Image desensitization method, device, computer device and storage medium

Abstract

This application relates to an image desensitization method, apparatus, computer device, and storage medium. The method includes: identifying a target object in a first image; matching the target object with multiple preset objects; if no match is found between the target object and the multiple preset objects, determining a first desensitization strategy based on the region type of the camera's monitoring area; and desensitizing the corresponding region of the target object in the first image based on the first desensitization strategy to obtain a second image. This method, applied to cameras, can avoid privacy leaks caused by network transmission at the image generation source, while also making the first desensitization strategy based on the region type of the camera's monitoring area more flexible, thereby significantly improving the security of personal privacy.

Description

Technical Field

[0001] This application relates to the field of image desensitization technology, and in particular to an image desensitization method, apparatus, computer equipment, and storage medium. Background Technology

[0002] With the widespread deployment of IoT cameras, there is a need for real-time analysis of video content in urban surveillance and enterprise production activities without compromising personal privacy.

[0003] Existing image desensitization methods typically involve retrieving a depth model from a cloud server to blur faces, license plates, or other content that needs desensitization. This method may inadvertently leak personal privacy due to network latency. Summary of the Invention

[0004] Therefore, it is necessary to provide an image desensitization method, apparatus, computer equipment, computer-readable storage medium, and computer program product that can improve the security of personal privacy in response to the above-mentioned technical problems.

[0005] In a first aspect, this application provides an image desensitization method applied to a camera, comprising:

[0006] Identify the target object in the first image;

[0007] Match the target object with multiple preset objects;

[0008] If the target object fails to match with multiple preset objects, the first desensitization strategy is determined based on the region type of the camera's monitoring area.

[0009] Based on the first desensitization strategy, the region corresponding to the target object in the first image is desensitized to obtain the second image.

[0010] In an exemplary embodiment, the first desensitization strategy is determined based on the region type of the camera-monitored area, including:

[0011] The target blur intensity and target kernel size are determined based on the region type of the monitored area by the camera;

[0012] The above-mentioned desensitization processing of the region corresponding to the target object in the first image based on the first desensitization strategy includes:

[0013] Based on a blur kernel with target blur intensity and target kernel size, the region corresponding to the target object in the first image is desensitized.

[0014] In one exemplary embodiment, the area types are office area, production area, public area, and remote monitoring area;

[0015] The above-mentioned area type based on the camera's monitoring area determines the target blur intensity and target kernel size, including:

[0016] When the area monitored by the camera is an office area, the target blur intensity is determined to be the first blur intensity and the target kernel size is determined to be the first size;

[0017] When the area monitored by the camera is a production area, the target blur intensity is determined to be the second blur intensity and the target kernel size is the second size.

[0018] When the area monitored by the camera is a public area, the target blur intensity is determined to be the third blur intensity and the target kernel size is the third size;

[0019] When the area type of the camera's monitoring area is a long-distance monitoring area, the target blur intensity is determined to be the fourth blur intensity and the target kernel size is the fourth size;

[0020] Among them, the third fuzziness intensity is greater than the first fuzziness intensity, the first fuzziness intensity is greater than the fourth fuzziness intensity, and the fourth fuzziness intensity is greater than the second fuzziness intensity; the third size is greater than the first size, the first size is greater than the fourth size, and the fourth size is greater than the second size.

[0021] In an exemplary embodiment, when the area type monitored by the camera is a production area, determining the target blur intensity as a second blur intensity and the target kernel size as a second size includes:

[0022] When the area monitored by the camera is a production area, the target blur intensity is determined to be 0 and the target kernel size is determined to be 0.

[0023] In one exemplary embodiment, the method further includes:

[0024] Determine the timestamp of the first image;

[0025] The above-mentioned matching of the target object with multiple preset objects includes:

[0026] If the timestamp meets the preset time conditions, the target object is matched with multiple preset objects.

[0027] In an exemplary embodiment, when the timestamp does not meet a preset time condition, the above-mentioned determination of the target object in the first image includes:

[0028] Identify the target object in the first image;

[0029] Based on the second desensitization strategy, the region corresponding to the target object in the first image is desensitized to obtain the third image.

[0030] In one exemplary embodiment, the method further includes:

[0031] The second image is sent to the target terminal so that the target terminal can display the second image.

[0032] Secondly, this application also provides an image desensitization device for use in a camera, comprising:

[0033] The first determining module is used to determine the target object in the first image;

[0034] The matching module is used to match a target object with multiple preset objects;

[0035] The second determining module is used to determine the first desensitization strategy based on the area type of the camera's monitoring area when the target object fails to match multiple preset objects.

[0036] The desensitization module is used to perform desensitization processing on the region corresponding to the target object in the first image based on the first desensitization strategy to obtain the second image.

[0037] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement some or all of the steps described in any method of the first aspect of the embodiments of this application.

[0038] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements some or all of the steps described in any method of the first aspect of the embodiments of this application.

[0039] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements some or all of the steps described in any method of the first aspect of the embodiments of this application.

