Infant monitor safety early warning method and system established based on three-dimensional scene

Abstract

The invention discloses a baby monitor safety early warning method and system established based on a three-dimensional scene, and the method comprises the steps: constructing an initial three-dimensional scene of a monitoring region; segmenting a target object in the initial three-dimensional scene to obtain contour information of the target object; extracting three-dimensional geometrical characteristic parameters of the target object; determining the type of each target object; marking a danger level for each target object; measuring and calculating the shortest distance between the infant and each target object in real time, and determining a corresponding safe distance threshold value; when it is detected that the shortest distance between the infant and any target object is smaller than a safety distance threshold value corresponding to the target object, a three-dimensional space movement track is obtained; calculating a collision risk coefficient based on the three-dimensional space motion trail; a comprehensive early warning index is obtained through multi-parameter weighted fusion operation; and when the comprehensive early warning index is greater than a preset threshold value, outputting a corresponding acousto-optic early warning signal according to an early warning level corresponding to the comprehensive early warning index. According to the invention, the accuracy of infant safety early warning is improved.

Description

Technical Field

[0001] The present application relates to the field of infant safety monitoring, and in particular to a three-dimensional scene-based baby monitor safety warning method and system. Background Art

[0002] Traditional baby monitors can only monitor through two-dimensional video images and cannot accurately determine the actual distance and spatial position relationship between the baby and surrounding objects. This can easily cause guardians to misjudge dangerous situations, thus affecting the baby's safety.

[0003] In related technologies, a simple three-dimensional scene can be created by acquiring depth information, which can calculate the distance between the baby and surrounding objects and issue an alarm when a dangerous distance is detected. This solution, to a certain extent, solves the problem that traditional two-dimensional monitoring cannot accurately determine spatial position relationships.

[0004] However, existing monitoring systems based on depth information have difficulty in fully understanding the semantics of the environment near the baby during the construction of three-dimensional scenes, making it difficult for the system to distinguish the danger levels of different objects, and thus difficult to provide targeted safety warnings. Summary of the Invention

[0005] The present application provides a baby monitor safety warning method and system based on a three-dimensional scene, which is used to improve the accuracy of baby safety warnings.

[0006] In a first aspect, the present application provides a baby monitor safety warning method based on a three-dimensional scene, which collects depth image information and color image information of the monitoring area; Based on the depth image information, the initial three-dimensional scene of the monitoring area is constructed through the spatial point cloud reconstruction algorithm; based on the color image information, the target object in the initial three-dimensional scene is segmented to obtain the contour information of the target object; Extracting the three-dimensional geometric feature parameters of the target object based on the contour information of the target object; Match the three-dimensional geometric feature parameters with the preset object feature database to determine the type of each target object in the monitoring area; Mark the danger level of each target object according to the preset danger level assessment rules; Calculate the shortest distance between the baby and each target object in real time, and determine the corresponding safe distance threshold based on the danger level of each target object; When it is detected that the shortest distance between the infant and any target object is less than a safety distance threshold corresponding to the target object, a three-dimensional spatial motion trajectory of the target object relative to the infant is obtained; Calculate the collision risk coefficient using the spatiotemporal collision prediction algorithm based on the three-dimensional motion trajectory; According to the collision risk coefficient, the danger level of the target object and the changing trend of the shortest distance, a comprehensive warning index is obtained through multi-parameter weighted fusion operation; When the comprehensive warning index is greater than a preset threshold, the corresponding sound and light warning signal is output according to the warning level corresponding to the comprehensive warning index.

[0007] By employing this technical solution, the system captures depth and color images of the monitored area and, in combination with a spatial point cloud reconstruction algorithm, constructs a 3D scene, accurately reconstructing the spatial structure and object distribution within the monitored area. 3D geometric feature parameters are extracted based on the target object's contours and matched against a pre-set database, enabling precise object identification and assessment of potential hazards. The system promptly detects potential hazardous contact by calculating the minimum distance between the infant and the object in real time and setting a corresponding safety threshold. When the distance falls below the threshold, the system accurately assesses the collision risk by analyzing the object's 3D trajectory and utilizing a spatiotemporal collision prediction algorithm. A comprehensive warning index is generated by weighted fusion of the collision risk factor, hazard level, and distance trend, providing a comprehensive picture of the current safety situation. Based on this comprehensive warning index, a graded warning signal is generated, preventing frequent triggering of warning signals while ensuring that caregivers take preventative measures before danger occurs, improving the accuracy and practicality of warnings.

[0008] In conjunction with some embodiments of the first aspect, in some embodiments, extracting three-dimensional geometric feature parameters of the target object based on the contour information of the target object specifically includes: Obtaining the three-dimensional coordinates of each surface point of the target object based on the contour information of the target object; Using triangulation algorithm to convert three-dimensional coordinate point set into polyhedral mesh structure; Calculate the normal vector and area of ​​each facet in the polyhedral mesh structure; Calculate the shape complexity of the target object based on the normal vector; Calculate the surface area of ​​the target object based on the area of ​​the polyhedral grid structure; Obtain the maximum distance of the target object on the three coordinate axes through the projection method and calculate the aspect ratio; The voxel counting method is used to calculate the volume of the target object and obtain the three-dimensional geometric feature parameters, which include surface area, aspect ratio, volume and shape complexity.

[0009] By adopting the above technical solution, the geometric shape of the object can be accurately reconstructed by obtaining the three-dimensional coordinates of the surface points of the target object and converting them into a polyhedral mesh structure using a triangulation algorithm. Calculating the normal vector and area of ​​each facet in the mesh structure helps to characterize the geometric features of the object's surface. Calculating the shape complexity through the normal vector can quantitatively characterize the object's potentially dangerous features such as edges and protrusions. Combining parameters such as surface area, aspect ratio, and volume, the geometric characteristics of the object can be fully described. These three-dimensional geometric feature parameters constitute a multidimensional feature space, enabling the system to more accurately identify and classify different types of objects and provide a reliable feature basis for the assessment of hazard levels. Compared with methods that rely solely on two-dimensional image features, analysis methods based on three-dimensional geometric features can more comprehensively assess the potential danger of objects and improve the accuracy of dangerous object identification and classification.

[0010] In conjunction with some embodiments of the first aspect, in some embodiments, calculating the collision risk coefficient based on the three-dimensional spatial motion trajectory using a spatiotemporal collision prediction algorithm specifically includes: Constructing a motion state vector according to the three-dimensional space motion trajectory, the motion state vector includes position coordinates, velocity and acceleration; The motion state vector is predicted based on the Kalman filter algorithm to obtain the state prediction values ​​at multiple sampling moments within the prediction time window; the shortest distance between the state prediction value at each sampling moment and the current position of the baby is calculated; According to the changing trend of the shortest distance, a distance-time function model is established using the polynomial fitting method. Calculate the collision time prediction value based on the distance-time function model; The collision risk coefficient is calculated by combining the collision time prediction value and the current motion trend.

