of the image to make image recognition more accurate. By performing noise reduction processing on the edge image data and the core image data, the dynamic and static noise and interference in the image data are removed, the quality and accuracy of image processing are improved, and the image data is made clearer and concise. , speed up the image processing speed and reduce the influence of uncertain factors; extract the text features in the image by performing feature information extraction processing after tilt correction on the image data, improve the accuracy of image recognition, enhance the ability of image processing, improve the efficiency of image processing, automatically process the image data through feature information extraction processing, reduce the cost and time of manual intervention, eliminate the uncertain factors in the image, thereby improving the stability and reliability of the system; by obtaining the registration text data transmitted by the user, through text comparison detection processing, it can accurately and quickly detect whether the registration information is correct, improve the accuracy and reliability of registration, and improve the efficiency and speed of merchant registration, automatically process the registration information, reduce the cost and time of manual intervention, eliminate the uncertain factors in the registration information, reduce the cumbersome steps when the user registers, and improve the user experience and satisfaction. Therefore, the content inspection of business license recognition and the merchant rapid registration method of the present invention can clearly identify the text in dynamic blur and static blur of the photographed business license, can clearly identify the business license photographed by tilt and the tilted font on the business license, and automatically fill in the information on the business license when the merchant registers, saving manual cumbersome steps.

[0064] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of a data analysis and processing method based on the business license data content of the present invention. In this example, the steps of the data analysis and processing method based on the business license data content include:

[0065] Step S1: acquiring image data of a business license, and performing image preprocessing on the image data to generate standard image data;

[0066] In the embodiment of the present invention, the image data of the business license photographed and uploaded by the user is obtained, and the image data is preprocessed, and the preprocessing includes: data cleaning processing and image binarization processing, so as to generate standard image data.

[0067] Step S2: performing image block core and edge area division processing on the standard image data to generate a divided image data set; performing data noise reduction processing on the divided image data set to generate noise-reduced image data;

[0068] In an embodiment of the present invention, the standard image data is evenly cut according to a fixed size to obtain a number of small image blocks, each of which is equal in size. Then, for each small image block, its central area and edge area are determined. The specific division method can adopt a pixel density distribution method to divide the area with high pixel density into the central area and the area with low pixel density into the edge area. For each small image block, different noise reduction methods are used for different types of pixel blocks to perform data noise reduction processing. For example, if there are fewer characters in the edge area, data noise reduction for the edge image is used. If there are more characters in the center part, data noise reduction for the center image is used, thereby generating noise-reduced image data.

[0069] Step S3: extracting feature information from the noise-reduced image data to generate image text data;

[0070] In the embodiment of the present invention, a convolutional neural network and a recurrent neural network are used to perform text information feature extraction processing on the denoised image data, thereby generating text data of the image.

[0071] Step S4: Obtain the registration text data, perform text comparison detection on the registration text data and the image text data, and generate a mall registration error message when the registration text data and the image text data are inconsistent, and generate merchant quick registration information when the registration text data and the image text data are consistent.

[0072] In an embodiment of the present invention, merchant registration text data input by a user is obtained, and text comparison detection processing is performed on the merchant registration text data and the image text data. When the merchant registration text data and the image text data are inconsistent, information about the mismatching area is generated, and a red warning is set for the mismatching area. The information and the warning are fed back to the terminal and transmitted to the user. When the merchant registration text data and the image text data are consistent, information about the merchant's quick registration success is fed back to the terminal and transmitted to the user.

[0073] Preferably, step S1 comprises the following steps:

[0074] Step S11: Acquire image data of the business license;

[0075] Step S12: performing data feasibility filtering processing on the image data to generate filtered image data;

[0076] Step S13: performing image data cleaning processing on the filtered image data to generate cleaned image data;

[0077] Step S14: performing image binarization processing on the cleaned image data to generate binary image data;

[0078] Step S15: Perform image edge detection processing on the binary image to generate standard image data.

