A network security data privacy protection management system

Abstract

The present invention discloses a network security data privacy protection management system, which belongs to the field of privacy protection technology, and specifically includes: obtaining an instruction of whether a user chooses to protect the privacy of an image location; when the user chooses to protect the location, obtaining the location information of the image uploaded by the user, and dividing the uploaded image into several sets according to the location information; identifying all images in any set, and identifying characteristic objects and characteristic texts from the image; dividing images containing the same characteristic object into the same subset, and for any subset, selecting the characteristic object that occupies the largest space of the entire image and marking it as a landmark object, retrieving a scene address similar to the landmark object from the network, randomly selecting any address as a virtual address of the subset image, and replacing the original location information, and replacing or erasing the characteristic text based on the virtual address; the present invention realizes the protection of user geographic information.

Description

Technical Field

[0001] The present invention relates to the field of privacy protection technology, and in particular to a network security data privacy protection management system. Background Art

[0002] With the popularization of the Internet, more and more users choose to share their life moments through online platforms. However, this also brings the problem of data privacy leakage. In particular, the location information and feature text in the image may expose sensitive information such as the user's living habits and work location. Therefore, how to protect user privacy while allowing users to enjoy the fun of sharing has become an urgent problem to be solved.

[0003] Some existing solutions mainly focus on blurring images or desensitizing location information. However, these methods protect user privacy while affecting the clarity and usability of images. In addition, these methods cannot prevent the actual location of users from being inferred through buildings, landscapes, and some direct text in the image.

[0004] Therefore, there is a need for a management system that can effectively protect user data privacy. The system can process images uploaded by users, replace or erase sensitive information in the images, while maintaining the clarity and usability of the images. Summary of the invention

[0005] The purpose of the present invention is to provide a network security data privacy protection management system to solve the following technical problems:

[0006] Some existing solutions mainly focus on blurring images or desensitizing location information. However, these methods protect user privacy while affecting the clarity and usability of images. In addition, these methods cannot prevent the actual location of users from being inferred through buildings, landscapes, and some direct text in the image.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] A network security data privacy protection management system, comprising:

[0009] An instruction collection module, used to obtain an instruction of whether the user chooses to protect the privacy of the image location, wherein the instruction includes a public location and a protected location;

[0010] A location division module, used to obtain location information of images uploaded by the user after the user selects a protected location, and divide the uploaded images into several sets according to the location information;

[0011] A scene recognition module is used to recognize all images in any set, identify characteristic objects from the images, including but not limited to buildings, natural landscapes, roads and signboards, and recognize characteristic text in the signboards, including but not limited to place names, signboard texts and company names;

[0012] The position protection module is used to divide images containing the same feature object into the same subset. For any subset, the feature object that occupies the largest area of ​​the entire image is selected and marked as a landmark object. The scene address similar to the landmark object is retrieved from the network, and any address is randomly selected as the virtual address of the subset image, and the original position information is replaced, and the feature text is replaced or erased based on the virtual address.

[0013] As a further solution of the present invention: the accuracy of the location information zoning level is set by the user.

[0014] As a further solution of the present invention: the recognition process of the characteristic object is:

[0015] A feature object recognition model is constructed based on a pre-trained CNN network. The feature object recognition model is based on labeled supervised learning and an edge detection algorithm to identify man-made objects in an image. The man-made objects include buildings, roads, and signboards. Corner points of man-made objects are identified through corner point detection. Man-made objects and natural landscapes are distinguished based on texture analysis. The image is divided into several regions, each region corresponds to a feature object, and any region is input into the feature object recognition model to output a recognition result.

[0016] As a further solution of the present invention: the recognition process of the characteristic text includes correcting the sign:

[0017] The signboards recording place names, signboard texts and company names are identified by frame selection, the signboard image is cropped, and the signboard image is converted into a binary image. The area with a larger pixel ratio in the binary image is marked as the background area, and the area with a smaller pixel ratio is marked as the text area. The text area is cut in sequence along the long side of the signboard in the short side direction to obtain a plurality of sub-areas, and all the single sub-areas are enlarged outward from the center until adjacent text areas touch each other to obtain a strip area, and the strip area is subjected to median filtering to generate a corresponding straight line equation, and the average slope of the straight line equation is calculated. The signboard image is angle corrected according to the average slope to obtain a signboard corrected image.

[0018] As a further solution of the present invention: the process of identifying characteristic characters is:

[0019] Scan the corrected image of the sign along the long side of the sign, obtain the column where the ratio of the number of pixels in the first text area to the number of pixels in the current column is greater than a preset threshold, which is the starting pixel column S of the text, continue to scan the corrected image of the sign until the column where the ratio of the number of pixels in the text area to the number of pixels in the current column is less than the preset threshold, which is the ending pixel column O of the first text, calculate the pixel width W between O and S, where W is the pixel width of a single text, continue to scan the corrected image of the license plate, identify text with a pixel ratio greater than the preset threshold and a pixel width of W, segment each text area to obtain several text images, recognize the text image based on OCR, obtain the sign text, perform natural language processing on the sign text, and identify place name entities, brand entities and corporate entities in the text, which are feature text.