[0011] By adopting the above technical solution, the motion characteristics of the object can be fully described by constructing a motion state vector containing position coordinates, velocity and acceleration. By using the Kalman filter algorithm to predict the state vector, multiple state prediction values ​​within the prediction time window can be obtained, thereby improving the accuracy of motion prediction. By calculating the shortest distance between the predicted state and the baby's position, and using polynomial fitting to establish a distance-time function model, the relative motion trend between the object and the baby can be accurately described. Based on the distance-time function model, the collision time prediction value is calculated, and the collision risk coefficient is calculated in combination with the current motion trend, which can accurately assess the risk level of potential collision. This risk assessment method based on spatiotemporal collision prediction not only takes into account the current spatial position relationship, but also takes into account the motion trend of the object, thereby improving the foresight and accuracy of the warning.

[0012] In conjunction with some embodiments of the first aspect, in some embodiments, after outputting an audible and visual warning signal corresponding to the warning level corresponding to the comprehensive warning index, the method further includes: Statistical time series of warning levels and corresponding danger levels; Calculate the warning frequency of each hazard level based on the time series; Identify high-incidence warning areas based on warning frequency; Set differentiated safety distance thresholds for high-incidence warning areas; Recalculate the comprehensive warning index based on differentiated safety distance thresholds.

[0013] By adopting the above technical solution, by statistically analyzing the time series of warning levels and the corresponding danger levels and calculating the warning frequency for each danger level, it is possible to identify areas in the system that are prone to danger. Setting differentiated safety distance thresholds for high-incidence warning areas enables more targeted warnings based on the risk characteristics of different areas. By recalculating the comprehensive warning index using differentiated safety distance thresholds, the system avoids excessive warnings while maintaining safety. This adaptive warning mechanism can dynamically adjust warning parameters based on actual usage, improving the practicality of the system. By continuously learning and optimizing warning parameters, the system makes the warning mechanism more consistent with the needs of actual usage scenarios, ensuring safety while improving the user experience.

[0014] In conjunction with some embodiments of the first aspect, in some embodiments, calculating the warning frequency of each danger level based on the time series specifically includes: Record the number of warnings for each danger level in each time series; Calculate the density of warning events per unit time; Determine the warning frequency based on the density of warning events.

[0015] By employing this technical solution, the system accurately quantifies the occurrence patterns of warning events at different hazard levels by recording the number of warnings issued for each hazard level within each time series and calculating the density of warning events per unit time. This calculation of warning event density takes into account the influence of the temporal dimension, avoiding the statistical bias that can result from simply accumulating the number of warnings. The warning frequency determined by warning event density reflects both the frequency of hazardous events and their temporal distribution, providing reliable data support for the system to establish a more accurate hazard warning model. This time-series-based warning frequency calculation method enables the system to more objectively assess the actual probability of occurrence of various hazardous situations, thereby optimizing the warning triggering mechanism, reducing false alarm rates, and improving the overall reliability and practicality of the warning system. Accurate calculation of warning frequency also provides a scientific basis for the subsequent setting of differentiated safety distance thresholds, enabling the system to dynamically adjust warning strategies based on actual conditions.

[0016] In conjunction with some embodiments of the first aspect, in some embodiments, after recalculating the comprehensive warning index based on the differentiated safety distance threshold, the method further includes: Record the transition time points between adjacent warning levels; Calculate the time interval between each increase in warning level based on the conversion time point; When the time interval is less than the preset duration, a higher level warning signal will be triggered in advance.

[0017] By adopting the above technical solution, the system can grasp the temporal patterns of the evolution of danger levels by recording the transition time points between adjacent warning levels and calculating the time intervals for warning level increases. When it is found that the time interval is less than the preset duration, a higher-level warning signal is triggered in advance, which can reserve more time for response in the event of a rapid deterioration of the dangerous situation. This time interval-based warning upgrade mechanism gives the system foresight, and can issue warnings in advance when the danger level is about to rise rapidly, avoiding the problem of delayed response of traditional warning systems. By analyzing the temporal characteristics of warning level transitions, the system can identify the pattern of rapid evolution of dangerous situations and adjust the timing of warning triggering accordingly, thereby ensuring the timeliness of warnings while maintaining system reliability. This improvement in the warning mechanism enables the system to better respond to sudden danger escalations, improving the real-time performance and protective effectiveness of the warning system.

[0018] In conjunction with some embodiments of the first aspect, in some embodiments, triggering a higher-level warning signal in advance specifically includes: Calculate the average value of the preset number of consecutive time intervals; Set half of the average value as the early warning lead time; When it is detected that the warning level begins to rise, the early warning advance amount outputs the next level warning signal; A higher level warning signal is triggered based on the next level warning signal.

[0019] By employing the above technical solution, the system establishes an adaptive early warning mechanism based on historical data by calculating the average of a preset number of consecutive time intervals and setting half of this average as the early warning lead time. This mechanism analyzes the temporal patterns of historical early warning escalations to determine an appropriate early warning lead time, avoiding the problem of setting the lead time too high or too low. When the system detects an increase in the early warning level, it outputs the next level of early warning signal in advance based on the calculated early warning lead time, thereby triggering a higher level of early warning signal, achieving a smooth transition between warning levels. This dynamic calculation method of the early warning lead time enables the system to automatically adjust the timing of early warnings based on actual operating data, ensuring the timeliness of early warnings while avoiding the potential interference caused by excessively early warnings. The early warning lead time, derived from historical data analysis, is highly adaptable and can automatically adjust as system operating conditions change, enhancing the scientific nature and practicality of the early warning mechanism.

[0020] In a second aspect, an embodiment of the present application provides a baby monitor safety warning system established based on a three-dimensional scene, and the baby monitor safety warning system established based on a three-dimensional scene includes: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, and the computer program code includes computer instructions, and one or more processors call the computer instructions to enable the system to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer-readable storage medium comprising instructions, which, when executed on a system, enables the system to execute the method described in the first aspect and any possible implementation of the first aspect.

[0022] In a fourth aspect, an embodiment of the present application provides a computer program product, characterized in that when the computer program product is run on a system, the system executes the method described in any possible implementation manner in the first aspect.