[0040] The aforementioned image desensitization method, apparatus, computer equipment, computer-readable storage medium, and computer program product, being applied to cameras, ensure that, on the one hand, the first image that has not yet undergone desensitization processing will not be uploaded to the server or sent to the terminal, thus preventing the leakage of personal privacy caused by network transmission from the source of image generation; on the other hand, the first desensitization strategy adopted for desensitizing the first image is determined based on the area type of the camera's monitoring area, making the desensitization processing more flexible. Therefore, the image desensitization method provided in this application embodiment can significantly improve the security of personal privacy. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is an application environment diagram of the image desensitization method in one embodiment;

[0043] Figure 2 This is a flowchart illustrating an image desensitization method in one embodiment;

[0044] Figure 3 This is a structural block diagram of an image desensitization device in one embodiment;

[0045] Figure 4 This is an internal structural diagram of a computer device in one embodiment;

[0046] Figure 5 This is a diagram of the internal structure of a computer device in another embodiment. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0048] The image desensitization method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, camera 102 communicates with server 104 via a network, and camera 102 can communicate with terminal 106 via wired or wireless means. Terminal 106 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. Head-mounted devices can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. Server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0049] In one exemplary embodiment, such as Figure 2As shown, an image desensitization method is provided, which is applied to... Figure 1 The camera in the process includes steps 202 to 208. Wherein:

[0050] Step 202: Determine the target object in the first image.

[0051] Among them, the camera can be a network camera (IP camera, IPC camera), which refers to a camera that transmits images and videos over a network.

[0052] Optionally, the image desensitization method provided in this application embodiment can be implemented by the control module in the camera.

[0053] The first image refers to the original image captured by the camera in real time and which has not yet undergone desensitization processing.

[0054] The target object refers to the object that exists in the first image and is required to be matched with multiple preset objects to determine whether it needs to be desensitized.

[0055] Optionally, the target object can be a face or a license plate.

[0056] In an exemplary embodiment, determining the target object in the first image includes: driving the target object in the first image through an object detection model.

[0057] In another exemplary embodiment, when the target object is a face or a ticket, the object detection model includes a face detection model and / or a license plate detection model. The above-mentioned determination of the target object in the first image includes: determining the target object in the first image through the face detection model and / or the license plate detection model.

[0058] The object detection model can be the YOLO Nano model.

[0059] Step 204: Match the target object with multiple preset objects.

[0060] Among them, multiple preset objects refer to multiple authorized objects that have been pre-entered into the authorized object database.

[0061] Optionally, an authorization object database storing multiple preset objects can be set on a server or on the camera itself.

[0062] Optionally, the preset target can be authorized personnel or authorized vehicles within the camera's monitoring area. Authorized personnel are those with legal access to the monitored area, and authorized vehicles are those with legal access to the monitored area.

[0063] Optionally, users can add, delete, or change multiple preset objects on the target terminal and synchronize them to the camera to ensure that the camera can match the target object in the first image with the latest multiple preset objects, thus avoiding the leakage of personal privacy due to failure to update the preset objects in time.

[0064] For example, authorized personnel in the monitored area may include employees, registered visitors, etc., and authorized vehicles in the monitored area may include internal official vehicles, employee registered vehicles, registered visitor vehicles, etc.

[0065] Optionally, the multiple preset objects can be obtained by the camera from the server, in order to match the target object with the multiple preset objects in the camera's view. Therefore, if the authorized object database storing the multiple preset objects is located on the server, then the multiple preset objects are obtained by the camera from the server.

[0066] In an exemplary embodiment, the above-described matching of the target object with multiple preset objects includes: determining the target object features of the target object; matching the target object features with the preset object features of each preset object among the multiple preset objects to obtain the feature similarity between the target object features and each preset object feature; determining that the target object and the multiple preset objects have not been successfully matched if the feature similarity of multiple features is less than a similarity threshold; and determining that the target object has been successfully matched with one of the multiple preset objects if there is a corresponding preset object feature with a feature similarity greater than or equal to the similarity threshold.

[0067] In another exemplary embodiment, the above-described determination of the target object features includes: determining the target object features of the target object through a feature extraction model.

[0068] The target object features and the preset object features can be presented as feature vectors, respectively.

[0069] Optionally, the feature extraction model can be the MobileViT model.

[0070] Optionally, the feature extraction model includes a face feature extraction model and / or a license plate feature extraction model.

[0071] Optionally, feature similarity can be the cosine similarity or Euclidean distance between the features of the target object and the features of the corresponding preset object.

[0072] It is easy to understand that since the image desensitization method provided in this application embodiment is applied to a camera, the object detection model and the feature extraction model are both set in the camera.

[0073] Step 206: If the target object fails to match with multiple preset objects, determine the first desensitization strategy based on the area type of the camera's monitoring area.

[0074] If the target object fails to match multiple preset objects, it means that the target object is an unauthorized object, that is, an unauthorized person or an unauthorized vehicle. In other words, in this case, the target object needs to be de-identified to protect its personal privacy.

[0075] The area type of the monitored area refers to the area divided by the monitored area corresponding to the installation location of the camera. Optionally, different area types can correspond to different security level requirements and thus have different levels of desensitization strategies.

[0076] The first desensitization strategy refers to the desensitization processing rules determined by the area type of the monitoring area corresponding to the camera, used to protect the personal privacy of unauthorized target objects.

[0077] Optionally, the first desensitization strategy can be set by the user in advance based on the area type of the monitoring area corresponding to the camera. Based on this, the user can change the specific content of the first desensitization strategy on the target terminal to ensure that the desensitization processing of the first image by the camera meets the user's personal privacy protection needs.

[0078] If the target object is successfully matched with one of the preset objects among multiple preset objects, the target object is identified as an authorized object. Since the authorized object is a legitimate monitoring and management object belonging to the camera's monitoring area, the target object does not need to be desensitized in order to facilitate the tracing of the authorized object's behavior.

[0079] Optionally, if the target object fails to match with multiple preset objects, the camera generates a notification event based on the event that the target object fails to match with multiple preset objects, and sends the notification event to the server, so that the server sends the first desensitization strategy to the camera.

[0080] Step 208: Based on the first desensitization strategy, the region corresponding to the target object in the first image is desensitized to obtain the second image.