[0079] The present invention can remove some unusable or non-compliant image data by performing data feasibility filtering processing on the original image data, thereby improving the efficiency and accuracy of subsequent processing; performing data cleaning processing on the filtered image data can remove some impurities, noise and other factors that interfere with the subsequent processing, thereby improving the robustness of the subsequent processing; performing binarization processing on the cleaned image data can convert the image into a black and white binary image, which is convenient for subsequent processing and recognition; performing edge detection processing on the binary image can detect the edges and contours in the image, thereby generating standard image data, which is convenient for subsequent processing and recognition. The image preprocessing process can improve the accuracy and robustness of image recognition, thereby more reliably performing automated processing and recognition.

[0080] As an example of the present invention, refer to Figure 2 As shown, Figure 1 Detailed implementation steps of step S1 in the flowchart, in this example, step S1 includes:

[0081] Step S11: Acquire image data of the business license;

[0082] In the embodiment of the present invention, image data of a business license photographed and uploaded by a user is obtained.

[0083] Step S12: performing data feasibility filtering processing on the image data to generate filtered image data;

[0084] In the embodiment of the present invention, the acquired image data is subjected to data feasibility filtering, which includes the following sub-steps:

[0085] Determine the format of the image to determine whether it is a common format, such as JPEG, PNG, etc.

[0086] Determine whether the resolution of the image is within a reasonable range;

[0087] Determine whether the image size is within a reasonable range;

[0088] Determine whether the color mode of the image is grayscale or RGB mode;

[0089] Determine whether the brightness and contrast of the image are within a reasonable range.

[0090] After data feasibility filtering, filtered image data is generated.

[0091] Step S13: performing image data cleaning processing on the filtered image data to generate cleaned image data;

[0092] In the embodiment of the present invention, the filtered image data is subjected to image data cleaning processing, including performing operations such as rotating, cropping, and scaling the image to adjust the image to a standard direction and size to generate cleaned image data.

[0093] Step S14: performing image binarization processing on the cleaned image data to generate binary image data;

[0094] In the embodiment of the present invention, the cleaned image data is subjected to image binarization processing to convert the color or grayscale image into a binary image. The binarization processing may use a common threshold segmentation algorithm, such as the OTSU algorithm, an adaptive threshold algorithm, and the like.

[0095] Step S15: Perform image edge detection processing on the binary image to generate standard image data.

[0096] In the embodiment of the present invention, image edge detection processing is performed on the binary image using common edge detection algorithms, such as the Canny algorithm, the Sobel algorithm, etc., so as to generate standard image data.

[0097] Preferably, the step S2 of dividing the image data set into core pixel block fuzziness data and edge pixel block fuzziness data comprises the following steps:

[0098] Step S21: using Fourier transform to perform image spectrum conversion processing on the standard image data to generate an image block spectrum;

[0099] Step S22: performing image block core and edge area division processing on the standard image data to generate core area pixel block data and edge area pixel block data;

[0100] Step S23: performing image blur calculation processing on the core area pixel block data and the edge area pixel block data based on the image block spectrum diagram, and generating core pixel block blur data and edge pixel block blur data respectively;

[0101] Step S24: performing image core area data noise reduction processing on the core pixel block blur data to generate core noise reduction pixel block data;

[0102] Step S25: performing image edge area data noise reduction processing on the edge pixel block blur data to generate edge noise reduction pixel block data;

[0103] Step S26: performing image pixel block integration on the core noise reduction pixel block data and the edge noise reduction pixel block data, thereby generating noise reduction image data.

[0104] The present invention can obtain the image block spectrum corresponding to the image data by performing Fourier transform and spectrum conversion processing on the standard image data. The image block spectrum can be used to calculate the blur of the image, which is convenient for calculating the noise data of the image and subsequent processing; the standard image data is divided into core and edge areas, which can extract the image feature information more finely; the image blur calculation processing is performed on the core area pixel block data and the edge area pixel block data, which can more accurately judge the clarity and quality of the pixel block, so as to select a suitable noise reduction formula for data noise reduction processing; the core pixel block blur data and the edge pixel block blur data are respectively Noise reduction processing, since the core area pixel block denoising formula has a better dynamic blur noise reduction effect and a more accurate result, and the edge area pixel block denoising formula has a fast processing speed for static blur noise reduction, it saves computing power and time, and reduces the load pressure of the system. Taking into account the application of denoising formula to ensure data accuracy, a large amount of resources can be saved, which can ensure that the data processing effects in different areas are more precise and accurate, and save computing power; image pixel block integration of core denoising pixel block data and edge denoising pixel block data can generate more accurate and complete denoised image data, which is convenient for subsequent processing and recognition, and more reliable for automated processing and recognition.

