A new type of anti-theft lock processing method and system based on privacy minimization for intelligent entry door

Abstract

This invention relates to the field of smart home control technology, and more particularly to a novel anti-theft lock processing method and system for smart entrance doors based on privacy-minimizing video processing. This invention involves capturing images within a limited time window triggered by an entry operation event, and performing spatial region cropping and irreversible pixel stripping on the captured images to convert the original images into abstract image data that does not contain reproducible information. Based on this abstract image data, statistical features characterizing the rhythm and continuity of entry operations are extracted to achieve anti-theft risk assessment. The assessment result controls whether to generate and upload minimal evidence data, thereby reducing the collection, storage, and transmission of original image data while ensuring anti-theft assessment capabilities, thus lowering the risk of privacy leakage.

Description

Technical Field

[0001] This invention relates to the field of smart home control technology, and in particular to a novel anti-theft lock processing method and system for smart entrance doors based on privacy-minimizing video processing. Background Technology

[0002] With the development of smart home technology, smart door locks integrating video capture and analysis functions are gradually becoming an important part of home security. Existing smart door lock systems typically use cameras installed on the outside of the door to capture images of the entry process, and then upload the captured complete images or processed image data to the backend for recording entry behavior, analyzing anomalies, or tracing back to the home.

[0003] However, the aforementioned technical solutions have significant shortcomings in practical applications. On the one hand, existing solutions often rely on continuous or quasi-continuous video capture, acquiring and processing complete video data even in normal home entry scenarios. This results in the collection and storage of a large amount of image information unrelated to anti-theft assessment, increasing the risk of privacy leaks and data compliance pressures. On the other hand, in order to improve the accuracy of analysis, some solutions focus on processing the image to make it clearer, enhanced, or smoother, thereby further improving the image's recognizability. This, to some extent, contradicts users' actual needs for privacy protection.

[0004] Furthermore, existing technologies generally lack effective constraints on the scope, duration, and semantic level of image data acquisition, failing to technically acquire only the information necessary for anti-theft assessment. This results in a difficulty in balancing privacy protection and anti-theft capabilities. Therefore, a new intelligent entrance door anti-theft lock solution is urgently needed, capable of reliably assessing entry risks without relying on complete video capture and uploading. Summary of the Invention

[0005] The purpose of this invention is to provide a novel anti-theft lock processing method and system for smart entrance doors based on privacy-minimizing video processing, addressing the shortcomings of existing technologies. The aim is to reduce the collection, processing, and uploading of complete image data by imposing multi-dimensional constraints on the timing, spatial range, and semantic level of entrance image acquisition, while ensuring the ability to determine theft prevention upon entry.

[0006] This invention achieves the above objectives through the following technical solution: a novel anti-theft lock processing method for smart entrance doors based on privacy-minimizing video processing, characterized in that it includes: Pre-triggering steps: The input detection unit of the smart entrance door detects whether an entry operation event has occurred. When an entry operation event is detected, the wireless monitoring device set on the outside of the door or opposite the door is triggered to start image acquisition, and the image acquisition is limited to a preset time window. Region cropping step: Perform spatial region positioning processing on the images captured by the wireless monitoring device within the preset time window, and based on preset anti-theft discrimination related region rules, crop at least one local region image from the images, wherein the local region image is smaller than the overall spatial range of the original image; Pixel stripping step: Perform irreversible unnecessary pixel stripping on the local area of ​​the image, convert the local area of ​​the image into abstract image data that does not contain information that can be restored to the original image, and discard the corresponding original pixel data after the conversion is completed; Risk assessment step: Based on the abstract image data, perform anti-theft risk analysis on the home entry process to generate a risk assessment result that characterizes whether the current home entry behavior is abnormal; Evidence control steps: Based on the risk assessment result, control whether to generate evidence data for remote verification. When the risk assessment result meets the preset risk conditions, generate minimized evidence data based on the abstract image data and perform upload processing. When the risk assessment result does not meet the preset risk conditions, do not generate or upload any original image data or complete screen data.

