Household intelligent security anti-theft system based on Internet of Things

Abstract

The invention discloses a household intelligent safety anti-theft system based on the Internet of Things, and relates to the technical field of intelligent household anti-theft, and the system comprises an input information monitoring module which is used for monitoring the password input information and speed, collecting a monitoring video, and recognizing the password correctness; the password correctness module is used for correcting the password correctness according to the input speed; the monitoring image identification module is used for adjusting an identification step length according to an input proficiency coefficient, extracting an image sequence and identifying a person suspicious degree; and the anti-theft alarm module is used for calculating the safety degree according to the correction password correctness and the person suspicious degree, and if the safety degree is lower than a threshold value, anti-theft alarm is triggered. The technical problems that in the prior art, a traditional home intelligent safety system often depends on a single judgment standard, and the accuracy and reliability of protection alarm are insufficient are solved, and the technical effects that through multi-level safety evaluation, the risk of false alarm and missing alarm is reduced, and the accuracy and reliability of home safety protection alarm are improved are achieved.

Description

Technical Field

[0001] This invention relates to the field of smart home anti-theft technology, specifically to a smart home security and anti-theft system based on the Internet of Things. Background Technology

[0002] The security of smart locks lies in their ability to authenticate users through multiple methods such as passwords, fingerprints, and facial recognition. However, due to factors such as password input error rates, attacks by malicious actors, and external interference, relying solely on password input is not very secure. Furthermore, while existing surveillance equipment can provide real-time video monitoring, judging the presence of security threats based solely on video stream recordings often lacks sufficient accuracy and intelligent analysis, easily leading to false alarms or missed detections.

[0003] Existing smart door locks rely solely on passwords or biometrics for authentication, making them vulnerable to password guessing, cracking, or brute-force attacks. Furthermore, most of the corresponding monitoring devices only have video recording capabilities and lack intelligent behavior recognition and anomaly detection capabilities. As a result, the overall accuracy and real-time performance of security alarms are poor, and false alarms and missed alarms are common. Summary of the Invention

[0004] This application provides an Internet of Things-based smart home security and anti-theft system to address the technical problem that traditional smart home security systems in the prior art often rely on a single judgment standard, resulting in insufficient accuracy and reliability of protection alarms.

[0005] This application provides an IoT-based smart home security and anti-theft system, comprising: an input information monitoring module, used to monitor and acquire input password information and password input speed information when the password authentication interface of the smart door lock is activated, and to collect monitoring video through the monitoring device, and to identify and acquire password correctness when the input password information is incorrect; a password correctness processing module, used to classify input proficiency coefficients according to the password input speed information, obtain input proficiency coefficients, correct the password correctness, and obtain corrected password correctness; a monitoring image recognition module, used to configure personnel recognition step size according to the input proficiency coefficients, extract monitoring images from the monitoring video, obtain monitoring image sequences, and identify personnel suspiciousness; and an anti-theft alarm module, used to calculate a security level according to the corrected password correctness and personnel suspiciousness, and to trigger an anti-theft alarm when the security level is less than a preset security level threshold.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application provides an IoT-based smart home security and anti-theft system, relating to the field of smart home anti-theft technology. By monitoring the password input information, input speed, and surveillance video of the smart door lock, it identifies and corrects the password's correctness. Based on an input proficiency coefficient, it adjusts the recognition step size of the surveillance image and analyzes the suspiciousness of personnel. Finally, it combines password correctness and suspiciousness to calculate the security level. If the level falls below a threshold, an anti-theft alarm is triggered. This solves the technical problem that traditional smart home security systems often rely on a single judgment standard, resulting in insufficient accuracy and reliability of alarms. It achieves multi-level security assessment by comprehensively considering password correctness and personnel suspiciousness, thereby reducing the risk of false alarms and missed alarms and improving the accuracy and reliability of home security alarms. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 A schematic diagram of a home smart security and anti-theft system based on the Internet of Things provided in this application embodiment; Figure 2 This is a schematic diagram illustrating the process of a monitoring image recognition module in an IoT-based smart home security and anti-theft system identifying the suspiciousness of a person, as provided in an embodiment of this application.