[0105] As an example of the present invention, refer to Figure 3 As shown, Figure 1 Detailed implementation steps of step S2 in the embodiment are shown in the flowchart. In this embodiment, step S2 includes:

[0106] Step S21: using Fourier transform to perform image spectrum conversion processing on the standard image data to generate an image block spectrum;

[0107] In the embodiment of the present invention, Fourier transform is used to perform image spectrum conversion processing on standard image data to generate an image block spectrum. Fourier transform can decompose a signal (including an image) into sine waves of different frequencies to obtain its frequency domain expression.

[0108] Step S22: performing image block core and edge area division processing on the standard image data to generate core area pixel block data and edge area pixel block data;

[0109] In the embodiment of the present invention, the standard image data is processed by dividing the image block core and edge area to generate core area pixel block data and edge area pixel block data. The standard image is divided into multiple pixel blocks according to a preset block size, and each pixel block is classified into a core area pixel block and an edge area pixel block.

[0110] Step S23: performing image blur calculation processing on the core area pixel block data and the edge area pixel block data based on the image block spectrum diagram, and generating core pixel block blur data and edge pixel block blur data respectively;

[0111] In the embodiment of the present invention, image blur calculation processing is performed on the core area pixel block data and the edge area pixel block data based on the image block spectrum diagram to generate core pixel block blur data and edge pixel block blur data respectively. Blur data of the core area pixel block and the edge area pixel block are obtained by calculating the energy distribution of the core area pixel block and the edge area pixel block in the frequency domain.

[0112] Step S24: performing image core area data noise reduction processing on the core pixel block blur data to generate core noise reduction pixel block data;

[0113] In an embodiment of the present invention, the core pixel block blur data is subjected to denoising processing for the core area data of the image. There is more text data in the core area, which causes more dynamic blur of the text. It is necessary to perform denoising processing using a specific core area pixel block denoising formula to eliminate dynamic blur noise data so that the text data can be restored to be clear and accurate, thereby generating core denoised pixel block data.

[0114] Step S25: performing image edge area data noise reduction processing on the edge pixel block blur data to generate edge noise reduction pixel block data;

[0115] In the embodiment of the present invention, the edge pixel block blur data is subjected to image edge area data denoising processing. There is less text data in the core area, resulting in less dynamic blur of the text, but more static blur noise caused by the image color. A specific edge area pixel block denoising formula is used to perform denoising processing. The amount of data processed is smaller and the processing speed is faster, so that the image style and color are restored clearly and accurately, thereby reducing the denoising pixel block data.

[0116] Step S26: performing image pixel block integration on the core noise reduction pixel block data and the edge noise reduction pixel block data, thereby generating noise reduction image data.

[0117] In the embodiment of the present invention, the core denoised pixel block data and the edge denoised pixel block data are merged according to the corresponding sequence before separation to form the final denoised image data.

[0118] Preferably, step S21 includes the following steps:

[0119] Step S211: performing image segmentation processing on the standard image data to generate image block data of the standard image data;

[0120] Step S213: performing a windowing operation on the image block data using a Hamming window function to generate windowed image block data;

[0121] Step S214: Performing image frequency domain mapping processing on the windowed image block data by Fourier transform to generate frequency domain mapping data of the windowed image block data;

[0122] Step S215: Calculate and process the amplitude spectrum and phase spectrum according to the frequency domain mapping data to generate an image block frequency spectrum diagram corresponding to the frequency domain mapping data.