[0007] Specifically, this invention proposes a complete intelligent entrance door anti-theft lock processing flow. The core is not to continuously or comprehensively monitor the entrance process with video, but to obtain only the minimum set of information necessary to complete the anti-theft judgment through a combination of event triggering, spatial shrinking, data abstraction and conditional output.

[0008] In practice, the system only activates the wireless monitoring device to capture images when an entry operation event is detected, and limits the acquisition duration through a preset time window. Then, it performs spatial area positioning on the captured images, retaining only the local areas related to door lock operation and personnel operation. Next, through irreversible pixel stripping processing, the local area images are converted into abstract image data that does not contain information that can be restored to the original image. Based on the abstract image data, an entry risk analysis is completed. And based on the risk analysis results, it is decided whether to generate and upload evidence data for remote verification.

[0009] Through the above process, the system can effectively avoid the storage and transmission of original images or complete video data in normal home entry scenarios while ensuring the anti-theft detection capability.

[0010] Specifically, this invention triggers image capture within a limited time window when an entry operation event is detected, and performs spatial region cropping and irreversible unnecessary pixel stripping on the captured image, so that the system completes risk analysis based only on abstract image data related to anti-theft judgment; at the same time, it controls the generation and uploading of evidence data according to the risk judgment results, thereby avoiding the retention of original image data or complete image data in normal entry scenarios.

[0011] Through the above-mentioned technical means, the present invention aims to solve the problems of high privacy leakage risk, redundant data processing, and difficulty in balancing privacy protection and anti-theft capabilities that are common in existing smart door lock systems.

[0012] Furthermore, in the pre-triggered step, the start time of the preset time window is determined when the input detection unit first detects the in-home operation event, and the end time of the preset time window is determined when either the in-home operation ends or the preset maximum collection duration is reached. Specifically, during the in-home operation-triggered screen capture process, the system clearly defines the start and end conditions for screen capture. Screen capture starts when the in-home operation event is first detected and terminates when the in-home operation ends or the collection duration reaches the preset upper limit. In this way, the screen capture time directly corresponds to the actual in-home operation process, avoiding uncontrollable screen capture time due to user pauses, accidental touches, or environmental changes, thus reducing the probability of capturing irrelevant screens from a time perspective.

[0013] Furthermore, in the area cropping step, the spatial area positioning processing includes determining the cropping area corresponding to the door lock panel based on pre-calibrated door lock operation area parameters, and determining the cropping area corresponding to the personnel operation area based on the inter-frame difference results. Specifically, when performing spatial area positioning on the acquired image, the system simultaneously uses both static area parameters and dynamic image change analysis to determine the cropping area. The door lock operation area is determined by pre-calibrated door lock operation area parameters and is used to cover fixed positions such as the door lock panel; the personnel operation area is determined by analyzing the inter-frame difference results and is used to cover the dynamic operation area generated by personnel during entry. Through these methods, the system can stably acquire image areas related to anti-theft judgment under different operating postures and different personnel conditions.

[0014] Furthermore, in the region cropping step, the cropping region corresponding to the personnel operation area is determined by performing grayscale difference processing on consecutive image frames and extracting connected regions with an area greater than a preset threshold from the difference results. Specifically, when determining the personnel operation area, the system performs grayscale difference processing on consecutive image frames and extracts connected regions with an area greater than a preset threshold from the difference results as the personnel operation area. This processing method can effectively eliminate minor lighting changes or noise interference, retaining only the areas of significant image changes caused by personnel operation. It is a mature and easily implemented image processing method in this field, and does not rely on complex models or deep learning algorithms.

[0015] Furthermore, in the pixel stripping step, the irreversible non-essential pixel stripping process includes converting the local area image into an edge contour image or a binary contour image, and deleting the corresponding grayscale or color image data after the conversion. Specifically, when performing pixel stripping on a local area image, the system converts the local area image into an edge contour image or a binary contour image, and deletes the corresponding grayscale or color image data after the conversion, thus achieving irreversible stripping of the original pixel information. Since the edge contour image or binary contour image does not contain the grayscale or color details of the original image, and the original image data is deleted after processing, the original image content cannot be restored through subsequent processing, thus achieving irreversible protection of privacy information from a technical structure perspective.