[0081] The region corresponding to the target object in the first image refers to the portion of the first image that includes the target object and requires desensitization processing. Optionally, the region corresponding to the target object in the first image can be presented in the form of a circular or rectangular detection box.

[0082] Desensitization processing of the target object region in the first image refers to modifying the pixel content of the target object region in the first image so that viewers cannot identify the specific content of the target object from the second image, thus protecting personal privacy. The desensitization process can be understood as an image processing process.

[0083] Optionally, the desensitization processing of the region corresponding to the target object in the first image can be performed by performing Gaussian blur processing or mosaic blur processing on the region corresponding to the target object in the first image.

[0084] The second image refers to a processed image captured in real time by the camera that has undergone anonymization. Uploading the second image to the server or sending it to the terminal will not leak the personal privacy of unauthorized individuals.

[0085] In a way that is easy to understand, the areas outside the target object area in the first image may not be desensitized, so that the second image can satisfy both the target object's need for personal privacy protection and the viewer's need to view the specific content of the surveillance footage.

[0086] Optionally, the first image will only be stored locally on the camera and will not be uploaded to the server or sent to the terminal. Only the second image can be uploaded to the server or sent to the terminal. Based on this, it can be ensured that the first image, which has not been anonymized, will not be spread outside the camera and cause leakage of the target's personal privacy. That is to say, only the second image, which has been anonymized and ensures that the target's personal privacy will not be leaked, can be spread outside the camera.

[0087] In the above-mentioned image desensitization method, since it is applied to a camera, on the one hand, the first image that has not yet been desensitized will not be uploaded to the server or sent to the terminal, thus avoiding the problem of personal privacy leakage caused by network transmission from the source of image generation. On the other hand, the first desensitization strategy adopted for desensitizing the first image is determined based on the area type of the camera's monitoring area, making the desensitization process more flexible. Therefore, the image desensitization method provided in this application embodiment can significantly improve the security of personal privacy.

[0088] In an exemplary embodiment, the first desensitization strategy is determined based on the region type of the camera-monitored area, including:

[0089] The target blur intensity and target kernel size are determined based on the region type of the monitored area by the camera;

[0090] The above-mentioned desensitization processing of the region corresponding to the target object in the first image based on the first desensitization strategy includes:

[0091] Based on a blur kernel with target blur intensity and target kernel size, the region corresponding to the target object in the first image is desensitized.

[0092] The target blur intensity characterizes the degree of image detail suppression after desensitization processing of the target object region in the first image. There is a positive correlation between the target blur intensity and the desensitization processing intensity.

[0093] The target kernel size refers to the size of the blur kernel, used to characterize the extent of the desensitization processing applied to the corresponding region of the target object in the first image. There is a positive correlation between the target kernel size and the intensity of the desensitization processing.

[0094] A blur kernel is an image processing operator used to blur the corresponding region of a target object in a first image to achieve desensitization. It is easy to understand that the blurring effect of a blur kernel is determined by the target blur intensity and the kernel size.

[0095] In this embodiment, different regions of the camera's monitoring area correspond to different blur intensities and kernel sizes. Therefore, the target blur intensity and target kernel size are determined based on the region type of the camera's monitoring area. Then, based on the blur kernel with the target blur intensity and target kernel size, the region corresponding to the target object in the first image is desensitized. Thus, hierarchical and refined desensitization processing can be achieved for different regions of the monitoring area. This satisfies the real-time monitoring requirements of the camera while meeting the different security level requirements of different region types, thereby significantly improving the rationality and security of personal privacy protection.

[0096] In one exemplary embodiment, the area types are office area, production area, public area, and remote monitoring area;

[0097] The above-mentioned area type based on the camera's monitoring area determines the target blur intensity and target kernel size, including:

[0098] When the area monitored by the camera is an office area, the target blur intensity is determined to be the first blur intensity and the target kernel size is determined to be the first size;

[0099] When the area monitored by the camera is a production area, the target blur intensity is determined to be the second blur intensity and the target kernel size is the second size.

[0100] When the area monitored by the camera is a public area, the target blur intensity is determined to be the third blur intensity and the target kernel size is the third size;

[0101] When the area type of the camera's monitoring area is a long-distance monitoring area, the target blur intensity is determined to be the fourth blur intensity and the target kernel size is the fourth size;

[0102] Among them, the third fuzziness intensity is greater than the first fuzziness intensity, the first fuzziness intensity is greater than the fourth fuzziness intensity, and the fourth fuzziness intensity is greater than the second fuzziness intensity; the third size is greater than the first size, the first size is greater than the fourth size, and the fourth size is greater than the second size.

[0103] Office area refers to a dedicated area within an enterprise used for internal management activities. For example, internal management activities can include routine office work, holding meetings, and other activities related to internal management.

[0104] A production area refers to a dedicated area within a company used for production activities. For example, production activities can include equipment operation, technological processes, and other activities related to the company's production activities.

[0105] Public areas refer to open areas that are accessible to the public, have high personnel mobility, and are frequently visited by unauthorized individuals. For example, public areas can be entrance / exit areas, lobby areas, or other areas where unauthorized individuals can also enter.

[0106] A long-distance monitoring area refers to a monitoring area where the camera is installed at a considerable distance from the monitored object, resulting in a small area of ​​the monitored object within the camera's monitoring range and lower resolution. For example, a long-distance monitoring area could be the area surrounding a business park, a distant road, or a distant entrance / exit. In essence, a long-distance monitoring area is an area where the area of ​​the monitored object within the camera's monitoring range is less than a certain threshold; the monitored object is typically a person or vehicle.