[0123] The present invention can divide a large image into several small image blocks by performing image cutting processing on standard image data, and each image block can be processed separately, thereby reducing the complexity of image processing and improving processing efficiency; performing windowing operation on image block data with Hamming window function can further improve the accuracy and reliability of image processing, reduce spectrum leakage, and thus ensure the accuracy of signals; performing image frequency domain mapping processing on windowed image block data using Fourier transform can improve the accuracy and efficiency of image processing, and convert it into frequency domain signals for better frequency domain analysis and processing; performing amplitude spectrum and phase spectrum calculation processing on frequency domain mapping data can effectively analyze the frequency domain information of the image block, the amplitude spectrum represents the amplitude of each frequency component in the frequency domain signal, and the phase spectrum represents the phase relationship between each frequency component, and by calculating the amplitude spectrum and the phase spectrum, the characteristics of the image block in the frequency domain can be further understood, providing an important basis for subsequent image processing.

[0124] In an embodiment of the present invention, the standard image data is cut and processed, and the image is cut into a number of image blocks, each of which is a specific pixel in size. During the cutting process, a sliding window can be used, and the window is moved to the next position by sliding a step each time. The image block data is windowed using a Hamming window function, and the Hamming window function is applied to each image block data to generate windowed image block data. The image frequency domain mapping process is performed on the windowed image block data using a Fourier transform formula to generate frequency domain mapping data of the windowed image block data, and the amplitude spectrum and phase spectrum are calculated and processed according to the frequency domain mapping data, and the amplitude spectrum and phase spectrum of each frequency domain mapping data are calculated, and the amplitude spectrum and phase spectrum are converted, so as to obtain the image block spectrum diagram corresponding to the frequency domain mapping data.

[0125] Preferably, step S3 comprises the following steps:

[0126] Step S33: performing image data tilt correction processing on the noise reduction image data to generate corrected image data;

[0127] Step S34: extracting feature information from the corrected image data to generate a character structure feature vector;

[0128] Step S35: Perform image text data generation processing according to the character structure feature vector, thereby generating image text data.

[0129] The present invention performs rotation correction processing on the image data through image data tilt correction, so that the objects in the image are aligned in the horizontal or vertical direction, thereby improving the clarity and readability of the image; performing feature information extraction processing on the corrected image data can help the recognition algorithm to better understand the image data and extract useful feature vectors, thereby improving the recognition accuracy; performing image text data generation processing based on the character structure feature vector can effectively convert the text data in the corrected image data, so that the text information in the image is easier to understand and process, thereby facilitating the extraction and application of information.

[0130] As an example of the present invention, refer to Figure 4 As shown, Figure 1 Detailed implementation steps of step S3 in the flowchart, in this example, step S3 includes:

[0131] Step S33: performing image data tilt correction processing on the noise reduction image data to generate corrected image data;

[0132] In the embodiment of the present invention, the image placement position of the image tilt correction template is used to perform placement position tilt correction processing on the noise reduction image data corresponding to the image placement position to generate corrected image data.

[0133] Step S34: extracting feature information from the corrected image data to generate a character structure feature vector;

[0134] In the embodiment of the present invention, a connected domain analysis is performed on the image data to obtain the connected regions of the characters, and the character structure feature vector is extracted according to the features such as the shape and size of the connected regions.

[0135] Step S35: Perform image text data generation processing according to the character structure feature vector, thereby generating image text data.

[0136] In the embodiment of the present invention, character statistics are performed based on the character structure feature vector, and the convolutional neural network model is used to sort and organize the characters to generate image text data corresponding to the image data.

[0137] Preferably, the image text data generation process includes character tilt structure feature vector correction process, character sequence prediction process, and character separation and connection process, and step S35 includes the following steps:

[0138] Step S351: using a deep convolutional neural network to perform character tilt structural feature vector correction processing on the character structural feature vector to generate a corrected character feature vector;

[0139] Step S352: using a convolutional neural network model to perform character sequence prediction processing on the corrected character feature vector to generate a sorted character feature vector;

[0140] Step S353: Use a fully convolutional neural network to perform character separation and connection processing on the sorted character feature vectors to generate image text data.