[0016] Furthermore, in the risk assessment step, the anti-theft risk analysis extracts statistical features representing the rhythm and continuity of the entry operation based on the abstract image data, and compares these statistical features with preset threshold rules to generate the risk assessment result. Specifically, after completing pixel stripping processing, the system performs risk analysis on the entry process based on the abstract image data. The risk analysis extracts statistical features that can represent the rhythm and continuity of the entry operation, and compares these statistical features with preset threshold rules to determine whether the current entry behavior deviates from the normal operating mode. This analysis method is based on calculable statistics, has a clear data source and judgment logic, does not rely on a black box model, and is easy to understand and implement.

[0017] Furthermore, the statistical features include at least the mean, variance, and peak count of the number of outline pixels in a local area of ​​the image over time. Specifically, during risk analysis, the system uses at least the mean, variance, and peak count of the number of outline pixels over time as statistical features. The mean reflects the average level of the overall movement amplitude during the in-home operation, the variance reflects the fluctuation of the movement amplitude over time, and the peak count reflects the rhythmic changes during the operation. Through comprehensive analysis of multi-dimensional statistical features, the characteristics of in-home operation behavior can be more accurately depicted.

[0018] Furthermore, in the evidence control step, the minimized evidence data is low-resolution contour sequence data generated based on the abstract image data, and the spatial resolution and temporal length of the minimized evidence data are both smaller than the original image data acquired within the preset time window. Specifically, when remote verification is required, the system generates low-resolution contour sequence data based on the abstract image data as minimized evidence data, and limits the spatial resolution and temporal length of this evidence data to be smaller than the original image data. By dually limiting the resolution and temporal range of the evidence data, even if data transmission occurs, it is impossible to restore the complete image details, thereby further reducing the risk of privacy leakage while meeting the anti-theft verification requirements.

[0019] A novel anti-theft lock processing system for smart entry doors based on privacy-minimizing video processing includes: The structure of an intelligent entrance door includes at least an input detection unit for detecting whether an entry operation event has occurred; A wireless monitoring device is installed on the outside of the door or opposite the door and is communicatively connected to the smart entrance door structure. It is used to perform image acquisition within a preset time window when the trigger command of the entrance operation event is received. A region cropping device, connected to the wireless monitoring device, is used to perform spatial region positioning processing on the images acquired within the preset time window, and to crop at least one local region image from the images based on preset anti-theft discrimination related region rules. A pixel stripping device, connected to the region cropping device, is used to perform irreversible and unnecessary pixel stripping processing on the local region image, converting the local region image into abstract image data that does not contain information that can be restored to the original image, and discarding the corresponding original pixel data after the conversion is completed. A risk discrimination device, connected to the pixel stripping device, is used to perform anti-theft risk analysis on the entry process based on the abstract image data, and generate a risk discrimination result to characterize whether the current entry behavior is abnormal. An evidence control device, connected to the risk discrimination device, is used to control whether to generate evidence data for remote verification based on the risk discrimination result. When the risk discrimination result meets the preset risk conditions, minimal evidence data is generated based on the abstract image data and uploaded. When the risk discrimination result does not meet the preset risk conditions, no original image data or complete image data is generated or uploaded.

[0020] Specifically, the system consists of multiple functional components, each corresponding to processing stages such as entry detection, image acquisition, region cropping, pixel stripping, risk analysis, and evidence control. During operation, the components work collaboratively according to a predetermined data flow sequence: the input detection unit provides the entry operation trigger signal; the wireless monitoring device performs image acquisition within a limited time window; the region cropping device, pixel stripping device, and risk assessment device sequentially complete image space shrinkage, data abstraction, and risk analysis; and the evidence control device determines whether to generate and upload evidence data based on the risk analysis results. Through the cooperation of these components, the system can fully implement the aforementioned anti-theft lock processing method.