[0107] Since the second fuzziness intensity is greater than the first fuzziness intensity, the first fuzziness intensity is greater than the fourth fuzziness intensity, the fourth fuzziness intensity is greater than the third fuzziness intensity, and the second size is greater than the first size, the first size is greater than the fourth size, and the fourth size is greater than the third size, therefore, among the four area types of office area, production area, public area, and remote monitoring area, the desensitization processing intensity of the production area is the lowest, while the desensitization processing intensity of the public area is the highest.

[0108] Optionally, the first, second, third, and fourth dimensions can be determined based on the image dimensions of the first image. It is readily understood that the first, second, third, and fourth dimensions are positively correlated with the image dimensions of the first image.

[0109] For example, if the image size of the first image is 320×320, then the first size can be 15×15, the second size can be 5×5 or 0, the third size can be 25×25, and the fourth size can be 9×9.

[0110] Since every action related to a company's production activities can affect the company's production and operation results, in order to ensure the traceability of production activities, the relevant operations of every person in the production area should be clearly visible to achieve behavioral traceability. Therefore, the desensitization treatment intensity in the production area is minimal.

[0111] Because the monitored object occupies a small area within the camera's monitoring range and has a low resolution in the long-distance monitoring area, it is difficult to fully identify the monitored object. Therefore, the desensitization intensity for long-distance monitoring areas can be relatively low, that is, it can be only slightly higher than the desensitization intensity for production areas among the four area types.

[0112] Since there may be a large number of unauthorized individuals in public areas, in order to prevent the personal privacy of a large number of unauthorized individuals from being uploaded to the server or distributed to the terminal and thus leaked, the desensitization process for public areas should be the most stringent among the four types of areas: office area, production area, public area, and remote monitoring area.

[0113] In this embodiment, the area types are office area, production area, public area, and remote monitoring area. Regarding the desensitization intensity, the desensitization intensity for public area is greater than that for office area, which is greater than that for remote monitoring area, and the remote monitoring area is greater than that for production area. This ensures that the desensitization intensity for the first image is precisely matched to the security level requirements, personnel flow characteristics, and monitoring scenario requirements of different area types. Based on this, not only can excessive desensitization prevent the monitored content from losing its viewing value, but insufficient desensitization intensity can also avoid the risk of personal privacy leakage. In other words, it significantly improves the security of personal privacy.

[0114] In an exemplary embodiment, when the area type monitored by the camera is a production area, determining the target blur intensity as a second blur intensity and the target kernel size as a second size includes:

[0115] When the area monitored by the camera is a production area, the target blur intensity is determined to be 0 and the target kernel size is determined to be 0.

[0116] The target blur intensity is 0 and the target kernel size is 0, which can be understood as not performing desensitization processing on the corresponding area of ​​the target object in the first image. That is, neither authorized nor unauthorized objects will be desensitized and will be fully presented in the second image.

[0117] In this embodiment, since every action of every person in the production area needs to be fully recorded for subsequent traceability, the target blur intensity and target kernel size are determined to be 0 when the area type of the camera's monitoring area is a production area. That is, when the area type of the camera's monitoring area is a production area, no desensitization processing is performed on the area corresponding to the target object in the first image. This ensures that the monitoring content of the production area can be fully presented, so that the monitoring content of the production area has traceability. Based on this, the flexibility of the desensitization processing is improved.

[0118] In one exemplary embodiment, the method further includes:

[0119] Determine the timestamp of the first image;

[0120] The above-mentioned matching of the target object with multiple preset objects includes:

[0121] If the timestamp meets the preset time conditions, the target object is matched with multiple preset objects.

[0122] The timestamp of the first image refers to the information used to characterize the time node when the camera captured the first image.

[0123] Optionally, the timestamp of the first image can indicate in which year, month, day, hour, minute, and second the first image was collected; that is, the timestamp of the first image can be specific to the second.

[0124] Preset time conditions refer to pre-defined time conditions used to determine whether the timestamp of the first image falls within a preset authorized time period, thus determining whether the first image needs to undergo desensitization processing. Preset time conditions can be understood as the authorized time period set by the user.

[0125] The authorized time period refers to the legal monitoring period during which authorized objects in the first image do not require desensitization processing. Conversely, the unauthorized time period refers to the time period outside the authorized time period, that is, the time period during which target objects appearing in the first image, whether authorized or unauthorized, require desensitization processing.

[0126] The timestamp meets the preset time conditions, which means that the acquisition time of the first image is within the authorized time period. During this time period, the behavior of the authorized object needs to be monitored. Therefore, if the target object is an authorized object, no desensitization processing is required. However, if the target object is an unauthorized object and does not match multiple preset objects, then the privacy of the unauthorized object needs to be protected and desensitization processing is required.

[0127] For example, the authorized time period can be a working time period; the unauthorized time period can be a non-working time period. Based on this, for example, the preset time condition can be a working time period within each day.

[0128] In this embodiment, when the timestamp meets the preset time condition, the target object is matched with multiple preset objects. Then, when the target object fails to match with any of the preset objects, a first desensitization strategy is determined based on the area type of the camera's monitoring area. Thus, through the dual matching mechanism of preset objects and preset time periods, the first desensitization strategy can be finely adapted to the time scene and the monitoring area scene respectively, so that the first desensitization strategy has high monitoring targeting. While improving the security of personal privacy, it also improves the targeting and flexibility of the desensitization processing of the first image.

[0129] In an exemplary embodiment, when the timestamp does not meet a preset time condition, the above-mentioned determination of the target object in the first image includes:

[0130] Identify the target object in the first image;

[0131] Based on the second desensitization strategy, the region corresponding to the target object in the first image is desensitized to obtain the third image.