[0023] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. This application provides a baby monitor safety warning method based on a three-dimensional scene. By collecting depth and color image information of the monitored area and combining it with a spatial point cloud reconstruction algorithm to construct a three-dimensional scene, the system can accurately restore the spatial structure and object distribution within the monitored area. Based on the target object's outline information, three-dimensional geometric feature parameters are extracted and matched against a preset database, enabling precise identification of object types and assessment of their potential danger. The system calculates the minimum distance between the baby and the object in real time and sets a corresponding safety threshold, enabling timely detection of potential dangerous contact. When the distance falls below the threshold, the system accurately assesses the collision risk by analyzing the object's three-dimensional motion trajectory and utilizing a spatiotemporal collision prediction algorithm. A comprehensive warning index is generated by weighted fusion of multiple parameters, including the collision risk coefficient, danger level, and distance trend, providing a comprehensive reflection of the current safety situation. Outputting graded warning signals based on the comprehensive warning index not only avoids frequent triggering of warning signals but also ensures that caregivers are promptly alerted to take preventive measures before danger occurs, improving the accuracy and practicality of warnings.

[0024] 2. The present application provides a baby monitor safety warning method based on a three-dimensional scene. By statistically analyzing the time series of warning levels and the corresponding danger levels and calculating the warning frequency of each danger level, areas in the system prone to danger can be identified. Differentiated safety distance thresholds are set for high-incidence warning areas, so that more targeted warnings can be made based on the risk characteristics of different areas. By recalculating the comprehensive warning index through differentiated safety distance thresholds, the system avoids excessive warnings while maintaining safety. This adaptive warning mechanism can dynamically adjust the warning parameters according to actual usage, thereby improving the practicality of the system. The system continuously learns and optimizes the warning parameters to make the warning mechanism more in line with the needs of actual usage scenarios, while ensuring safety and improving the user experience.

[0025] 3. The present application provides a baby monitor safety warning method based on a three-dimensional scene. By recording the conversion time points between adjacent warning levels and calculating the time interval for the warning level to increase, the system can grasp the time pattern of the evolution of the danger level. When it is found that the time interval is less than the preset time length, a higher level warning signal is triggered in advance, which can reserve more response time when the dangerous situation deteriorates rapidly. This time interval-based warning upgrade mechanism enables the system to have predictiveness and can issue warnings in advance when the danger level is about to increase rapidly, avoiding the problem of delayed response of traditional warning systems. By analyzing the time characteristics of the warning level conversion, the system can identify the pattern of rapid evolution of dangerous situations and adjust the warning triggering timing accordingly, thereby ensuring the timeliness of the warning while maintaining the reliability of the system. The improvement of this warning mechanism enables the system to better respond to sudden danger upgrades and improves the real-time performance and protection effect of the warning system. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a flow chart of a baby monitor safety warning method based on a three-dimensional scene in an embodiment of the present application.

[0027] Figure 2 This is a flow chart of an adaptive optimization method based on historical warning data in an embodiment of the present application.

[0028] Figure 3 This is a schematic diagram of the physical device structure of a baby monitor safety warning system based on a three-dimensional scene provided in an embodiment of the present application. DETAILED DESCRIPTION

[0029] The terms used in the following examples of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification and appended claims of this application, the singular expressions "a," "an," "said," "above," "the," and "this" are intended to include plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in this application refers to any or all possible combinations comprising one or more of the listed items.

[0030] In the following, the terms "first" and "second" are used for descriptive purposes only and should not be understood to imply or suggest relative importance or implicitly indicate the number of the technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of this application, unless otherwise specified, "plurality" means two or more.

[0031] The following uses an embodiment and combines Figure 1 , a baby monitor safety warning method based on a three-dimensional scene in an embodiment of the present application is described: See also Figure 1 , which is a flow chart of a baby monitor safety warning method based on a three-dimensional scene in an embodiment of the present application.

[0032] S101, collecting depth image information and color image information of the monitoring area; The system uses depth cameras and color cameras installed in the monitored area to collect depth image information and color image information of the area, respectively. The depth camera can be based on principles such as structured light, ToF (Time of Flight), and binocular vision, and is used to obtain depth information of objects within the monitored area. The color camera can be a conventional visible light camera, used to obtain color images of the monitored area. The system can also use an RGBD camera, which integrates a depth camera and a color camera to simultaneously obtain depth and color images. In addition, the system can also select an appropriate camera layout based on actual needs, such as using multiple cameras to form a camera array to expand the monitoring range and improve the quality of the collected images.

[0033] When capturing image information, the system can set an appropriate sampling frequency and resolution. A higher sampling frequency helps capture details of an object's motion, while a higher resolution provides more spatial detail. The system can also preprocess the captured images, such as removing noise and correcting distortion, to improve the accuracy of subsequent processing. Furthermore, considering that changing lighting conditions may affect image acquisition quality, the system can employ adaptive exposure control technology to automatically adjust the camera's exposure parameters based on ambient light intensity to achieve stable and clear images.

[0034] In some special scenarios, such as low-light environments, ordinary visible light color cameras may have difficulty capturing clear images. To address this issue, the system can be equipped with an additional infrared camera, which uses infrared light sources to actively illuminate and obtain clear images even in low-light conditions. By fusing visible light and infrared images, the system can stably capture image information of the monitored area around the clock.

[0035] S102, constructing an initial three-dimensional scene of the monitoring area using a spatial point cloud reconstruction algorithm based on the depth image information; segmenting the target object in the initial three-dimensional scene based on the color image information to obtain contour information of the target object; The system uses the acquired depth image information to construct a 3D scene model of the monitored area using a spatial point cloud reconstruction algorithm. Common spatial point cloud reconstruction algorithms include Poisson surface reconstruction, Marching Cubes, and Alpha Shapes. These algorithms transform the 3D coordinates of the pixels in the depth image to obtain the corresponding 3D point cloud data. They then use the spatial relationship information between the point clouds to construct a triangular mesh surface, forming an initial 3D model of the monitored area. The system can also refine and smooth the initial 3D model using algorithms such as moving least squares and Laplace smoothing to obtain a more refined and smooth 3D scene model.

[0036] After constructing the initial 3D scene, the system uses color image information to segment the target objects within the scene. Common target segmentation algorithms include region growing-based segmentation, graph cut-based segmentation, and clustering-based segmentation. The system can group pixels belonging to the same object into the same region based on features such as color, texture, and edges, thereby obtaining the contours of each object. Furthermore, given the dynamic nature of the surveillance scene, the system can also incorporate motion detection and tracking algorithms to accurately segment and identify targets by analyzing the motion changes of objects between consecutive frames.

[0037] In some complex scenes, a single object segmentation algorithm may struggle to accurately extract object contours, for example due to interfering factors such as occlusion and shadows between objects. To address this issue, the system can employ a multi-feature fusion segmentation strategy. By comprehensively leveraging various feature information such as an object's color, texture, depth, and motion, a robust object segmentation model is constructed to improve segmentation accuracy in complex scenes. Furthermore, the system can incorporate artificial intelligence technologies, such as deep learning algorithms, which, through training and learning from massive amounts of data, automatically extract and identify object contour features, further enhancing the intelligent level of segmentation.