[0009] Explanation of reference numerals in the attached diagram: Input information monitoring module 10, password accuracy verification module 20, monitoring image recognition module 30, and burglar alarm module 40. Detailed Implementation

[0010] This application provides an Internet of Things-based smart home security and anti-theft system to address the technical problem that traditional smart home security systems in the prior art often rely on a single judgment standard, resulting in insufficient accuracy and reliability of protection alarms.

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0012] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0013] Example 1, as Figure 1 As shown, this application provides a smart home security and anti-theft system based on the Internet of Things (IoT). The system includes smart door locks and monitoring devices interconnected via the IoT. The system includes: The input information monitoring module 10 is used to monitor and obtain the input password information and password input speed information when the password authentication interface of the smart door lock is started, and to collect monitoring video through the monitoring device, and to identify and obtain the password correctness when the input password information is incorrect.

[0014] Furthermore, the input information monitoring module 10 is also used to perform the following steps: P11: When the password authentication interface of the smart door lock is started, monitor and obtain the input password information and input time information, and collect monitoring video through the monitoring device; P12: Calculate the password input speed information based on the input time information; P13: Determine whether the input password information is correct. If yes, unlock the door; if no, identify the correctness of the input password information.

[0015] It should be understood that the input information monitoring module 10 of this application is responsible for comprehensively monitoring and analyzing the user's input behavior when the password authentication interface of the smart door lock is started. Its main objective is to determine the normality of the input behavior and the correctness of the password by monitoring the password input information and input speed, combined with the video data collected by the monitoring equipment, thereby providing an effective basis for subsequent anti-theft judgments.

[0016] Specifically, when a user enters a password through the smart lock, the input information monitoring module is activated and begins monitoring the input behavior. At this time, the system records the password information entered by the user in real time. This password information refers to the numbers or characters the user presses one by one in the password box, which may include numeric passwords, letter passwords, or other types of character passwords. Simultaneously, the module also records the input time information, that is, the time spent from when the user begins inputting to when they finish. This time information is crucial for subsequent analysis of the user's input speed. At the same time, the monitoring equipment simultaneously collects surveillance video of the user's surrounding environment for subsequent person recognition and behavior analysis, especially when the password is entered incorrectly, as the video data can provide crucial auxiliary information.

[0017] Furthermore, based on the input time information, input speed information is calculated. Input speed information refers to the speed at which a user enters the password within a certain time. If the user's input time is short, it indicates that the user is entering the password quickly, possibly meaning the user is familiar with the password. Conversely, a longer input time may indicate that the user is entering the password slowly, possibly because an unauthorized person is attempting to enter the password. Password input speed not only reflects the user's input habits but also provides important clues for subsequent behavioral analysis. High-speed input may indicate a legitimate user, while slow input is more likely to indicate an unauthorized user.

[0018] Next, the system checks if the entered password matches the preset correct password. If the password is correct, the system will directly unlock the door, allowing the authorized user to enter. If the password is incorrect, the system will initiate an error detection process. During this process, the password correctness is calculated and evaluated. This process uses a password correctness algorithm, which calculates a numerical ratio based on the degree of match between the actual entered password and the correct password, typically the percentage of the correct password's length. For example, if the correct password is "1234" and the user enters "1237", the algorithm will calculate a correctness of 75%. This correctness value, combined with input speed information, can further adjust the evaluation of password correctness, accurately determining whether the user is an authorized person.

[0019] In this step, the input password and input time are crucial data sources, while password accuracy is a key indicator for determining whether an intrusion has occurred. By comprehensively considering both the password input error rate and input speed, the system can more intelligently determine whether the behavior is from an unauthorized user, thereby improving system security.

[0020] Furthermore, to identify the correctness of the input password information, the input information monitoring module 10 is also used to perform the following steps: P13-1: Obtain the correct password; P13-2: Identify the similarity between the input password information and the correct password to obtain the password correctness.

[0021] Optionally, before verifying password correctness, the system must first obtain the correct password. This step is a prerequisite for the entire password verification process. The correct password refers to the system's preset password, typically stored in the smart lock's database as the standard for subsequent verification. The system can obtain the correct password through local storage or by synchronizing the corresponding password data with the cloud. It is important to note that the process of obtaining the correct password should be secure, ensuring that only authorized users or administrators can access this information to prevent password leakage or unauthorized access.