[0020] As a further solution of the present invention: the process of replacing or erasing the characteristic text is:

[0021] The place name in the characteristic text is replaced with a virtual address. If the characteristic text does not include a place name, the characteristic text is erased.

[0022] As a further solution of the present invention: the process of replacing or erasing the characteristic text also includes:

[0023] When the name of the sign is identified as a chain enterprise or a store, only the form of the sign is changed, and the characteristic text remains unchanged. The form includes but is not limited to color, shape and pattern.

[0024] Beneficial effects of the present invention:

[0025] The present invention can effectively perform privacy protection processing on images uploaded by users through modules such as instruction collection, location division, scene recognition, and location protection, including replacing or erasing characteristic text and virtualizing geographic location information, while allowing users to customize privacy protection levels and intelligently identify and process image content. While ensuring user privacy, this system also maintains image quality, provides personalized privacy protection, enhances user experience, and improves processing efficiency and accuracy through intelligent technology. It is a safe and convenient privacy protection solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The present invention will be further described below in conjunction with the accompanying drawings.

[0027] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0029] See also Figure 1 As shown, the present invention is a network security data privacy protection management system, comprising:

[0030] An instruction collection module, used to obtain an instruction of whether the user chooses to protect the privacy of the image location, wherein the instruction includes a public location and a protected location;

[0031] A location division module, used to obtain location information of images uploaded by the user after the user selects a protected location, and divide the uploaded images into several sets according to the location information;

[0032] A scene recognition module is used to recognize all images in any set, identify characteristic objects from the images, including but not limited to buildings, natural landscapes, roads and signboards, and recognize characteristic text in the signboards, including but not limited to place names, signboard texts and company names;

[0033] The position protection module is used to divide images containing the same feature object into the same subset. For any subset, the feature object that occupies the largest area of ​​the entire image is selected and marked as a landmark object. The scene address similar to the landmark object is retrieved from the network, and any address is randomly selected as the virtual address of the subset image, and the original position information is replaced, and the feature text is replaced or erased based on the virtual address.

[0034] The present invention analyzes images uploaded by users, identifies, replaces or erases location information and feature text, effectively preventing leakage of personal location information and related sensitive data, thereby enhancing user data privacy protection.

[0035] While protecting privacy, the system also maintains the clarity and usability of the image, allowing users to enjoy the fun of sharing without having to worry about their personal privacy being violated, thus improving the user experience.

[0036] By randomly selecting similar scene addresses to replace the original location information, the difficulty of inferring the original location information is increased, and the security of privacy protection is improved.

[0037] In another preferred embodiment of the present invention, the accuracy of the location information zoning level is set by the user.

[0038] Allowing users to set the accuracy of location information and whether to protect specific feature objects provides a high degree of flexibility and personalized options.

[0039] In another preferred embodiment of the present invention, the recognition process of the characteristic object is:

[0040] A feature object recognition model is constructed based on a pre-trained CNN network. The feature object recognition model is based on labeled supervised learning and an edge detection algorithm to identify man-made objects in an image. The man-made objects include buildings, roads, and signboards. Corner points of man-made objects are identified through corner point detection. Man-made objects and natural landscapes are distinguished based on texture analysis. The image is divided into several regions, each region corresponds to a feature object, and any region is input into the feature object recognition model to output a recognition result.

[0041] The pre-trained CNN network and supervised learning model are used to automatically identify feature objects and text in images, reducing manual intervention and improving processing efficiency and accuracy.

[0042] In another preferred embodiment of the present invention, the recognition process of the characteristic text includes correcting the sign:

[0043] The signboards recording place names, signboard texts and company names are identified by frame selection, the signboard image is cropped, and the signboard image is converted into a binary image. The area with a larger pixel ratio in the binary image is marked as the background area, and the area with a smaller pixel ratio is marked as the text area. The text area is cut in sequence along the long side of the signboard in the short side direction to obtain a plurality of sub-areas, and all the single sub-areas are enlarged outward from the center until adjacent text areas touch each other to obtain a strip area, and the strip area is subjected to median filtering to generate a corresponding straight line equation, and the average slope of the straight line equation is calculated. The signboard image is angle corrected according to the average slope to obtain a signboard corrected image.

[0044] In another preferred embodiment of the present invention, the process of identifying characteristic characters is as follows:

[0045] Scan the corrected image of the sign along the long side of the sign, obtain the column where the ratio of the number of pixels in the first text area to the number of pixels in the current column is greater than a preset threshold, which is the starting pixel column S of the text, continue to scan the corrected image of the sign until the column where the ratio of the number of pixels in the text area to the number of pixels in the current column is less than the preset threshold, which is the ending pixel column O of the first text, calculate the pixel width W between O and S, where W is the pixel width of a single text, continue to scan the corrected image of the license plate, identify text with a pixel ratio greater than the preset threshold and a pixel width of W, segment each text area to obtain several text images, recognize the text image based on OCR, obtain the sign text, perform natural language processing on the sign text, and identify place name entities, brand entities and corporate entities in the text, which are feature text.