[0021] Furthermore, the wireless monitoring device is configured to determine the start time of the preset time window when the input detection unit first detects an entry operation event, and to determine the end time of the preset time window when either an entry operation end event is detected or a preset maximum collection duration is reached. The region clipping device is configured to determine the clipping region corresponding to the door lock panel based on pre-calibrated door lock operation region parameters, and to determine the clipping region corresponding to the personnel operation region based on the difference results between consecutive frames in the image. The region cropping device performs grayscale difference processing on consecutive image frames and extracts connected regions with an area greater than a preset threshold from the difference results to determine the cropping region corresponding to the personnel operation area. The pixel stripping device is configured to convert the local area image into an edge contour image or a binary contour image, and delete the corresponding grayscale image data or color image data after the conversion is completed. The risk discrimination device is configured to extract statistical features representing the rhythm and continuity of the in-home operation based on the abstract image data, and compare the statistical features with a preset threshold rule to generate the risk discrimination result; The statistical features include at least the mean, variance, and peak number of contour pixels in a local area of ​​the image as a function of time. The minimized evidence data is low-resolution contour sequence data generated based on the abstract image data, and the spatial resolution and time length of the low-resolution contour sequence data are both smaller than the original image data acquired within the preset time window.

[0022] The beneficial effects of this invention are: First, this invention limits the start and end time of image capture by triggering an in-home operation event, ensuring that video capture only occurs within the necessary time period of the in-home operation, thus avoiding long-term or continuous monitoring of the external environment and reducing the capture of irrelevant images from a time perspective.

[0023] Secondly, this invention introduces a spatial region cropping mechanism during image processing, retaining only the local area image related to door lock operation and personnel operation, thereby reducing the amount of irrelevant information contained in the image from a spatial dimension and reducing the possibility of personal privacy information being collected.

[0024] Furthermore, this invention transforms local areas of the image into abstract image data that does not contain information that can be restored to the original image through irreversible, unnecessary pixel stripping. While completing the extraction of information required for anti-theft analysis, it eliminates the risk of the original image being restored or re-identified, thus achieving privacy protection at the semantic level.

[0025] Furthermore, this invention completes the risk assessment of home entry based on the abstract image data, and controls the generation and uploading of evidence data according to the risk assessment results. It generates and uploads minimal evidence data only when preset risk conditions are met, and does not upload any original image data or complete screen data in normal home entry scenarios, thereby effectively reducing the data storage and transmission burden while ensuring the practicality of anti-theft measures.

[0026] In summary, this invention achieves effective anti-theft judgment of entry behavior without relying on complete video capture and uploading, taking into account both the security and privacy protection needs of smart door locks, and has good practical value and promotion significance. Attached Figure Description

[0027] Figure 1 This is a schematic flowchart of one method of the present invention.

[0028] Figure 2 This is a schematic flowchart of another method of the present invention.

[0029] Figure 3 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. It is understood that the accompanying drawings are provided for reference and illustration only, and are not intended to limit the present invention. The connection relationships shown in the accompanying drawings are only for clear description and do not limit the connection method.

[0031] like Figures 1-2 As shown, a novel anti-theft lock processing method for smart entrance doors based on privacy-minimizing video processing is characterized by comprising: Pre-triggering steps: The input detection unit of the smart entrance door detects whether an entry operation event has occurred. When an entry operation event is detected, the wireless monitoring device set on the outside of the door or opposite the door is triggered to start image acquisition, and the image acquisition is limited to a preset time window. Region cropping step: Perform spatial region positioning processing on the images captured by the wireless monitoring device within the preset time window, and based on preset anti-theft discrimination related region rules, crop at least one local region image from the images, wherein the local region image is smaller than the overall spatial range of the original image; Pixel stripping step: Perform irreversible unnecessary pixel stripping on the local area of ​​the image, convert the local area of ​​the image into abstract image data that does not contain information that can be restored to the original image, and discard the corresponding original pixel data after the conversion is completed; Risk assessment step: Based on the abstract image data, perform anti-theft risk analysis on the home entry process to generate a risk assessment result that characterizes whether the current home entry behavior is abnormal; Evidence control steps: Based on the risk assessment result, control whether to generate evidence data for remote verification. When the risk assessment result meets the preset risk conditions, generate minimized evidence data based on the abstract image data and perform upload processing. When the risk assessment result does not meet the preset risk conditions, do not generate or upload any original image data or complete screen data.