[0022] After obtaining the correct password, the system will evaluate the password information entered by the user. The entered password information refers to the actual sequence of characters entered by the user, which may contain numbers, letters, or other symbols. The system calculates the similarity between the entered password and the correct password by comparing their differences, and ultimately obtains the password's correctness. For example, a string matching algorithm, such as Levenshtein distance (edit distance), is used to calculate the minimum number of operations (including insertion, deletion, and replacement operations) required to transform one string into another, to measure the degree of difference between the two strings. The smaller the edit distance, the more similar the two passwords are, and the higher the password's correctness.

[0023] Using this type of algorithm, the system can output a password correctness value, typically a floating value between 0 and 1, representing the degree of matching between the input password and the correct password. For example, if the user enters "1237" and the correct password is "1234", the system might calculate a similarity value of 0.75 (i.e., 75% similarity) based on the string matching algorithm, thus concluding that the password is 75% correct. Through this method, the system can not only identify completely incorrect input but also accurately assess the accuracy of the entered password, further enhancing security capabilities when combined with other data (such as input speed and surveillance video).

[0024] The password accuracy module 20 is used to classify the input proficiency coefficient based on the password input speed information, obtain the input proficiency coefficient, correct the password accuracy, and obtain the corrected password accuracy.

[0025] Furthermore, the password correctness verification module 20 is also used to perform the following steps: P21: Construct an input proficiency classifier; P22: Input the password input speed information into the input proficiency classifier to classify and obtain the input proficiency coefficient; P23: Multiply the input proficiency coefficient by the password correctness to complete the correction calculation and obtain the corrected password correctness.

[0026] Specifically, the password correctness module 20 in this application calculates and corrects the password's correctness by combining the user's input speed information. The corrected password correctness will more accurately reflect the user's input behavior, thereby improving the system's accuracy in detecting unauthorized intrusions.

[0027] First, the password accuracy module 20 needs to build an input proficiency classifier. This classifier determines a user's input proficiency based on their input speed. Input proficiency refers to a user's familiarity with the password input process, usually measured by their input speed. Highly proficient users will input passwords quickly, while users unfamiliar with passwords or those attempting to impersonate others will input more slowly.

[0028] The input proficiency classifier can be constructed using machine learning algorithms, particularly supervised learning algorithms, to identify different input speed characteristics through a training dataset. By analyzing behavioral patterns at different input speeds, the user's input speed information is mapped to a preset proficiency level. Common classification algorithms include Support Vector Machines (SVM), Decision Trees, and K-Nearest Neighbors (KNN), with the specific choice depending on the characteristics of the data and actual needs.

[0029] As a user enters their password, their input speed information is collected in real time and transmitted to an input proficiency classifier. Input speed information refers to the time it takes for the user to type their password (e.g., the time to type each character) and the total duration of the entire input process. Based on this data, the classifier analyzes the user's input speed and categorizes them into different proficiency levels (e.g., high proficiency, medium proficiency, and low proficiency). For example, if a user types their password quickly, the classifier might label them as "high proficiency" and assign them a higher proficiency coefficient. If a user types slowly or hesitates, the classifier will label them as "low proficiency" and assign them a lower proficiency coefficient accordingly.

[0030] After obtaining the input proficiency coefficient, the system will adjust the password correctness based on this coefficient. The adjustment is calculated by multiplying the input proficiency coefficient by the original password correctness to obtain the adjusted password correctness. For example, if the original password correctness is 80% (i.e., 0.8), and the input proficiency coefficient calculated based on input speed is 1.2 (indicating that the user is proficient in input), then the adjusted password correctness is: Adjusted Password Correctness = Password Correctness × Input Proficiency Coefficient = 0.8 × 1.2 = 0.96. This means that the user's input password is very close to the correct password, and the system can consider the user to be a legitimate user, and may even allow access when the password is close to correct.

[0031] If the input proficiency coefficient is low (e.g., 0.7, indicating that the user is not proficient in input or the password has been seen by others), then the password correction accuracy will be adjusted to a lower value. This means that the system will consider that there is a large error in the password input, thereby triggering further security warnings or anti-theft measures, which can significantly improve the security of the system.