[0132] In another exemplary embodiment, the third image is obtained by performing desensitization processing on the region corresponding to the target object in the first image based on the second desensitization strategy, including:

[0133] Based on a blur kernel with a fifth blur intensity and a fifth size, the region corresponding to the target object in the first image is desensitized to obtain the third image.

[0134] The timestamp not meeting the preset time conditions means that the acquisition time of the first image is an unauthorized time period. Therefore, the target objects in the first image, whether authorized or unauthorized, need to be desensitized to protect the personal privacy of the target objects.

[0135] Optionally, the fifth fuzziness intensity can be the same as the first fuzziness intensity or the second fuzziness intensity.

[0136] Alternatively, the fifth dimension may be the same as the first or second dimension.

[0137] The third image refers to the image captured in real time by the camera and after the desensitization process has been completed. To put it simply, in the second image, only the target objects that did not successfully match multiple preset objects were desensitized, while in the third image, all the target objects detected by the object detection model were desensitized.

[0138] In this embodiment, if the timestamp of the first image does not meet the preset time condition, then as long as the target object is identified in the first image, the corresponding area of ​​the target object in the first image needs to be desensitized indiscriminately. This ensures that the personal privacy of any target object will not be leaked during the unauthorized time period without the authorization of any target object, thereby improving the security of personal privacy.

[0139] In one exemplary embodiment, the method further includes:

[0140] The second image is sent to the target terminal so that the target terminal can display the second image.

[0141] The target terminal can be a terminal used by the camera administrator or a person who needs to view the camera's monitoring content.

[0142] Optionally, the target terminal can be connected to a display device, so that after the second image is sent to the target terminal, the target terminal can display the second image through the display device.

[0143] Optionally, the target terminal may send a viewing request to the camera, and the camera, in response to the viewing request, sends the second image, which has been de-identified, to the target terminal.

[0144] In this embodiment, the second image, which has undergone desensitization processing, is sent to the target terminal so that the target terminal displays the second image. This ensures that the personal privacy of the unauthorized object in the first image is not leaked to the user using the target terminal, thus ensuring the security of the personal privacy of the unauthorized object.

[0145] In an exemplary embodiment, the above-described desensitization process, based on the first desensitization strategy, desensitizes the region corresponding to the target object in the first image to obtain the second image, including:

[0146] Based on the first desensitization strategy and the camera's device status information, the corresponding area of ​​the target object in the first image is desensitized to obtain the second image.

[0147] Specifically, if the camera's device status information indicates that the number of frames processed per second is greater than or equal to a first frame rate threshold, then deformable blurring or multi-target tracking blurring is performed on the region corresponding to the target object in the first image based on the first desensitization strategy; if the camera's device status information indicates that the number of frames processed per second is greater than or equal to a second frame rate threshold but less than the first frame rate threshold, then Gaussian blurring is performed on the region corresponding to the target object in the first image based on the first desensitization strategy; if the camera's device status information indicates that the number of frames processed per second is less than the second frame rate threshold, then mosaic blurring is performed on the region corresponding to the target object in the first image based on the first desensitization strategy.

[0148] Frames per second (fps) refers to the number of image frames that a camera can de-identify within a unit of time. fps characterizes a camera's real-time processing capability, and it's easy to understand that there is a positive correlation between fps and real-time processing capability.

[0149] Deformable blurring refers to a refined desensitization method that adaptively adjusts the blurring shape based on the shape of the target object in the monitoring screen.

[0150] Multi-target tracking blurring refers to a method for tracking target objects to achieve stable desensitization processing of continuous image frames.

[0151] In one exemplary embodiment, the method further includes:

[0152] Based on the number of target objects in the second image, the first desensitization strategy, and the camera's device status, a log event is generated; the log event is then uploaded to the server or sent to the target terminal.

[0153] In another exemplary embodiment, the above-mentioned uploading of log events to the server or sending them to the target terminal includes: uploading log events to the server or sending them to the target terminal every preset interval.

[0154] The camera's device status refers to the device status of the camera when it captures the first image. Optionally, the camera's device status may include the camera's load rate.

[0155] In this embodiment, the log events generated by the camera can be periodically uploaded to the server or sent to the terminal, thereby meeting the requirements of relevant laws and regulations for minimizing monitoring data and tracing responsibility.

[0156] In an exemplary embodiment, the image desensitization method provided in this application can be implemented using the following program code:

[0157] Initialization: Load the object detection model and the authorized object database storing multiple preset objects; obtain multiple first desensitization strategies; process each frame of the first image in a loop:

[0158] Step 1: data_frame ← Collect the first image;

[0159] Step 2: objects ← object detection model.detect(data_frame); matches the target object with multiple preset objects;

[0160] Step 3: authorized_map ← {}; For each obj in objects: If obj.category∈{face, license plate}: Match the target object with multiple preset objects using the feature extraction model; If the target object has a successfully matched preset object and the timestamp meets the preset time condition: authorized_map[obj] ← True; Otherwise: authorized_map[obj] ← False;

[0161] Step 4: For each obj in objects: If authorized_map[obj] == False: Based on the first desensitization strategy, perform desensitization processing on obj.bounding_box, which is the region corresponding to the target object in the first image;

[0162] Step 5: Send the second image to the target terminal so that the target terminal can display the second image;

[0163] Step 6: Based on the number of target objects in the second image, the first desensitization strategy, and the camera's device status, generate log events; upload the log events to the server or send them to the target terminal;

[0164] Step 7: Periodically check and update the first desensitization strategy to reload it.