[0032] In this embodiment, as Figure 3 As shown, the new intelligent entrance door anti-theft lock processing system includes an intelligent entrance door structure, a wireless monitoring device, an area cropping device, a pixel stripping device, a risk discrimination device, and an evidence control device.

[0033] The intelligent entrance door structure includes at least an input detection unit, which is connected to the digital input or operation components of the door lock to detect whether an entry operation event has occurred. The wireless monitoring device is located on the outside of the door or opposite the door, and its installation angle and field of view are pre-adjusted so that its captured image can cover the door lock operation area and the area where people outside the door are operating. The wireless monitoring device is connected to the intelligent entrance door structure via wireless communication.

[0034] In actual deployment, the wireless monitoring device can be fixed on the wall directly in front of the door or above the door frame, so that it can complete the perception of the entry operation process outside the door without collecting images of the indoor environment.

[0035] When a user enters the home at the door lock, the input detection unit detects password input, touch operation, or other preset entry operation behavior, and sends the detection result as an entry operation event to the wireless monitoring device.

[0036] Upon receiving the entry operation event, the wireless monitoring device does not perform continuous recording, but only starts image acquisition within a preset time window. The start time of the preset time window is determined when the entry operation event is first detected, and its end time is determined by either detecting an entry operation completion event or reaching a preset maximum acquisition duration.

[0037] By using the above method, the system only acquires screen data during the necessary time period when the in-home operation occurs, thus avoiding the collection of irrelevant screen data from a temporal perspective.

[0038] After completing the image acquisition within the preset time window, the region cropping device performs spatial region positioning processing on the acquired image.

[0039] Specifically, the area trimming device first determines the trimming area corresponding to the door lock panel in the screen based on the pre-calibrated door lock operation area parameters; the door lock operation area parameters can be determined manually during the equipment installation and debugging stage and stored in the system for subsequent use.

[0040] Meanwhile, the region cropping device performs grayscale difference processing on consecutive frame images. By comparing the pixel changes between adjacent frame images, it extracts connected regions with an area greater than a preset threshold in the difference results, thereby determining the cropping region corresponding to the personnel operation area.

[0041] After completing the above positioning, the area cropping device retains only the cropped images of the door lock operation area and the personnel operation area, and the remaining image areas are no longer involved in subsequent processing.

[0042] After obtaining the local area image, the pixel stripping device performs irreversible, unnecessary pixel stripping processing on the local area image.

[0043] Specifically, the pixel stripping device converts the local area image into an edge contour image or a binary contour image. This conversion process only retains the structural contour information in the image, without retaining grayscale texture information or color information. After the conversion is completed, the system immediately deletes the corresponding grayscale image data or color image data, so that the original image cannot be restored through subsequent processing.

[0044] By using the aforementioned irreversible pixel stripping method, the system retains the structural information required for anti-theft analysis while eliminating image information that can identify individuals from the source.

[0045] After completing the pixel stripping process, the risk discrimination device performs anti-theft risk analysis on the entry process based on the abstract image data.

[0046] In this embodiment, the risk discrimination device extracts statistical features from the abstract image data that characterize the rhythm and continuity of the in-home operation. The statistical features include at least the mean, variance, and peak number of contour pixels as a function of time.

[0047] The risk discrimination device compares the statistical features with preset threshold rules to determine whether the current home visit behavior deviates from the normal operation mode, and generates a risk discrimination result to characterize whether the home visit behavior is abnormal.

[0048] After obtaining the risk assessment result, the evidence control device controls whether to generate evidence data for remote verification based on the risk assessment result.

[0049] When the risk assessment result meets the preset risk conditions, the evidence control device generates minimized evidence data based on the abstract image data and performs upload processing; the minimized evidence data is low-resolution contour sequence data, whose spatial resolution and time length are both smaller than the original image data acquired within the preset time window.