[0032] Furthermore, to construct an input proficiency classifier, the password correctness module 20 is also used to perform the following steps: P21-1: Based on the historical usage data of the smart lock, obtain the sample password input speed information set of the smart lock, calculate the average password input speed information, calculate the ratio of each sample password input speed information to the average password input speed information, and obtain the sample input proficiency coefficient set; P21-2: Construct the mapping relationship between the sample password input speed information set and the sample input proficiency coefficient set to obtain the input proficiency classifier.

[0033] In one possible embodiment of this application, in order to construct an input proficiency classifier, the system needs to use the historical usage data of the smart lock, i.e., past password input records, in order to generate a set of sample password input speed information and a set of sample input proficiency coefficients, thereby providing training data for the classifier.

[0034] Specifically, the first step is to access the historical usage data of the smart lock. This data typically includes the password input time from multiple past uses of the lock, recording the time spent each time the user entered the password. By collecting and organizing this historical data, a sample password input speed information set is obtained. This set contains input time information for each record from multiple historical data points, which the system converts into input speed to obtain the sample password input speed information set.

[0035] Next, the average password input speed information of these samples is calculated, which is the average password input speed across all historical records. This average password input speed information serves as a baseline for the entire system, representing the typical input speed of an average user using this smart lock.

[0036] Then, the ratio between the sample password input speed information and the average password input speed information in each historical record is calculated. This ratio reflects the difference in each user's input speed relative to the average speed. For example, if a user's input speed is faster than the average speed, the ratio will be greater than 1; if it is slower, the ratio will be less than 1. Based on these ratios, a set of sample input proficiency coefficients is derived, with each ratio corresponding to a user's proficiency in inputting passwords. Users with faster input speeds (ratio greater than 1) are generally considered proficient users, while users with slower input speeds (ratio less than 1) may be new users, slower users, or unauthorized users.

[0037] Next, machine learning techniques are used to establish the relationship between the set of sample password input speed information and the set of sample input proficiency coefficients, forming an input proficiency classifier. This is a supervised learning process; the system uses known historical data (sample password input speed information and corresponding input proficiency coefficients) to train the classifier, enabling it to predict the user's input proficiency based on new input speed information.

[0038] For example, when building a classifier, various common classification algorithms can be used, such as Support Vector Machines (SVM), which find an optimal hyperplane to map input speed information and proficiency coefficients to different categories. Decision tree algorithms, on the other hand, construct a tree structure to recursively determine a user's input proficiency based on the characteristics of input speed. The goal of these algorithms is to train the classifier to accurately map new password input speed information to the corresponding input proficiency. For instance, the system might classify users with faster input speeds and a larger ratio to the average speed as highly proficient, while classifying slower users as less proficient.

[0039] Once training is complete, the system can use the input proficiency classifier to classify users' input speed in real time and generate corresponding input proficiency coefficients. These coefficients are then used to adjust the user's password accuracy, thereby improving the system's recognition accuracy.

[0040] The monitoring image recognition module 30 is used to configure the personnel recognition step size according to the input proficiency coefficient, extract monitoring images from the monitoring video, obtain a monitoring image sequence, and identify the suspiciousness of the personnel.

[0041] Furthermore, such as Figure 2 As shown, the monitoring image recognition module 30 is also used to perform the following steps: P31: Obtain a preset personnel identification step size, which includes the step size for extracting monitoring images from the monitoring video; P32: Multiply the input proficiency coefficient by the preset personnel identification step size to obtain the personnel identification step size; P33: Extract multiple frames of monitoring images from the monitoring video according to the personnel identification step size to obtain a monitoring image sequence; P34: Identify the monitoring image sequence to obtain the personnel suspiciousness level.

[0042] It should be understood that the function of the surveillance image recognition module 30 in this application is to determine whether a user's identity and behavior are suspicious by analyzing the surveillance video and combining it with the user's input proficiency coefficient. This module adjusts the video frame extraction frequency by adjusting the personnel recognition step size, thereby reducing the waste of computing resources while ensuring recognition accuracy.

[0043] First, the monitoring image recognition module 30 needs to obtain a preset personnel recognition step size. The personnel recognition step size refers to the frequency at which image frames are extracted from the monitoring video, that is, the time interval between video frame extractions. Typically, the video stream is continuous, containing multiple frames per second. If the step size is set small, more image frames are extracted per second, resulting in higher image analysis accuracy, but the computational load also increases; conversely, if the step size is large, fewer image frames are extracted, saving computational resources, but some important details may be missed. Therefore, choosing an appropriate step size is key to optimizing system performance.