[0038] S103, extracting three-dimensional geometric feature parameters of the target object based on the contour information of the target object; The system extracts the three-dimensional geometric feature parameters of the target object based on the contour information of the target object. Specifically: the three-dimensional coordinates of each surface point of the target object are obtained based on the contour information of the target object; Using triangulation algorithm to convert three-dimensional coordinate point set into polyhedral mesh structure; Calculate the normal vector and area of ​​each facet in the polyhedral mesh structure; Calculate the shape complexity of the target object based on the normal vector; Calculate the surface area of ​​the target object based on the area of ​​the polyhedral grid structure; Obtain the maximum distance of the target object on the three coordinate axes through the projection method and calculate the aspect ratio; The voxel counting method is used to calculate the volume of the target object and obtain the three-dimensional geometric feature parameters, which include surface area, aspect ratio, volume and shape complexity.

[0039] After obtaining the contour information of the target object, the system further extracts the three-dimensional geometric feature parameters of the object for subsequent object identification and risk assessment. Three-dimensional geometric feature parameters usually include the object's size, shape, topological structure and other attributes. To extract these feature parameters, the system first constructs a triangular mesh model of the object's surface based on the three-dimensional coordinates of the object's contour points. Then, the system can calculate geometric quantities such as the normal vector and curvature of the triangular facet to obtain parameters reflecting the shape characteristics of the object, such as surface area, volume, convex hull, shape complexity, etc. In addition, by analyzing the span of the object on the three coordinate axes, the system can also calculate the aspect ratio of the object to reflect its size characteristics.

[0040] When extracting three-dimensional geometric features, the system can select different feature descriptors and calculation methods according to actual needs. For example, in order to characterize the shape complexity of an object, the system can use shape complexity indicators to quantify the complexity of its shape by calculating the average curvature and Gaussian curvature of the object's surface. In order to accurately calculate the volume of an object, the system can use triangulation to discretize the object's surface into a polyhedral mesh model, and then obtain the object's volume through integral calculation. In addition, the system can also introduce more advanced geometric shape analysis methods, such as harmonic analysis and wavelet analysis, to extract the multi-scale shape features of the object.

[0041] In some cases, there may be subtle defects or irregularities on the surface of an object, resulting in inaccurate extracted geometric features. To address this problem, the system can adopt a feature extraction method based on statistical learning. By collecting a large amount of 3D model data of different objects, the system can use machine learning algorithms, such as support vector machines and random forests, to automatically learn and extract the key geometric features of the objects. This data-driven feature extraction method can effectively adapt to the shape changes of different objects and improve the robustness and generalization ability of feature extraction. At the same time, the system can also introduce shape matching and recognition technology in graphics to achieve rapid classification and recognition of objects by comparing the similarity between the geometric features of the object and the standard model.

[0042] S104, matching the three-dimensional geometric feature parameters with a preset object feature database to determine the type of each target object in the monitoring area; After extracting the target object's three-dimensional geometric feature parameters, the system matches these feature parameters with a preset object feature database to determine the object's specific type. This object feature database stores standard three-dimensional geometric features for various common objects, such as size ranges and shape descriptors. The system calculates the similarity between the target object's feature parameters and the features of each object category in the database to identify the most suitable object type. Common feature matching algorithms include nearest neighbor algorithms, support vector machines, and decision trees. The system can select appropriate matching strategies based on actual needs, such as Euclidean distance metrics or Mahalanobis distance metrics, to measure the similarity between feature parameters.

[0043] When identifying object types, the system can employ multi-feature fusion, leveraging information such as the object's size, shape, and texture to improve recognition accuracy. For example, the system can combine 3D geometric features with 2D image features (such as color histograms and SIFT features) to construct a multimodal object feature representation, enhancing recognition reliability. Furthermore, the system can incorporate rule-based reasoning mechanisms to refine and optimize initial recognition results based on prior knowledge such as spatial relationships between objects and their functional attributes.

[0044] Given the diversity of actual monitoring scenarios, a pre-set object feature database may not cover all possible object types. To address this open-set recognition problem, the system can employ an incremental learning strategy to dynamically update and expand the feature database. When an object of an unknown category is identified, the system can prompt the user to label it and add the object's feature parameters to the database, continuously enriching and improving the coverage of object categories. Furthermore, the system can utilize unsupervised learning algorithms, such as cluster analysis and anomaly detection, to automatically discover and identify new object categories, enabling adaptive recognition of unknown objects.

[0045] S105, marking the danger level of each target object according to the preset danger level assessment rules; After identifying the types of target objects in the monitoring area, the system marks the corresponding hazard level for each object based on the preset hazard level assessment rules. Hazard level assessment rules generally take into account the inherent properties and potential risk factors of the object, such as size, material hardness, sharpness, etc. For example, large and hard objects may have a higher collision risk, while sharp and fragile objects may cause injuries such as cuts and punctures. The system can pre-assess the hazard of different object categories and develop a set of hazard level classification standards. Then, based on the object type information, the system automatically matches the corresponding hazard level and marks the hazard level for it, such as high risk, medium risk, low risk, etc.

[0046] When assessing an object's hazard level, the system can comprehensively consider multiple factors. In addition to the object's inherent properties, the system can also analyze environmental factors such as its spatial location and placement. For example, unstable objects placed around a crib may pose a higher potential hazard. Furthermore, the system can incorporate risk assessment models, such as fault tree analysis and Bayesian networks, to comprehensively assess the probability of an object's hazard and the consequences of its damage, resulting in a more comprehensive and quantitative hazard level.

[0047] Considering that different families may have different criteria for assessing danger levels, the system provides personalized risk assessment rule configuration. Users can adjust and optimize the risk assessment rules based on their own security needs and priorities. For example, users can set stricter risk thresholds for specific types of objects or define special risk assessment strategies for certain areas. Through personalized configuration, the system can better adapt to different users' security preferences and provide more personalized and detailed safety warning services.

[0048] S106: Calculate the shortest distance between the infant and each target object in real time, and determine a corresponding safe distance threshold based on the danger level of each target object; The system analyzes the coordinates of the infant and target objects in three-dimensional space and calculates the Euclidean distance between them in real time, determining the shortest distance between the infant and each object. Simultaneously, the system determines whether the infant is in a safe zone based on a pre-set safety distance threshold. The size of the safety distance threshold is related to the danger level of the object; the higher the danger level, the larger the corresponding safety distance threshold. For example, for high-risk objects, the system may require the infant to maintain a greater distance from them to ensure safety; for low-risk objects, the system may set a smaller safety distance threshold, allowing the infant to approach within a certain range.