[0165] In an exemplary embodiment, after matching the target object with multiple preset objects, the method further includes:

[0166] Determine the confidence level of the matching results between the target object and multiple preset objects;

[0167] Based on the first desensitization strategy, the corresponding region of the target object in the first image is desensitized to obtain the second image, including:

[0168] If the confidence level of the matching result is less than or equal to the confidence threshold, a first adjustment coefficient is generated;

[0169] The target blur intensity and target kernel size are adjusted based on the first adjustment coefficient to obtain the adjusted target blur intensity and the adjusted target kernel size; the adjusted target blur intensity is greater than the unadjusted target blur intensity, and the adjusted target kernel size is greater than the unadjusted target kernel size.

[0170] Based on a blur kernel with adjusted target blur intensity and adjusted target kernel size, the region corresponding to the target object in the first image is desensitized to obtain the second image.

[0171] The confidence level of the matching results between the target object and multiple preset objects can be determined by the object detection model.

[0172] Optionally, the first adjustment coefficient is determined based on the confidence difference between the confidence level of the matching result and the confidence threshold, and there is a positive correlation between the first adjustment coefficient and the confidence difference. The larger the first adjustment coefficient, the larger the difference in fuzziness intensity between the adjusted target fuzziness intensity and the target fuzziness intensity before adjustment. Similarly, the larger the first adjustment coefficient, the larger the size difference between the adjusted target kernel size and the target kernel size before adjustment.

[0173] Optionally, the adjusted target fuzziness intensity = first adjustment coefficient × target fuzziness intensity, and the adjusted target kernel size = first adjustment coefficient × target kernel size; the first adjustment coefficient is greater than 1.

[0174] In another exemplary embodiment, the above-mentioned adjustment of the target blur intensity and target kernel size based on the first adjustment coefficient to obtain the adjusted target blur intensity and adjusted target kernel size includes:

[0175] A second adjustment coefficient is generated based on the area type of the monitored area by the camera;

[0176] Based on the first adjustment coefficient and the second adjustment coefficient, the target fuzziness intensity and the target kernel size are adjusted respectively to obtain the adjusted target fuzziness intensity and the adjusted target kernel size.

[0177] This can be understood as follows: based on the area type of the monitored area by the camera, not only will the target blur intensity and target kernel size be determined, but a second adjustment coefficient will also be generated. Only when the confidence level of the matching result is less than or equal to the confidence level threshold will the second adjustment coefficient work together with the first adjustment coefficient to adjust the target blur intensity and target kernel size respectively. If the confidence level of the matching result is greater than the confidence level threshold, the second adjustment coefficient will not adjust the target blur intensity and target kernel size respectively.

[0178] The second adjustment coefficient is greater when the area type is a public area than when the area type is an office area; the second adjustment coefficient is greater when the area type is an office area than when the area type is a remote monitoring area; and the second adjustment coefficient is greater when the area type is a remote monitoring area than when the area type is a production area.

[0179] The larger the second adjustment coefficient, the greater the difference in fuzziness intensity between the adjusted target fuzziness intensity and the original target fuzziness intensity. Similarly, the larger the second adjustment coefficient, the greater the difference in size between the adjusted target kernel size and the original target kernel size.

[0180] Optionally, the adjusted target fuzziness intensity = first adjustment coefficient × second adjustment coefficient × target fuzziness intensity, and the adjusted target kernel size = first adjustment coefficient × second adjustment coefficient × target kernel size; the second adjustment coefficient is greater than 1.

[0181] In this embodiment, when the confidence level of the matching result is less than or equal to the confidence level threshold, it indicates that although the target object and multiple preset objects have not been successfully matched, this matching result cannot guarantee that the target object is necessarily not a preset object, that is, it cannot guarantee that the target object is definitely not an authorized object. It can be understood that the target object at this time is an unknown object whose identity cannot be confirmed and is completely unfamiliar to the user using the camera. Therefore, in order to avoid the misjudgment of the matching result from causing leakage of the target object's personal privacy, the target blur intensity and the target kernel size should be strengthened and adjusted respectively. That is, based on the blur kernel with the adjusted target blur intensity and the adjusted target kernel size, the corresponding area of ​​the target object in the first image is subjected to a higher desensitization intensity desensitization processing to improve the security of personal privacy.

[0182] In one exemplary embodiment, the above image desensitization method includes:

[0183] Identify the target object in the first image;

[0184] The target object is matched with multiple preset objects to determine the matching results and the confidence level of the matching results.

[0185] If the target object fails to match with multiple preset objects, the target blur intensity, target kernel size, and second adjustment coefficient are determined based on the region type of the camera's monitoring area.

[0186] If the confidence level of the matching result is less than or equal to the confidence threshold, a first adjustment coefficient is generated;

[0187] Based on the first adjustment coefficient and the second adjustment coefficient, the target blur intensity and the target kernel size are adjusted respectively to obtain the adjusted target blur intensity and the adjusted target kernel size; the adjusted target blur intensity is greater than the unadjusted target blur intensity, and the adjusted target kernel size is greater than the unadjusted target kernel size;

[0188] Based on a blur kernel with adjusted target blur intensity and adjusted target kernel size, the region corresponding to the target object in the first image is desensitized to obtain the second image.

[0189] In this context, it's easy to understand that when the area type is an office area, the adjusted target fuzz intensity is the adjusted first fuzz intensity, and the adjusted target kernel size is the adjusted first size. The same applies to other area types, so it won't be elaborated further here. Similarly, the adjusted first fuzz intensity is greater than the unadjusted first fuzz intensity, and the adjusted first size is greater than the unadjusted first size. The same applies to the third fuzz intensity, fourth fuzz intensity, third size, and fourth size, so it won't be elaborated further here. The adjusted second fuzz intensity is greater than or equal to the unadjusted second fuzz intensity, and the adjusted second size is greater than or equal to the unadjusted second size.