[0044] In practical applications, the input proficiency coefficient (derived from the aforementioned password input speed analysis) adjusts the personnel recognition step size. Specifically, the system multiplies the input proficiency coefficient by the preset personnel recognition step size to obtain a dynamic personnel recognition step size. This adjustment aims to adjust the image extraction frequency based on the user's input habits: for proficient users (with a higher input proficiency coefficient), the system considers their password input behavior to be relatively standardized, thus reducing the image extraction frequency to save computational resources. For inexperienced users or potential criminals (with a lower input proficiency coefficient), the system increases the image extraction frequency to enhance the accuracy of user behavior monitoring.

[0045] Once the personnel recognition step size is determined, the system extracts image frames from the surveillance video according to that step size. Specifically, multiple frames are extracted from the video stream at set time intervals (determined by the personnel recognition step size). These extracted image frames form a sequence of surveillance images. For example, if the step size is 2 seconds, the system will extract one frame from the video every 2 seconds. If the step size is smaller (e.g., 1 second), the system will extract more image frames; if the step size is larger (e.g., 5 seconds), fewer images will be extracted each time. By adjusting the step size, the system can balance the accuracy of image recognition with the consumption of computational resources.

[0046] Finally, the acquired surveillance image sequences are identified to determine whether the individuals appearing in the images are suspicious. The identification process typically includes steps such as object detection, face recognition, and behavior analysis. The system can use deep learning algorithms, such as convolutional neural networks (CNNs), to analyze image content, thereby determining the identity and behavioral patterns of individuals.

[0047] During this process, the system calculates the suspiciousness level of each identified individual based on the image content. Suspiciousness level is a numerical value representing the degree to which a person's behavior in an image deviates from normal behavior. If the identification results indicate that a user's behavior does not conform to normal usage patterns (e.g., entering passwords too slowly, peeking at passwords, or other abnormal behavior), the system will mark that person as suspicious and issue an alarm signal to enhance system security.

[0048] Furthermore, the monitoring image recognition module 30 is also used to perform the following steps: P34-1: Based on the monitoring and identification data within a historical time period, a set of sample monitoring image sequences is collected, and the suspiciousness of personnel within each sample monitoring image sequence is identified to obtain a set of sample personnel suspiciousness, wherein each sample personnel suspiciousness includes a suspicious value; P34-2: Using the set of sample monitoring image sequences and the set of sample personnel suspiciousness, a personnel suspiciousness identifier is trained based on a convolutional neural network; P34-3: The monitoring image sequences are input into the trained personnel suspiciousness identifier, and the identification output obtains the personnel suspiciousness.

[0049] Optionally, a suspicious person identifier can be trained using historical data to improve the system's accuracy in detecting suspicious behavior.

[0050] First, the system needs to collect a large set of sample surveillance image sequences from historical monitoring data. This data comes from the system's previous monitoring records, including image sequences extracted from all surveillance videos. Each image sequence contains consecutive video frames within a certain time range, representing a monitoring period.

[0051] During the acquisition of sample surveillance image sequences, the system identifies individuals in each sequence and assesses and labels whether their behavior is abnormal. To facilitate subsequent learning and training, each surveillance image sequence is accompanied by a suspicion level label, indicating whether the individuals in that sequence exhibit suspicious behavior. The suspicion level is a quantitative indicator, typically ranging from 0 to 1, where 0 represents no suspicious behavior and 1 represents highly suspicious behavior. For each sample surveillance image sequence, the system records its corresponding individual suspicion level, thus forming a complete set of sample individual suspicion levels.

[0052] Next, using the collected set of sample surveillance image sequences and the corresponding set of suspiciousness scores for individuals, a suspicious person identifier is constructed by training a convolutional neural network (CNN). A convolutional neural network is a deep learning algorithm, particularly suitable for image processing tasks. It automatically extracts features from images by simulating the neuronal connection mechanism of the human brain, thereby performing classification or regression tasks.