[0049] When determining the safe distance threshold, the system can take into account multiple factors, such as the infant's age, developmental stage, and behavioral abilities. For example, for younger infants, due to their limited mobility, the system can relax the safe distance threshold accordingly; whereas for older infants, due to their greater curiosity and exploratory abilities, the system may need to set stricter safe distance limits. Furthermore, the system can dynamically adjust the safe distance threshold based on the infant's historical behavioral patterns and interaction habits, providing more personalized and intelligent safety protection.

[0050] Considering the potential for occlusion and interference in real-world environments, which can lead to errors or inaccuracies in distance measurement, the system can incorporate multi-sensor fusion technology to address this issue. This technology leverages data from multiple sensors, including visual, infrared, and ultrasonic sensors, to cross-validate and optimize distance information, improving the stability and reliability of distance measurement. Furthermore, the system can employ algorithms such as Kalman filtering and particle filtering to predict and track the movement of infants and objects, continuously estimating the distance between them even in the presence of brief occlusions, ensuring consistent safety monitoring.

[0051] S107: When it is detected that the shortest distance between the infant and any target object is less than a safety distance threshold corresponding to the target object, obtaining a three-dimensional spatial motion trajectory of the target object relative to the infant; When the system detects that the distance between an infant and a target object is less than the corresponding safety distance threshold for that object, it indicates that the infant may be facing a potential safety risk. At this point, the system begins to focus on the target object and obtain its three-dimensional motion trajectory relative to the infant. By analyzing continuous depth image frames, the system can extract information about the object's position changes in three-dimensional space and reconstruct its motion trajectory. Common motion trajectory extraction methods include tracking based on optical flow and tracking based on feature point matching. The system can select an appropriate tracking algorithm, combining the object's appearance features and motion pattern, to accurately estimate its motion state in three-dimensional space.

[0052] When extracting an object's motion trajectory, the system can employ a multi-scale, layered tracking strategy. First, the system roughly tracks the object within a large spatiotemporal range, quickly identifying its approximate direction of motion and position changes. Then, the system performs refined motion analysis within a localized area, extracting detailed motion parameters such as velocity, acceleration, and rotation. This layered tracking approach allows the system to capture the object's motion details at varying spatiotemporal granularities, ensuring both real-time tracking and accurate motion state estimation.

[0053] Considering the potential for occlusion and overlap during the movement of infants and objects, which can lead to interruptions or confusion in the extraction of motion trajectories, the system can introduce constraint-based motion estimation technology to address this issue. By incorporating prior knowledge, such as the continuity and smoothness of an object's motion, the system can perform constrained optimization of the motion trajectory, filling in trajectory gaps during occlusions and eliminating noise interference during tracking. Furthermore, the system can utilize prediction algorithms such as Kalman filtering and particle filtering to predict the object's future motion trends based on its historical motion state, thereby improving the consistency and stability of the motion trajectory.

[0054] S108. Calculate the collision risk coefficient using a spatiotemporal collision prediction algorithm based on the three-dimensional spatial motion trajectory; The system calculates the collision risk coefficient based on the three-dimensional space motion trajectory using a space-time collision prediction algorithm, specifically including: Constructing a motion state vector according to the three-dimensional space motion trajectory, the motion state vector includes position coordinates, velocity and acceleration; The motion state vector is predicted based on the Kalman filter algorithm to obtain the state prediction values ​​at multiple sampling moments within the prediction time window; the shortest distance between the state prediction value at each sampling moment and the current position of the baby is calculated; According to the changing trend of the shortest distance, a distance-time function model is established using the polynomial fitting method. Calculate the collision time prediction value based on the distance-time function model; The collision risk coefficient is calculated by combining the collision time prediction value and the current motion trend.

[0055] After determining the three-dimensional trajectory of the target object relative to the infant, the system uses a spatiotemporal collision prediction algorithm to assess potential collision risks. This algorithm analyzes the movement trends of the object and infant over a period of time to determine whether a collision is likely and quantify the risk of collision.

[0056] Specifically, the system first constructs motion models of the object and infant based on their current motion states, such as position, velocity, and acceleration. The system then projects these motion models forward into the future, generating predicted positions at different moments. By comparing the distances between the predicted positions, the system can determine whether there is a likelihood of a collision between the object and infant.

[0057] To quantify collision risk, the system comprehensively assesses the severity of the collision by calculating parameters such as collision time, collision speed, and collision angle. For example, shorter collision time and higher collision speed generally indicate a higher collision risk. The system can design a multi-dimensional collision risk assessment model that maps these parameters to a scalar collision risk coefficient, intuitively reflecting the degree of collision risk.

[0058] During the actual prediction process, the system employs techniques such as Monte Carlo simulation and probabilistic graphical models to comprehensively account for the uncertainty and randomness of motion. By generating a large number of motion trajectory samples, the system can estimate the probability distribution of collisions, resulting in a more robust and reliable collision risk assessment. Furthermore, the system can update the motion status of the object and infant in real time, dynamically adjusting the collision prediction model to adapt to the changes and uncertainties in the actual scenario.

[0059] Considering that the shape, size, material, and other properties of different objects can affect the consequences and severity of a collision, the system also needs to comprehensively consider these factors when calculating the collision risk factor. For example, for sharp, hard objects, even a relatively low collision velocity can cause serious injury, so the system needs to increase the collision risk factor accordingly. The system can establish a comprehensive collision hazard assessment model to estimate the potential harm to the infant based on the object's attributes and collision parameters, providing a more comprehensive and accurate basis for subsequent safety warnings.

[0060] S109, obtaining a comprehensive warning index through a multi-parameter weighted fusion operation based on the collision risk coefficient, the danger level of the target object, and the changing trend of the shortest distance; After calculating the collision risk factor, the system also considers other relevant factors, such as the danger level of the target object and the changing trend of the shortest distance between the infant and the object, to arrive at a final comprehensive warning index. The comprehensive warning index is a quantitative safety risk assessment that reflects the overall safety threat level faced by the infant in the current situation.

[0061] To calculate the comprehensive warning index, the system uses a multi-parameter weighted fusion method. First, the system normalizes parameters such as collision risk factor, hazard level, and distance trend, mapping them to the same numerical scale. Then, the system assigns a weight to each parameter, indicating its importance in the comprehensive warning index. Through weighted summation or other fusion operations, the system generates a comprehensive warning index value.

[0062] When determining the weight coefficients of each parameter, the system can take into account a variety of factors. For example, the collision risk coefficient should generally be given a higher weight because it directly reflects the possibility and degree of danger of a collision. The danger level of the target object is also an important reference factor. The higher the danger level of the object, the larger the corresponding weight coefficient should be. In addition, the changing trend of the shortest distance can also provide valuable information. If the distance shows a trend of rapid shortening, it may mean that the risk is rapidly increasing and needs to be given a higher weight.

[0063] S110. When the comprehensive warning index is greater than a preset threshold, a corresponding sound and light warning signal is output according to the warning level corresponding to the comprehensive warning index.