[0050] When the risk assessment result does not meet the preset risk conditions, the system does not generate or upload any original image data or complete screen data, thereby avoiding unnecessary data retention and transmission in normal home entry scenarios.

[0051] Through the above technical solution, this embodiment, while ensuring the ability to detect burglaries, processes image data only at the necessary time, space, and semantic level, effectively reducing the risk of privacy leakage and data compliance costs. At the same time, it avoids the collection and uploading of complete video footage, making it suitable for smart door burglar lock application scenarios with high privacy protection requirements.

[0052] In detail: In this embodiment, the new smart entrance door anti-theft lock processing system based on privacy-minimizing video processing achieves the minimized acquisition of information required for entrance anti-theft judgment by constraining the timing, range, and semantic level of image acquisition.

[0053] When the system is running, after the input detection unit in the smart entrance door structure detects an entry operation event, it triggers the wireless monitoring device to start image acquisition. The image acquisition is only performed within a preset time window. The preset time window covers the necessary time period for the entry operation to occur. Image acquisition is terminated after an entry operation completion event is detected or the preset maximum acquisition time is reached, thereby avoiding long-term or continuous monitoring of the external environment.

[0054] After completing the image acquisition within the preset time window, the region cropping device performs spatial region positioning processing on the acquired images. This spatial region positioning processing includes determining the image region corresponding to the door lock panel based on pre-calibrated door lock operation area parameters, and determining the image region corresponding to the personnel operation area based on the difference results between consecutive image frames. The region cropping device only performs cropping processing on the image regions related to anti-theft detection; other image regions are not included in subsequent analysis.

[0055] After obtaining the local area image, the pixel stripping device performs irreversible unnecessary pixel stripping processing on the local area image. The irreversible unnecessary pixel stripping processing includes converting the local area image into an edge contour image or a binary contour image, and deleting the corresponding grayscale image data or color image data after the conversion is completed, so that the original image data cannot be recovered or reconstructed.

[0056] After pixel stripping, the risk discrimination device performs anti-theft risk analysis on the entry process based on the abstract image data. The risk discrimination device extracts statistical features representing the rhythm and continuity of the entry operation from the abstract image data, and compares these statistical features with preset threshold rules to generate a risk discrimination result indicating whether the current entry behavior is abnormal.

[0057] After obtaining the risk assessment result, the evidence control device controls whether to generate evidence data for remote verification based on the risk assessment result. When the risk assessment result meets the preset risk conditions, the evidence control device generates minimal evidence data based on the abstract image data and performs upload processing; when the risk assessment result does not meet the preset risk conditions, the system does not generate or upload any original image data or complete image data.

[0058] By employing the above processing method, the system can obtain the information required for intrusion detection while avoiding the collection, storage, and transmission of complete video footage data, effectively reducing the risk of privacy leaks and minimizing the burden of data processing and transmission.

[0059] In this embodiment, the risk discrimination device performs anti-theft risk analysis on the entry process based on the abstract image data obtained after pixel stripping. The abstract image data is a sequence of contour images formed by edge contour extraction or binarization of a local area of ​​the image.

[0060] In a specific implementation, the system performs frame-by-frame processing on consecutive image frames in the contour image sequence and counts the number of contour pixels in each frame to obtain a contour pixel count value that characterizes the degree of change in the current image structure.

[0061] Assume that a total of [number] samples are collected within the preset time window. Frame contour image, number The number of contour pixels in a frame contour image is denoted as ,in .

[0062] Based on the number of contour pixels, the system calculates the following statistical features: Average number of outline pixels ; Contour pixel count variance ; Peak number of outline pixels, where, when satisfying and At that time, the judgment of the first A frame represents a peak value, and the total number of times the peak value occurs is counted as the peak value quantity feature.

[0063] The above statistical characteristics are used to characterize the changes in the amplitude of personnel's movements and the continuity of the operational rhythm during the home visit operation.