[0141] The present invention uses a deep convolutional neural network model to correct the character tilt structure feature vector, which can effectively correct the recognition error caused by character tilt; the convolutional neural network model performs character sequence prediction processing on the corrected character feature vector, which can help identify and correct problems such as character misalignment and missing; the full convolutional neural network model performs character separation and connection processing on the sorted character feature vector, which can effectively deal with the connectivity problem between characters, thereby generating image text data more accurately.

[0142] In an embodiment of the present invention, a deep convolutional neural network model is trained using a character structure feature vector, a portion of the character structure feature vector is taken as a training set for deep convolutional neural network training, and the input character structure feature vector is corrected according to the trained deep convolutional neural network model to generate a corrected character feature vector. A convolutional neural network model is trained using a corrected character feature vector, a portion of the corrected character feature vector is taken as a training set for convolutional neural network model training, and the input corrected character feature vector is sorted according to the trained convolutional neural network model to generate a sorted character feature vector. A full convolutional neural network model is trained using a sorted character feature vector, a portion of the sorted character feature vector is taken as a training set for full convolutional neural network model training, and character separation and connection processing is performed on the input sorted character feature vector according to the trained full convolutional neural network model, thereby generating image text data, and three models are used to process the character structure feature vector, thereby converting the text data in the image into text data for storage.

[0143] Preferably, step S4 comprises the following steps:

[0144] Step S41: Obtaining registration text data;

[0145] Step S42: Perform text comparison detection processing using the registered text data and the image text data. When the registered text data and the image text data are inconsistent, a mall registration error message is generated. When the registered text data and the image text data are consistent, a merchant quick registration message is generated.

[0146] The present invention obtains registration text data transmitted by the user, and uses the registration text data to perform text comparison detection processing with the image text data. When the registration text data and the image text data are inconsistent, a mall registration error message is generated. When the registration text data and the image text data are consistent, a merchant quick registration information is generated. It can accurately and quickly detect whether the registration information is correct, improve the accuracy and reliability of registration, and improve the efficiency and speed of merchant registration. It automatically processes registration information, reduces the cost and time of manual intervention, eliminates uncertain factors in registration information, reduces cumbersome steps when users register, and improves user experience and satisfaction.

[0147] In an embodiment of the present invention, merchant registration text data input by a user is obtained, and text comparison detection processing is performed on the merchant registration text data and the image text data. When the merchant registration text data and the image text data are inconsistent, information about the mismatching area is generated, and a red warning is set for the mismatching area. The information and the warning are fed back to the terminal and transmitted to the user. When the merchant registration text data and the image text data are consistent, information about the merchant's quick registration success is fed back to the terminal and transmitted to the user.

[0148] In this specification, a data analysis and processing system based on business license data content is provided, including:

[0149] at least one processor; and,

[0150] a memory communicatively connected to the at least one processor; wherein,

[0151] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the data analysis and processing method based on the business license data content as described in any of the above items.

[0152] The present invention obtains the image data of the business license to check the content. Since the obtained business license may be blurred or direct recognition may cause the input of erroneous information, we need to process the image data. First, the image data is pre-processed to reduce the impact of useless, redundant and abnormal data, and the image data is divided into a core part and an edge part, and different noise reduction processing methods are applied to blur noise points in different areas. Since the core area noise reduction processing has a better dynamic blur noise reduction effect and a more accurate result, the edge area noise reduction processing has a fast processing speed for the static blur noise reduction effect, which saves computing power and time and reduces the load pressure of the system. Comprehensively considering the application of the noise reduction formula to ensure data accuracy, a large amount of resources is saved, and the image data is tilted according to the historical image data, and then the tilted fonts in the image are tilted, so that the content on the business license can be more accurately identified, and the errors and misjudgments that may occur in manual recognition are avoided, the time and workload of manual processing are greatly reduced, and the work efficiency is improved. The relevant laws, regulations and industry standards are more effectively complied with to ensure the compliance of subsequent operations. Therefore, the content inspection of business license identification and the quick merchant registration method of the present invention can clearly identify the text in dynamic blur and static blur in the photographed business license, can clearly identify the business license photographed at an angle and the tilted fonts on the business license, and automatically fill in the information on the business license when the merchant registers, saving tedious manual steps.