[0190] It is easy to understand that the adjusted third blur intensity is greater than the adjusted first blur intensity, the adjusted first blur intensity is greater than the adjusted fourth blur intensity, and the adjusted fourth blur intensity is greater than the adjusted second blur intensity; the adjusted third size is greater than the adjusted first size, the adjusted first size is greater than the adjusted fourth size, and the adjusted fourth size is greater than the adjusted second size.

[0191] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0192] Based on the same inventive concept, this application also provides an image desensitization apparatus for implementing the image desensitization method described above. The solution provided by this apparatus is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more image desensitization apparatus embodiments provided below can be found in the limitations of the image desensitization method described above, and will not be repeated here.

[0193] In one exemplary embodiment, such as Figure 3 As shown, an image desensitization device is provided for use with a camera, including: a first determining module 302, a matching module 304, a second determining module 306, and a desensitization module 308, wherein:

[0194] The first determining module 302 is used to determine the target object in the first image.

[0195] Matching module 304 is used to match the target object with multiple preset objects.

[0196] The second determining module 306 is used to determine the first desensitization strategy based on the area type of the camera's monitoring area when the target object fails to match with multiple preset objects.

[0197] The desensitization module 308 is used to perform desensitization processing on the region corresponding to the target object in the first image based on the first desensitization strategy to obtain the second image.

[0198] In an exemplary embodiment, regarding the determination of the first desensitization strategy based on the region type of the monitored area of ​​the camera, the second determination module 306 is specifically used to determine the target blur intensity and the target kernel size based on the region type of the monitored area of ​​the camera; regarding the desensitization processing of the region corresponding to the target object in the first image based on the first desensitization strategy, the desensitization module 308 is specifically used to perform desensitization processing on the region corresponding to the target object in the first image based on the blur kernel having the target blur intensity and the target kernel size.

[0199] In an exemplary embodiment, the area types are office area, production area, public area, and remote monitoring area. Regarding determining the target blur intensity and target kernel size based on the area type of the camera's monitoring area, the second determining module 306 is specifically used to determine the target blur intensity as a first blur intensity and the target kernel size as a first size when the area type of the camera's monitoring area is an office area; determine the target blur intensity as a second blur intensity and the target kernel size as a second size when the area type of the camera's monitoring area is a production area; determine the target blur intensity as a third blur intensity and the target kernel size as a third size when the area type of the camera's monitoring area is a public area; and determine the target blur intensity as a fourth blur intensity and the target kernel size as a fourth size when the area type of the camera's monitoring area is a remote monitoring area. Wherein, the third blur intensity is greater than the first blur intensity, the first blur intensity is greater than the fourth blur intensity, and the fourth blur intensity is greater than the second blur intensity; the third size is greater than the first size, the first size is greater than the fourth size, and the fourth size is greater than the second size.

[0200] In an exemplary embodiment, when the area type of the camera's monitoring area is a production area, the second determining module 306 is specifically used to determine that the target blur intensity is 0 and the target kernel size is 0 when the area type of the camera's monitoring area is a production area.

[0201] In an exemplary embodiment, the first determining module 302 is further configured to determine the timestamp of the first image; in terms of matching the target object with multiple preset objects, the second determining module 306 is specifically configured to match the target object with multiple preset objects when the timestamp meets the preset time condition.

[0202] In an exemplary embodiment, when the timestamp does not meet the preset time condition, the first determining module 302 is specifically used to determine the target object in the first image; based on the second desensitization strategy, the corresponding area of ​​the target object in the first image is desensitized to obtain the third image.

[0203] In an exemplary embodiment, the desensitization module 308 is further configured to send the second image to the target terminal so that the target terminal displays the second image.

[0204] In one exemplary embodiment, the image desensitization device described above includes:

[0205] The first determining module 302 is used to determine the target object in the first image;

[0206] The matching module 304 is used to match the target object with multiple preset objects, determine the matching result of the target object with the multiple preset objects and the confidence level of the matching result of the target object with the multiple preset objects;

[0207] The second determining module 306 is used to determine the target blur intensity, target kernel size, and second adjustment coefficient based on the region type of the camera's monitoring area when the target object fails to match multiple preset objects.

[0208] The desensitization module 308 is used to generate a first adjustment coefficient when the confidence level of the matching result is less than or equal to the confidence level threshold.

[0209] The desensitization module 308 is also used to adjust the target blur intensity and the target kernel size based on the first adjustment coefficient and the second adjustment coefficient, respectively, to obtain the adjusted target blur intensity and the adjusted target kernel size; the adjusted target blur intensity is greater than the unadjusted target blur intensity, and the adjusted target kernel size is greater than the unadjusted target kernel size;

[0210] The desensitization module 308 is also used to perform desensitization processing on the region corresponding to the target object in the first image based on a blur kernel with adjusted target blur intensity and adjusted target kernel size, to obtain a second image.

[0211] Each module in the aforementioned image desensitization device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0212] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 4As shown, this computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data related to image desensitization methods. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements an image desensitization method.

[0213] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 5 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When executed by the processor, the computer program implements an image desensitization method. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0214] Those skilled in the art will understand that Figure 5The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0215] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0216] In the above computer device embodiments, any of the computer devices can be a camera.

[0217] In one exemplary embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above-described method embodiments.