[0064] During the risk assessment process, the system compares the mean number of contour pixels, the variance of the number of contour pixels, and the peak number of contour pixels with preset thresholds. When at least one of the statistical features exceeds the corresponding preset threshold, the system determines that the current entry behavior deviates from the normal operation mode and generates an abnormal risk assessment result; when none of the statistical features exceed the corresponding preset thresholds, a normal risk assessment result is generated.

[0065] After generating the risk assessment result, the system controls whether to generate evidence data for remote verification based on the risk assessment result. When the risk is determined to be abnormal, low-resolution minimized evidence data is generated based on the contour image sequence and uploaded. When the risk is determined to be normal, no original image data or complete screen data is generated or uploaded.

[0066] Based on the above embodiments, the preset thresholds and normal operating modes involved in the risk identification step can be determined by adaptive adjustment during initialization learning and operation.

[0067] During the initial deployment phase of the system, the intelligent entrance door anti-theft lock processing system can collect abstract image data corresponding to multiple normal entrance operations within a preset initialization period, and calculate statistical characteristics such as the mean, variance, and peak number of contour pixels based on the abstract image data, and use the statistical distribution range of the statistical characteristics as the reference interval for the normal operation mode.

[0068] During subsequent operation, when the system continuously determines that multiple home visits are normal risks, the reference interval can be updated based on the newly obtained statistical features, thereby achieving adaptive adjustment to different user operating habits and different installation environments; when the system determines that the home visit behavior has abnormal risks, the corresponding statistical features will not participate in the update of the reference interval.

[0069] In this way, the preset threshold is not a fixed value, but an adjustable parameter based on historical normal operation data, enabling the system to maintain a stable risk assessment capability under different application scenarios. Those skilled in the art can, based on the above description and in conjunction with specific application environments, set the initialization cycle length, update conditions, and adjustment range to implement the risk assessment process of this invention.

[0070] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A novel anti-theft lock processing method for smart entrance doors based on privacy-minimizing video processing, characterized in that, include: Pre-triggering steps: The input detection unit of the smart entrance door detects whether an entry operation event has occurred. When an entry operation event is detected, the wireless monitoring device set on the outside of the door or opposite the door is triggered to start image acquisition, and the image acquisition is limited to a preset time window. Region cropping step: Perform spatial region positioning processing on the images captured by the wireless monitoring device within the preset time window, and based on preset anti-theft discrimination related region rules, crop at least one local region image from the images, wherein the local region image is smaller than the overall spatial range of the original image; Pixel stripping step: Perform irreversible unnecessary pixel stripping on the local area of ​​the image, convert the local area of ​​the image into abstract image data that does not contain information that can be restored to the original image, and discard the corresponding original pixel data after the conversion is completed; Risk assessment step: Based on the abstract image data, perform anti-theft risk analysis on the home entry process to generate a risk assessment result that characterizes whether the current home entry behavior is abnormal; Evidence control steps: Based on the risk assessment result, control whether to generate evidence data for remote verification. When the risk assessment result meets the preset risk conditions, generate minimized evidence data based on the abstract image data and perform upload processing. When the risk assessment result does not meet the preset risk conditions, do not generate or upload any original image data or complete screen data.

2. The method for processing a novel anti-theft lock for a smart entrance door based on privacy-minimizing video processing according to claim 1, characterized in that, In the pre-triggered step, the start time of the preset time window is determined when the input detection unit first detects the in-home operation event, and the end time of the preset time window is determined when either the in-home operation end event is detected or the preset maximum collection time is reached.

3. The method for processing a novel anti-theft lock for a smart entrance door based on privacy-minimizing video processing according to claim 2, characterized in that, In the area cropping step, the spatial area positioning process includes determining the cropping area corresponding to the door lock panel based on the pre-calibrated door lock operation area parameters, and determining the cropping area corresponding to the personnel operation area based on the inter-frame difference results of the image. The cropping area corresponding to the personnel operation area is determined by performing grayscale difference processing on continuous image frames and extracting connected regions with an area greater than a preset threshold from the difference results.