[0064] When the system's calculated comprehensive warning index exceeds a preset threshold, it indicates the current security risk has reached a critical level, necessitating prompt action. Based on the range of the comprehensive warning index, the system can categorize warning levels into low, medium, and high. Each warning level corresponds to a distinct audio and visual warning signal, alerting users to potential safety hazards and enabling them to take appropriate countermeasures.

[0065] When outputting audio and visual warning signals, the system can use different warning methods based on the warning level. For example, for low-risk levels, the system may only need to use a soft warning sound or flashing indicator light to alert the user. For high-risk levels, the system may issue a more eye-catching and rapid alarm sound, accompanied by a warning light with a faster flashing frequency, to attract the user's immediate attention.

[0066] In addition to audio and visual warning signals, the system can also deliver warning information through other means. For example, the system can promptly send warning information to the baby's parents or other designated emergency contacts through push notifications, text messages, and phone calls. This way, even if the parent is temporarily away from the baby, they can still be informed of the baby's safety risk and take necessary intervention measures remotely.

[0067] In the above-described embodiment, by collecting depth and color image information of the monitored area and combining it with a spatial point cloud reconstruction algorithm to construct a three-dimensional scene, the spatial structure and object distribution within the monitored area can be accurately restored. By extracting three-dimensional geometric feature parameters based on the target object's contour information and matching them with a preset database, the system can accurately identify the object type and assess its potential danger. By calculating the minimum distance between the infant and the object in real time and setting a corresponding safety threshold, the system can promptly detect potential dangerous contact. When the distance is less than the threshold, the collision risk can be accurately assessed by analyzing the object's three-dimensional motion trajectory and utilizing a spatiotemporal collision prediction algorithm. A comprehensive warning index is generated by weightedly integrating the collision risk coefficient, danger level, and distance change trend through multi-parameter fusion, providing a comprehensive reflection of the current safety situation. Outputting graded warning signals based on the comprehensive warning index not only avoids frequent triggering of warning signals but also ensures that caregivers are promptly alerted to take preventive measures before danger occurs, thereby improving the accuracy and practicality of warnings.

[0068] It can be seen from the first embodiment above that the basic early warning process can realize real-time monitoring and early warning of the baby's safety status. However, in actual applications, different areas may have different frequencies and patterns of dangerous occurrences, and the use of a unified safety distance threshold may affect the accuracy of the early warning. In order to further improve the early warning effect of the system, this application also provides an adaptive optimization method based on historical early warning data. This method analyzes the time distribution characteristics of early warning events and implements differentiated early warning strategies for different areas, so that the system's early warning mechanism can better adapt to the actual use environment. The following is combined with Figure 2 , an adaptive optimization method based on historical warning data in an embodiment of the present application is described: See also Figure 2 , which is a flow chart of an adaptive optimization method based on historical warning data in an embodiment of the present application.

[0069] S201. Calculate the time series of warning levels and corresponding danger levels; In this step, the system compiles statistics on the time series of warning levels and the corresponding danger levels. Specifically, the system can obtain warning data over a period of time, including the occurrence time of each warning event, the warning level, and the corresponding danger level. The system can then arrange this data in chronological order to form a time series of warning levels. The system can also record the danger level corresponding to each warning event and establish a mapping relationship between warning levels and danger levels.

[0070] To achieve statistical analysis of warning level time series and hazard levels, the system can employ a variety of data processing techniques. For example, the system can use database technology to store and manage warning data, extracting the required time series and hazard level information through SQL queries. The system can also utilize data structures and algorithms, such as linked lists and arrays, to efficiently organize and process warning data. Furthermore, the system can apply techniques such as time series analysis and data mining to uncover hidden patterns and regularities within warning data.

[0071] When compiling the time series of warning levels and danger levels, the system may encounter data quality issues such as missing or inconsistent data. To address this, the system can introduce a data preprocessing mechanism to clean, complete, and standardize the raw warning data. For example, if missing warning times or danger levels are missing, the system can make reasonable estimates and fill in the missing data based on contextual information. For inconsistent data formats, the system can convert and unify them using techniques such as regular expressions. Through data preprocessing, the system can improve the quality and reliability of warning data, laying the foundation for subsequent analysis and optimization.

[0072] S202. Calculate the warning frequency of each hazard level based on the time series; The system calculates the warning frequency of each hazard level based on the time series, including: Record the number of warnings for each danger level in each time series; Calculate the density of warning events per unit time; Determine the warning frequency based on the density of warning events.

[0073] In this step, the system calculates the warning frequency of each hazard level based on the warning level time series obtained in step S201. The warning frequency reflects the frequency of warning events of different hazard levels and is an important indicator for evaluating the performance of the warning system and optimizing the warning strategy.

[0074] To calculate the warning frequency, the system first needs to record the number of warnings for each hazard level within each time series. This can be achieved by traversing the time series and counting the number of occurrences of each hazard level. Next, the system calculates the density of warning events per unit time, which is the number of warnings divided by the length of the time series. Finally, the system determines the warning frequency based on the warning event density. Different density thresholds can be set to correspond to different warning frequency levels.

[0075] When calculating warning frequencies, the system can employ sliding window technology to dynamically adjust the range of the time series. This allows it to capture changing trends in warning event patterns and promptly update warning frequencies. Furthermore, the system can incorporate time weighting, assigning higher weights to recent warning events to reflect the current warning status.

[0076] S203, identifying high-incidence warning areas based on warning frequencies; In this step, the system uses the warning frequencies for each hazard level calculated in step S202 to identify areas with a high incidence of warning events. High-incidence warning areas are geographic locations or locations where the frequency of warning events is significantly higher than in other areas. By identifying these high-incidence warning areas, the system can take targeted preventive and response measures, improving the effectiveness of warnings.

[0077] To identify high-incidence warning areas, the system can correlate warning frequency with geographic location. Specifically, the system clusters warning events by location and calculates the average warning frequency within each cluster. The system then compares the warning frequencies across clusters and identifies areas with significantly higher frequencies than other clusters, marking them as high-incidence warning areas.

[0078] During implementation, the system can utilize spatial database technology to store and manage geographically attributed warning data. It can also apply spatial clustering algorithms, such as DBSCAN and K-Means, to geographically cluster warning events. Furthermore, the system can incorporate statistical methods, such as hypothesis testing and confidence interval estimation, to assess the significance of differences in warning frequency between clusters.

[0079] S204. Setting differentiated safety distance thresholds for high-incidence warning areas; In this step, the system sets differentiated safety distance thresholds for the high-risk warning areas identified in step S203. The safety distance threshold is the minimum safe distance between an infant and the dangerous area. When the distance falls below this threshold, the system triggers an alert. By setting differentiated thresholds for different areas, the system can more precisely control the triggering conditions for alerts, improving their relevance and effectiveness.