Claims

1. A data analysis and processing method based on business license data content, characterized in that: The method comprises the following steps: Step S1: acquiring image data of a business license, and performing image preprocessing on the image data to generate standard image data; Step S2: performing image block core and edge area division processing on the standard image data, thereby generating a divided image data set, wherein the divided image data set includes core area pixel block data and edge area pixel block data; performing data denoising processing on the divided image data set, generating denoised image data; wherein step S2 includes the following steps: Step S21: using Fourier transform to perform image spectrum conversion processing on the standard image data to generate an image block spectrum; Step S22: performing image block core and edge area division processing on the standard image data, dividing the standard image into a plurality of pixel blocks according to a preset block size, classifying each pixel block, and generating core area pixel block data and edge area pixel block data; Step S23: performing image blur calculation processing on the core area pixel block data and the edge area pixel block data based on the image block spectrum diagram, and generating core pixel block blur data and edge pixel block blur data respectively; Step S24: using the core pixel block fuzziness data to perform image core area data denoising processing on the core area pixel block data to generate core denoised pixel block data; Step S25: using the edge pixel block fuzziness data to perform image edge area data denoising processing on the edge area pixel block data to generate edge denoised pixel block data; Step S26: performing image pixel block integration on the core noise reduction pixel block data and the edge noise reduction pixel block data, thereby generating noise reduction image data; Step S3: extracting feature information from the noise-reduced image data to generate image text data; wherein step S3 includes the following steps: Step S33: performing image data tilt correction processing on the noise reduction image data to generate corrected image data; Step S34: extracting feature information from the corrected image data to generate a character structure feature vector; Step S35: performing image text data generation processing according to the character structure feature vector, thereby generating image text data; wherein step S35 includes the following steps: Step S351: using a deep convolutional neural network to perform character tilt structural feature vector correction processing on the character structural feature vector to generate a corrected character feature vector; Step S352: using a convolutional neural network model to perform character sequence prediction processing on the corrected character feature vector to generate a sorted character feature vector; Step S353: using a fully convolutional neural network to perform character separation and connection processing on the sorted character feature vectors, thereby generating image text data; Step S4: Obtain the registration text data, perform text comparison detection on the registration text data and the image text data, and generate a mall registration error message when the registration text data and the image text data are inconsistent, and generate merchant quick registration information when the registration text data and the image text data are consistent.

2. The data analysis and processing method based on the business license data content according to claim 1 is characterized in that: Step S1 includes the following steps: Step S11: Acquire image data of the business license; Step S12: performing data feasibility filtering processing on the image data to generate filtered image data; Step S13: performing image data cleaning processing on the filtered image data to generate cleaned image data; Step S14: performing image binarization processing on the cleaned image data to generate binary image data; Step S15: Perform image edge detection processing on the binary image to generate standard image data.

3. The data analysis and processing method based on the business license data content according to claim 1 is characterized in that: Step S21 includes the following steps: Step S211: performing image segmentation processing on the standard image data to generate image block data of the standard image data; Step S213: performing a windowing operation on the image block data using a Hamming window function to generate windowed image block data; Step S214: Performing image frequency domain mapping processing on the windowed image block data by Fourier transform to generate frequency domain mapping data of the windowed image block data; Step S215: Calculate and process the amplitude spectrum and phase spectrum according to the frequency domain mapping data to generate an image block frequency spectrum diagram corresponding to the frequency domain mapping data.

4. The data analysis and processing method based on the business license data content according to claim 1 is characterized in that: Step S4 includes the following steps: Step S41: Obtaining registration text data; Step S42: Perform text comparison detection processing using the registered text data and the image text data. When the registered text data and the image text data are inconsistent, a mall registration error message is generated. When the registered text data and the image text data are consistent, a merchant quick registration message is generated.

5. A data analysis and processing system based on business license data content, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the data analysis and processing method based on the business license data content as described in any one of claims 1 to 4.

Images