4. The method for processing a novel anti-theft lock for a smart entrance door based on privacy-minimizing video processing according to claim 3, characterized in that, In the pixel stripping step, the irreversible non-essential pixel stripping process includes converting the local area image into an edge contour image or a binary contour image, and deleting the corresponding grayscale image or color image data after the conversion is completed.

5. The method for processing a novel anti-theft lock for a smart entrance door based on privacy-minimizing video processing according to claim 4, characterized in that, In the risk assessment step, the anti-theft risk analysis extracts statistical features representing the rhythm and continuity of entry operations based on the abstract image data, and compares the statistical features with preset threshold rules to generate the risk assessment result.

6. The method for processing a novel anti-theft lock for a smart entrance door based on privacy-minimizing video processing according to claim 5, characterized in that, The statistical features include at least the mean, variance, and peak number of contour pixels in a local area of ​​the image as a function of time. In the evidence control step, the minimized evidence data is low-resolution contour sequence data generated based on the abstract image data, and the spatial resolution and time length of the minimized evidence data are both smaller than the original image data acquired within the preset time window.

7. A novel anti-theft lock processing system for smart entrance doors based on privacy-minimizing video processing, characterized in that, include: The structure of an intelligent entrance door includes at least an input detection unit for detecting whether an entry operation event has occurred; A wireless monitoring device is installed on the outside of the door or opposite the door and is communicatively connected to the smart entrance door structure. It is used to perform image acquisition within a preset time window when the trigger command of the entrance operation event is received. A region cropping device, connected to the wireless monitoring device, is used to perform spatial region positioning processing on the images acquired within the preset time window, and to crop at least one local region image from the images based on preset anti-theft discrimination related region rules. A pixel stripping device, connected to the region cropping device, is used to perform irreversible and unnecessary pixel stripping processing on the local region image, converting the local region image into abstract image data that does not contain information that can be restored to the original image, and discarding the corresponding original pixel data after the conversion is completed. A risk discrimination device, connected to the pixel stripping device, is used to perform anti-theft risk analysis on the entry process based on the abstract image data, and generate a risk discrimination result to characterize whether the current entry behavior is abnormal. An evidence control device, connected to the risk discrimination device, is used to control whether to generate evidence data for remote verification based on the risk discrimination result. When the risk discrimination result meets the preset risk conditions, minimal evidence data is generated based on the abstract image data and uploaded. When the risk discrimination result does not meet the preset risk conditions, no original image data or complete image data is generated or uploaded.

8. The novel anti-theft lock processing system for smart entrance doors based on privacy-minimizing video processing according to claim 7, characterized in that, The wireless monitoring device is configured to determine the start time of the preset time window when the input detection unit first detects an entry operation event, and to determine the end time of the preset time window when either an entry operation end event is detected or a preset maximum collection duration is reached. The region clipping device is configured to determine the clipping region corresponding to the door lock panel based on pre-calibrated door lock operation region parameters, and to determine the clipping region corresponding to the personnel operation region based on the difference results between consecutive frames in the image.

9. The novel anti-theft lock processing system for smart entrance doors based on privacy-minimizing video processing according to claim 8, characterized in that, The region cropping device performs grayscale difference processing on consecutive image frames and extracts connected regions with an area greater than a preset threshold from the difference results to determine the cropping region corresponding to the personnel operation area. The pixel stripping device is configured to convert the local area image into an edge contour image or a binary contour image, and delete the corresponding grayscale image data or color image data after the conversion is completed. The risk discrimination device is configured to extract statistical features representing the rhythm and continuity of the in-home operation based on the abstract image data, and compare the statistical features with a preset threshold rule to generate the risk discrimination result.

10. The novel anti-theft lock processing system for smart entrance doors based on privacy-minimizing video processing according to claim 9, characterized in that, The statistical features include at least the mean, variance, and peak number of contour pixels in a local area of ​​the image as a function of time. The minimized evidence data is low-resolution contour sequence data generated based on the abstract image data, and the spatial resolution and time length of the low-resolution contour sequence data are both smaller than the original image data acquired within the preset time window.

Images