[0080] To set differentiated safety distance thresholds, the system needs to comprehensively consider multiple factors. First, the system must analyze the risk level of high-risk warning areas, setting lower thresholds for high-risk areas and higher thresholds for low-risk areas. Second, the system must also consider infant behavioral characteristics, dynamically adjusting thresholds based on factors such as the infant's movement speed and range of activity. Furthermore, the system can incorporate risk assessment models, combining historical warning data and expert knowledge, to quantify the risk level of different areas and provide a reference for threshold setting.

[0081] During implementation, the system can create a profile for safe distance thresholds, assigning an appropriate threshold to each high-incidence warning area. The system can also design a threshold adjustment algorithm to dynamically optimize threshold settings based on real-time warning feedback and infant behavior data. For example, if the frequency of warnings in a certain area consistently exceeds expectations, the system can automatically lower the safe distance threshold for that area, increasing the sensitivity of the warnings.

[0082] S205. Recalculate the comprehensive warning index based on the differentiated safety distance threshold.

[0083] In the final step, the system recalculates the comprehensive warning index based on the differentiated safety distance thresholds set in step S204. The comprehensive warning index is a quantitative indicator used to assess the overall safety risk of the infant's environment. By introducing differentiated thresholds, the system can more accurately assess the safety conditions in different areas and generate a more targeted comprehensive warning index.

[0084] To calculate the comprehensive early warning index, the system needs to collect and process multi-source, heterogeneous early warning data. In addition to distance warning data based on differentiated thresholds, the system can also incorporate other types of early warning data, such as infant physiological condition warnings and behavioral abnormalities. The system can assign weights to different types of early warning data, determining their impact on the overall safety risk. The system then takes the weighted sum of these various types of warning data to generate a comprehensive early warning index.

[0085] During implementation, the system can design a warning data fusion framework to standardize and integrate warning data from different sources and formats. The system can also establish a calculation model for a comprehensive warning index, defining weight coefficients and calculation formulas for various types of warning data. Furthermore, the system can incorporate machine learning algorithms to automatically optimize the calculation method for the comprehensive warning index by training historical warning data and security incident data.

[0086] In the above embodiment, by statistically analyzing the time series of warning levels and the corresponding danger levels and calculating the warning frequency of each danger level, areas in the system that are prone to danger can be identified. By setting differentiated safety distance thresholds for high-incidence warning areas, more targeted warnings can be made based on the risk characteristics of different areas. By recalculating the comprehensive warning index through differentiated safety distance thresholds, the system avoids excessive warnings while maintaining safety. This adaptive warning mechanism can dynamically adjust the warning parameters according to actual usage, thereby improving the practicality of the system. By continuously learning and optimizing the warning parameters, the system makes the warning mechanism more in line with the needs of actual usage scenarios, improving the user experience while ensuring safety.

[0087] Furthermore, in another embodiment, after recalculating the comprehensive warning index according to the differentiated safety distance threshold, the method further includes: Record the transition time points between adjacent warning levels; Calculate the time interval between each increase in warning level based on the conversion time point; When the time interval is less than the preset time, a higher level of warning is triggered in advance, including: Calculate the average value of the preset number of consecutive time intervals; Set half of the average value as the early warning lead time; When it is detected that the warning level begins to rise, the early warning advance amount outputs the next level warning signal; A higher level warning signal is triggered based on the next level warning signal.

[0088] The system records the time when the warning level changes, that is, the moment when it switches from one warning level to the next higher or lower level. By tracking the time when the warning level changes, the system can analyze the development trend and dynamic characteristics of the warning event.

[0089] To record the time of transitions, the system can compare the current index with the warning level corresponding to the previous index after each calculation of the comprehensive warning index. If the two warning levels differ, the current time is marked as the transition time. The system can store the transition time and the corresponding warning level in a log file or database for subsequent analysis and query.

[0090] To calculate time intervals, the system can traverse the warning level conversion log and extract all time points when the warning level was raised. The system then sorts these time points in chronological order and calculates the time difference between adjacent warning level increase time points, which is the time interval between warning level increases. The system can store the calculated time interval sequence for subsequent analysis and decision-making. The system introduces an early warning trigger mechanism, which triggers higher-level warnings in advance based on the time interval between warning level increases. If the time interval between multiple consecutive warning level increases is less than the preset duration, the system determines that the warning event may enter a rapid development stage and requires early action.

[0091] To trigger an early warning, the system first sets a preset duration threshold to determine whether an interval is too short. The system then counts a preset number of consecutive intervals (e.g., three) and calculates their average. If the average is less than the preset duration threshold, the system triggers the early warning mechanism.

[0092] In the early warning mechanism, the system sets half the average time interval as the early warning lead, which is the length of time before the warning is triggered. When the system detects that the current warning level begins to increase, it advances the warning lead by the next level and outputs the next level of warning signal. This allows the system to take countermeasures before the warning event fully develops to a higher level.

[0093] Finally, the system triggers the corresponding higher-level warning based on the next-level warning signal output in advance. This can be achieved by calling the warning signal generation and output module, which uses the pre-generated warning signal as input to generate corresponding warning prompts and control instructions.

[0094] It's important to note that when setting the lead time for early warnings, the system needs to balance the timeliness and accuracy of early warnings. If the lead time is set too high, it may lead to excessive warnings; if it is set too low, it may not buy enough time for response measures. Therefore, the system can adaptively adjust the lead time by analyzing historical warning data to find the optimal balance. Furthermore, the system can incorporate a manual intervention mechanism, allowing users to manually adjust the lead time based on actual conditions to address specific warning scenarios.

[0095] In the above embodiment, by recording the transition time points between adjacent warning levels and calculating the time intervals for the warning level to increase, the system can grasp the time pattern of the evolution of the danger level. When it is found that the time interval is less than the preset time length, a higher-level warning signal is triggered in advance, which can reserve more response time in the event of a rapid deterioration of the dangerous situation. This time interval-based warning upgrade mechanism enables the system to be predictive and can issue warnings in advance when the danger level is about to increase rapidly, avoiding the problem of delayed response of traditional warning systems. By analyzing the time characteristics of the warning level conversion, the system can identify the pattern of rapid evolution of the dangerous situation and adjust the warning triggering timing accordingly, thereby ensuring the timeliness of the warning while maintaining the reliability of the system. This improvement in the warning mechanism enables the system to better respond to sudden danger upgrades and improves the real-time performance and protection effect of the warning system.

[0096] The following describes the system in the embodiment of the present invention from the perspective of hardware processing. Figure 3 , which is a schematic diagram of the physical device structure of a baby monitor safety warning system based on a three-dimensional scene provided in an embodiment of the present application.