[0218] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0219] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0220] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0221] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0222] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. An image desensitization method, characterized in that, Applied to cameras, including: Identify the target object in the first image; the first image will only be stored locally on the camera and will not be uploaded to the server or sent to the terminal; Match the target object with multiple preset objects; If the target object fails to match any of the preset objects, the target blur intensity and target kernel size are determined based on the region type of the camera's monitoring area; the region type includes office area, production area, public area, and long-distance monitoring area. Based on a blur kernel having the target blur intensity and the target kernel size, the region corresponding to the target object in the first image is desensitized to obtain a second image; The determination of target blur intensity and target kernel size based on the region type of the monitored area of ​​the camera includes: If the area monitored by the camera is an office area, the target blur intensity is determined to be a first blur intensity and the target kernel size is a first size; When the area type of the monitoring area of ​​the camera is the production area, the target blur intensity is determined to be the second blur intensity and the target kernel size is the second size; When the area type of the monitoring area of ​​the camera is the public area, the target blur intensity is determined to be the third blur intensity and the target kernel size is the third size; When the area type of the camera's monitoring area is the long-distance monitoring area, the target blur intensity is determined to be the fourth blur intensity and the target kernel size is the fourth size; Wherein, the third blur intensity is greater than the first blur intensity, the first blur intensity is greater than the fourth blur intensity, and the fourth blur intensity is greater than the second blur intensity; the third size is greater than the first size, the first size is greater than the fourth size, and the fourth size is greater than the second size; After matching the target object with multiple preset objects, the method further includes: Determine the confidence level of the matching results between the target object and multiple preset objects; The process of desensitizing the region corresponding to the target object in the first image based on a blur kernel having the target blur intensity and the target kernel size to obtain a second image includes: If the confidence level of the matching result is less than or equal to the confidence threshold, a first adjustment coefficient is generated; The target blur intensity and the target kernel size are adjusted based on the first adjustment coefficient to obtain the adjusted target blur intensity and the adjusted target kernel size; the adjusted target blur intensity is greater than the unadjusted target blur intensity, and the adjusted target kernel size is greater than the unadjusted target kernel size. Based on a blur kernel with adjusted target blur intensity and adjusted target kernel size, the region corresponding to the target object in the first image is desensitized to obtain the second image.

2. The method according to claim 1, characterized in that, The target object is either a human face or a license plate.

3. The method according to claim 2, characterized in that, Determining the target object in the first image includes: The target object in the first image is determined using a face detection model or a license plate detection model.

4. The method according to claim 1, characterized in that, When the area type of the monitoring area of ​​the camera is the production area, determining that the target blur intensity is a second blur intensity and the target kernel size is a second size includes: If the area type of the monitoring area of ​​the camera is a production area, the target blur intensity is determined to be 0 and the target kernel size is determined to be 0.

5. The method according to claim 1, characterized in that, The method further includes: Determine the timestamp of the first image; The step of matching the target object with multiple preset objects includes: If the timestamp meets the preset time conditions, the target object is matched with multiple preset objects.

6. The method according to claim 5, characterized in that, If the timestamp does not meet the preset time condition, determining the target object in the first image includes: Identify the target object in the first image; Based on the second desensitization strategy, the region corresponding to the target object in the first image is desensitized to obtain the third image.

7. The method according to claim 1, characterized in that, The method further includes: The second image is sent to the target terminal so that the target terminal displays the second image.

8. An image desensitization device, characterized in that, Applied to cameras, including: The first determining module is used to determine the target object in the first image; the first image will only be stored locally on the camera and will not be uploaded to the server or sent to the terminal. A matching module is used to match the target object with multiple preset objects; The second determining module is used to determine the target blur intensity and target kernel size based on the region type of the monitoring area of ​​the camera when the target object fails to match with any of the preset objects; the region type is an office area, a production area, a public area, or a long-distance monitoring area. The desensitization module is used to perform desensitization processing on the region corresponding to the target object in the first image based on a blur kernel having the target blur intensity and the target kernel size, to obtain a second image; Specifically, regarding the determination of target blur intensity and target kernel size based on the region type of the monitored area of ​​the camera, the second determining module is used to determine the target blur intensity as a first blur intensity and the target kernel size as a first size when the region type of the monitored area of ​​the camera is an office area; to determine the target blur intensity as a second blur intensity and the target kernel size as a second size when the region type of the monitored area of ​​the camera is a production area; to determine the target blur intensity as a third blur intensity and the target kernel size as a third size when the region type of the monitored area of ​​the camera is a public area; and to determine the target blur intensity as a fourth blur intensity and the target kernel size as a fourth size when the region type of the monitored area of ​​the camera is a long-distance monitoring area. The third blur intensity is greater than the first blur intensity, the first blur intensity is greater than the fourth blur intensity, and the fourth blur intensity is greater than the second blur intensity; the third size is greater than the first size, the first size is greater than the fourth size, and the fourth size is greater than the second size. After matching the target object with multiple preset objects, the matching module is further used to determine the confidence level of the matching results between the target object and the multiple preset objects; The second determining module is specifically used to generate a first adjustment coefficient when the confidence level of the matching result is less than or equal to the confidence level threshold; adjust the target blur intensity and the target kernel size based on the first adjustment coefficient to obtain the adjusted target blur intensity and the adjusted target kernel size; the adjusted target blur intensity is greater than the unadjusted target blur intensity, and the adjusted target kernel size is greater than the unadjusted target kernel size; based on the blur kernel with the adjusted target blur intensity and the adjusted target kernel size, perform desensitization processing on the region corresponding to the target object in the first image to obtain the second image.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images