[0053] For example, the training process can begin by inputting the collected surveillance image sequence into a CNN model and performing data standardization (e.g., uniform image size and color) to ensure data consistency. Further, the CNN extracts spatial features (e.g., shape, texture, color) from the surveillance images layer by layer through multi-layer convolution and pooling operations, and combines this with temporal information to understand the dynamic behavior in each image frame. The suspiciousness label of the sample is used as the target value to help the network learn how to predict the corresponding suspicious value from the input surveillance image sequence. Through backpropagation, the system continuously adjusts the network's weight parameters based on the error between each prediction result and the true label until the network can accurately identify and classify suspicious behavior. After training, the network can output a suspiciousness value for a person's behavior based on the input surveillance image sequence, providing a basis for decision-making.

[0054] After training is complete, the system inputs new surveillance image sequences into the already trained suspicious person identifier. The CNN network then automatically evaluates the behavior of individuals within the input image sequences and outputs a suspiciousness score based on the previous training. This suspiciousness score represents the degree of abnormality in the individual's behavior. If the suspiciousness score is high, the system may trigger an alarm, indicating that the individual may be a potential threat. For example, if the system detects abnormal behavior in the image sequence (such as stealing items or lingering in inappropriate places), the output suspiciousness score will be close to 1. Conversely, if the behavior is normal, the output suspiciousness score will be close to 0.

[0055] Through this process, the monitoring image recognition module 30 can more intelligently identify suspicious behavior, promptly detect potential security threats, and effectively optimize the system's alarm strategy.

[0056] The burglar alarm module 40 is used to calculate the security level based on the correctness of the corrected password and the suspiciousness of the person, and to trigger a burglar alarm when the security level is less than a preset security level threshold.

[0057] Furthermore, the burglar alarm module 40 is also used to perform the following steps: P41: Subtract the personnel's suspiciousness from 1 to obtain the personnel's security level; P42: Perform a weighted calculation on the correctness of the corrected password and the personnel's security level to obtain the security level; P43: Determine whether the security level is less than a preset security level threshold. If yes, then trigger an anti-theft alarm; otherwise, do not trigger an anti-theft alarm.

[0058] Specifically, the core function of the burglar alarm module 40 in this application is to calculate the security level based on the correctness of the corrected password and the suspiciousness of the person, and to trigger an alarm when the security level is lower than a preset threshold.

[0059] First, based on the suspiciousness level of personnel obtained from the monitoring image recognition module 30, the personnel safety level is calculated. Suspiciousness level is a quantitative indicator representing abnormal personnel behavior; a higher value indicates more suspicious behavior. To obtain the personnel safety level, the system uses a simple method of subtracting the suspiciousness level from 1. This means that if a person's suspiciousness level is high (e.g., close to 1), the personnel safety level is low (close to 0), indicating that the person's behavior may pose a threat. Conversely, if a person's behavior is normal (suspiciousness level close to 0), the safety level is close to 1, indicating that the person does not pose a threat to safety.

[0060] Next, a weighted calculation is performed on the corrected password accuracy and personnel security level to arrive at the final security score. The purpose of this process is to combine these two pieces of information to assess the overall security of the system. Corrected password accuracy reflects the accuracy of the user's entered password; incorrect or non-standard password input may indicate a security vulnerability. Personnel security level reflects the degree of abnormal behavior identified after analyzing surveillance images. The weighted calculation assigns different weights to the two indicators, combining them into a final security assessment value. By adjusting these two weights, the system's sensitivity can be adjusted for different application scenarios. For example, in scenarios where password input accuracy is critical, the system can increase the weight of corrected password accuracy.

[0061] Finally, the system compares the calculated security level with a preset security threshold. The security threshold is a system-defined standard representing the minimum level of security the system considers acceptable. If the security level is below this threshold, the system considers a security risk and triggers an anti-theft alarm to alert relevant personnel. Conversely, if the security level is greater than or equal to the threshold, it indicates that the current password input and personnel behavior do not show any obvious abnormalities, and the system will not issue an alarm. Through these steps, the anti-theft alarm module 40 ensures that the system responds promptly and takes security measures when password input is abnormal or personnel behavior is suspicious, preventing potential security threats.

[0062] In summary, the embodiments of this application have at least the following technical effects: This application monitors the input password information, password input speed, and surveillance video when the smart door lock password authentication interface is started, identifies the password correctness, obtains an input proficiency coefficient based on the input speed, corrects the password correctness, configures the personnel identification step size based on the proficiency coefficient, extracts image sequences from the surveillance video and identifies personnel suspiciousness, calculates the security level based on the corrected password correctness and personnel suspiciousness, and triggers an anti-theft alarm if the security level is lower than the threshold.