[0097] It should be noted that Figure 3 The structure of the system shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0098] like Figure 3 As shown, the system includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes, such as the methods described in the above embodiments, based on programs stored in a read-only memory (ROM) 302 or programs loaded from a storage unit 308 into a random access memory (RAM) 303. RAM 303 also stores various programs and data required for system operation. CPU 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to bus 304.

[0099] The following components are connected to the I / O interface 305: an input section 306 including a camera, infrared sensor, and the like; an output section 307 including a liquid crystal display (LCD) and speakers; a storage section 308 including a hard disk and the like; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. Removable media 311, such as a magnetic disk, optical disk, magneto-optical disk, or semiconductor memory, is installed in the drive 310 as needed, so that computer programs read from the media can be installed in the storage section 308 as needed.

[0100] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for executing the methods illustrated in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via the communication section 309 and / or installed from removable media 311. When executed by the central processing unit (CPU) 301, the computer program performs the various functions defined in the present invention.

[0101] It should be noted that the computer-readable medium described in the embodiments of the present invention may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present invention, a computer-readable signal medium may include a data signal transmitted in baseband or as part of a carrier wave, which carries a computer-readable computer program. Such a propagated data signal may take any of a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof.

[0102] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram can represent a module, program segment, or part of the code, and the above-mentioned module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the boxes can also occur in an order different from that marked in the accompanying drawings. For example, two boxes shown in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.

[0103] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the system described in the above embodiments, or may exist independently and not incorporated into the system. The storage medium carries one or more computer programs, and when executed by a processor of a system, the system implements the methods provided in the above embodiments.

[0104] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0105] As used in the above embodiments, the term “when” may be interpreted to mean “if” or “after” or “in response to determining that” or “in response to detecting that”, depending on the context. Similarly, the phrases “upon determining that” or “if (stated condition or event) is detected” may be interpreted to mean “if determining that” or “in response to determining that” or “upon detecting (stated condition or event)” or “in response to detecting (stated condition or event)”, depending on the context.

[0106] In the above embodiments, all or part of the embodiments can be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., a floppy disk, hard disk, tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive).

[0107] Those skilled in the art will appreciate that all or part of the process steps in the above-described method embodiments can be implemented by a computer program instructing the relevant hardware. The program can be stored in a computer-readable storage medium, and when executed, the program can include the process steps in the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A baby monitor safety warning method based on a three-dimensional scene, characterized in that: include: Collect depth image information and color image information of the monitored area; Constructing an initial three-dimensional scene of the monitoring area through a spatial point cloud reconstruction algorithm according to the depth image information; Segmenting the target object in the initial three-dimensional scene based on the color image information to obtain contour information of the target object; Extracting three-dimensional geometric feature parameters of the target object according to the contour information of the target object; Matching the three-dimensional geometric feature parameters with a preset object feature database to determine the type of each target object in the monitoring area; Marking a danger level for each target object according to a preset danger level assessment rule; Calculating the shortest distance between the infant and each target object in real time, and determining a corresponding safe distance threshold based on the danger level of each target object; When it is detected that the shortest distance between the infant and any of the target objects is less than a safety distance threshold corresponding to the target object, obtaining a three-dimensional spatial motion trajectory of the target object relative to the infant; Calculating a collision risk coefficient using a spatiotemporal collision prediction algorithm based on the three-dimensional spatial motion trajectory; Obtaining a comprehensive warning index through multi-parameter weighted fusion calculation based on the collision risk coefficient, the danger level of the target object, and the change trend of the shortest distance; When the comprehensive warning index is greater than a preset threshold, a corresponding sound and light warning signal is output according to the warning level corresponding to the comprehensive warning index.

2. The method according to claim 1, characterized in that Extracting the three-dimensional geometric feature parameters of the target object according to the contour information of the target object specifically includes: Acquiring three-dimensional coordinates of each surface point of the target object based on the contour information of the target object; Converting the three-dimensional coordinate point set into a polyhedral mesh structure using a triangulation algorithm; Calculating the normal vector and area of ​​each facet in the polyhedral mesh structure; Calculating the shape complexity of the target object according to the normal vector; Calculating the surface area of ​​the target object based on the area of ​​the polyhedral grid structure; Obtain the maximum distance of the target object on the three coordinate axes by projection method, and calculate the aspect ratio; The volume of the target object is calculated using a voxel counting method to obtain three-dimensional geometric feature parameters, which include the surface area, the aspect ratio, the volume, and the shape complexity.

3. The method according to claim 1, characterized in that The calculating of the collision risk coefficient based on the three-dimensional space motion trajectory using a spatiotemporal collision prediction algorithm specifically includes: Constructing a motion state vector according to the three-dimensional space motion trajectory, wherein the motion state vector includes position coordinates, velocity and acceleration; Predicting the motion state vector based on a Kalman filter algorithm to obtain state prediction values ​​at multiple sampling moments within a prediction time window; calculating the shortest distance between the state prediction value at each sampling moment and the current position of the infant; According to the changing trend of the shortest distance, a distance-time function model is established by using a polynomial fitting method; Calculating a collision time prediction value based on the distance-time function model; The collision risk coefficient is calculated by combining the collision time prediction value and the current motion trend.

4. The method according to claim 1, wherein After outputting the corresponding sound and light warning signal according to the warning level corresponding to the comprehensive warning index, the method further includes: Counting the time series of the warning levels and the corresponding danger levels; Calculating the warning frequency of each of the danger levels according to the time series; Identify high-incidence warning areas based on the warning frequency; Setting differentiated safety distance thresholds for the high-incidence warning areas; The comprehensive warning index is recalculated according to the differentiated safety distance threshold.

5. The method according to claim 4, characterized in that Calculating the warning frequency of each danger level according to the time series specifically includes: Recording the number of warnings for each danger level within each time series; Calculate the density of warning events per unit time; The warning frequency is determined according to the warning event density.

6. The method according to claim 4, characterized in that After recalculating the comprehensive warning index according to the differentiated safety distance threshold, the method further includes: Recording the transition time points between adjacent warning levels; Calculate the time interval for each increase in warning level based on the conversion time point; When the time interval is less than the preset time length, a higher level warning signal is triggered in advance.

7. The method according to claim 6, characterized in that The triggering of a higher level early warning signal in advance specifically includes: Calculate the average value of the time interval for a preset number of consecutive times; Half of the average value is set as the early warning lead time; When it is detected that the warning level begins to increase, outputting the next level warning signal in advance by the warning lead amount; A higher level warning signal is triggered according to the lower level warning signal.

8. A baby monitor safety warning system based on a three-dimensional scene, characterized in that: The system comprises: One or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to cause the system to execute the method according to any one of claims 1 to 7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a system, the system is caused to perform the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product is run on a system, the system is caused to perform the method according to any one of claims 1 to 7.

Images