[0046] The system can identify and process place name entities, brand entities, corporate entities, etc. in the text, further enriching the dimension of privacy protection.

[0047] In another preferred embodiment of the present invention, the process of replacing or erasing the characteristic text is as follows:

[0048] The place name in the characteristic text is replaced with a virtual address. If the characteristic text does not include a place name, the characteristic text is erased.

[0049] In another preferred embodiment of the present invention, the process of replacing or erasing the characteristic text further includes:

[0050] When the name of the sign is identified as a chain enterprise or a store, only the form of the sign is changed, and the characteristic text remains unchanged. The form includes but is not limited to color, shape and pattern.

[0051] The system design takes into account a variety of scenarios. For example, when identifying a signboard of a chain enterprise or store, only the form can be changed without changing the characteristic text, thus retaining a certain degree of information authenticity while taking into account privacy protection.

[0052] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A network security data privacy protection management system, characterized in that: include: An instruction collection module, used to obtain an instruction of whether the user chooses to protect the privacy of the image location, wherein the instruction includes a public location and a protected location; A location division module, used to obtain location information of images uploaded by the user after the user selects a protected location, and divide the uploaded images into several sets according to the location information; A scene recognition module is used to recognize all images in any set, identify characteristic objects from the images, the characteristic objects include buildings, natural landscapes, roads and signboards, and recognize characteristic texts in the signboards, the characteristic texts include place names, signboard texts and company names; A position protection module is used to divide images containing the same characteristic object into the same subset. For any subset, the characteristic object that occupies the largest area of ​​the entire image is selected and marked as a landmark object. A scene address similar to the landmark object is retrieved from the network, and any address is randomly selected as a virtual address of the subset image, and the original position information is replaced, and the characteristic text is replaced or erased based on the virtual address; The process of replacing or erasing feature text is as follows: The place name in the characteristic text is replaced with a virtual address. If the characteristic text does not include a place name, the characteristic text is erased.

2. A network security data privacy protection management system according to claim 1, characterized in that: The accuracy of the zone level of the location information is set by the user.

3. A network security data privacy protection management system according to claim 1, characterized in that: The recognition process of the feature object is as follows: A feature object recognition model is constructed based on a pre-trained CNN network. The feature object recognition model is based on labeled supervised learning and an edge detection algorithm to identify man-made objects in an image. The man-made objects include buildings, roads, and signboards. Corner points of man-made objects are identified through corner point detection. Man-made objects and natural landscapes are distinguished based on texture analysis. The image is divided into several regions, each region corresponds to a feature object, and any region is input into the feature object recognition model to output a recognition result.

4. A network security data privacy protection management system according to claim 3, characterized in that: The recognition process of characteristic characters includes correcting the sign: The signboards recording place names, signboard texts and company names are identified by frame selection, the signboard image is cropped, and the signboard image is converted into a binary image. The area with a larger pixel ratio in the binary image is marked as the background area, and the area with a smaller pixel ratio is marked as the text area. The text area is cut in sequence along the long side of the signboard in the short side direction to obtain a plurality of sub-areas, and all the single sub-areas are enlarged outward from the center until adjacent text areas touch each other to obtain a strip area, and the strip area is subjected to median filtering to generate a corresponding straight line equation, and the average slope of the straight line equation is calculated. The signboard image is angle corrected according to the average slope to obtain a signboard corrected image.

5. A network security data privacy protection management system according to claim 4, characterized in that: The process of identifying characteristic characters is as follows: Scan the corrected image of the sign along the long side of the sign to obtain the column where the ratio of the number of pixels in the first text area to the number of pixels in the current column is greater than a preset threshold, which is the starting pixel column S of the text. Continue scanning the corrected image of the sign until a column where the ratio of the number of pixels in the text area to the number of pixels in the current column is less than a preset threshold is identified, which is the ending pixel column O of the first text. Calculate the pixel width W between O and S, where W is the pixel width of a single text. Continue scanning the corrected image of the sign to identify text with a pixel ratio greater than a preset threshold and a pixel width of W. Segment each text area to obtain several text images, recognize the text image based on OCR to obtain the sign text, perform natural language processing on the sign text, and identify place name entities, brand entities, and corporate entities in the text, which are feature text.

6. A network security data privacy protection management system according to claim 1, characterized in that: The process of replacing or erasing feature text also includes: When the sign is identified as a chain enterprise or a store, only the form of the sign is changed, and the characteristic text remains unchanged. The form includes color, shape and pattern.

Images