[0063] This technology achieves the goal of conducting multi-level security assessments by integrating password accuracy and personnel suspiciousness, thereby reducing the risk of false alarms and missed alarms and improving the accuracy and reliability of home security alarms.

[0064] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0065] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0066] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A smart home security and anti-theft system based on the Internet of Things, characterized in that, The system includes smart door locks and monitoring equipment interconnected via the Internet of Things (IoT), and also includes: The input information monitoring module is used to monitor and obtain the input password information and password input speed information when the password authentication interface of the smart door lock is started, and to collect monitoring video through the monitoring device, and to identify and obtain the password correctness when the input password information is incorrect; The password correctness module is used to classify the input proficiency coefficient based on the password input speed information, obtain the input proficiency coefficient, correct the password correctness, and obtain the corrected password correctness. The surveillance image recognition module is used to configure the personnel recognition step size according to the input proficiency coefficient, extract surveillance images from the surveillance video, obtain a surveillance image sequence, and identify the suspiciousness of personnel. The burglar alarm module is used to calculate the security level based on the correctness of the corrected password and the suspiciousness of the person, and to trigger a burglar alarm when the security level is less than a preset security threshold.

2. The IoT-based smart home security and anti-theft system according to claim 1, characterized in that, The input information monitoring module is also used for: When the password authentication interface of the smart door lock is started, the system monitors and acquires the input password information and input time information, and collects monitoring video through the monitoring device. Based on the input time information, the password input speed information is calculated. Determine whether the input password information is correct. If yes, unlock the door; otherwise, verify the correctness of the input password information.

3. The IoT-based smart home security and anti-theft system according to claim 2, characterized in that, The input information monitoring module is also used for: Get the correct password; The similarity between the input password information and the correct password is identified to obtain the password correctness.

4. The home smart security and anti-theft system based on the Internet of Things according to claim 1, characterized in that, The password correctness verification module is also used for: Build an input proficiency classifier; The password input speed information is input into the input proficiency classifier, and the input proficiency coefficient is obtained by classification. The corrected password accuracy is obtained by multiplying the input proficiency coefficient by the password accuracy.

5. The IoT-based smart home security and anti-theft system according to claim 4, characterized in that, The password correctness verification module is also used for: Based on the historical usage data of the smart lock, a set of sample password input speed information of the smart lock is obtained, and the average password input speed information is calculated. The ratio of each sample password input speed information to the average password input speed information is calculated to obtain a set of sample input proficiency coefficients. Construct a mapping relationship between the sample password input speed information set and the sample input proficiency coefficient set to obtain an input proficiency classifier.

6. The IoT-based smart home security and anti-theft system according to claim 1, characterized in that, The monitoring image recognition module is also used for: Obtain a preset personnel identification step size, wherein the preset personnel identification step size includes the step size for extracting monitoring images from the monitoring video; The input proficiency coefficient is multiplied by the preset personnel recognition step size to obtain the personnel recognition step size; According to the personnel identification step size, multiple frames of monitoring images in the monitoring video are extracted to obtain a monitoring image sequence; The suspiciousness of individuals is determined by identifying the sequence of surveillance images.

7. The IoT-based smart home security and anti-theft system according to claim 6, characterized in that, The monitoring image recognition module is also used for: Based on historical monitoring and identification data, a set of sample monitoring image sequences is collected, and the suspiciousness of personnel in each sample monitoring image sequence is identified to obtain a set of sample personnel suspiciousness, wherein each sample personnel suspiciousness includes a suspicious value. Using the sample surveillance image sequence set and the sample personnel suspicion set, a personnel suspicion identifyr is trained based on a convolutional neural network; The surveillance image sequence is input into the trained personnel suspiciousness detector, and the suspiciousness level of the personnel is obtained from the identification output.

8. The IoT-based smart home security and anti-theft system according to claim 1, characterized in that, The burglar alarm module is also used for: The personnel safety score is obtained by subtracting the personnel's suspiciousness score from 1. The correctness of the modified password and the security level of the personnel are weighted and calculated to obtain the security level; Determine whether the security level is less than a preset security threshold. If yes, then activate the burglar alarm; otherwise, do not activate the burglar alarm.

Images