Human body posture detection method based on privacy protection camera and radar

Abstract

The invention discloses a human body posture detection method based on a privacy protection camera and a radar, which is applied to a human body posture detection system based on the privacy protection camera and the radar, and comprises the radar used for obtaining radar three-dimensional data and detecting an abnormal human body posture through the radar three-dimensional data; the privacy protection camera is arranged in the same monitoring area of the radar and is used for acquiring a fuzzy space image and detecting an abnormal human body posture through the fuzzy space image; the method comprises the steps that space radar three-dimensional data are collected through a radar, and radar human body posture features are obtained through calculation of the space radar three-dimensional data; a space blurred image is collected through the privacy protection camera, and human body posture features of the image are recognized through the space blurred image; and judging whether the current human body posture is abnormal or not according to the radar human body posture features and the image human body posture features, and if yes, giving an alarm.

Description

Technical Field

[0001] The present invention relates to the technical field of human body posture detection, and in particular to a human body posture detection method based on a privacy - protected camera and radar. Background Art

[0002] In a home environment, human body posture and behavior recognition technology has gradually become an important means to improve safety and quality of life. For example, by real - time monitoring of the activity state of the human body, intelligent home control, health risk reminder, and emergency warning can be realized.

[0003] In the prior art, microwave radar is usually used for human body posture detection. The radar judges the volume and dynamic characteristics of the target through signal reflection, and has advantages such as being insensitive to the light environment and having strong penetration. However, due to limited resolution, it is difficult to accurately capture the subtle movements or complex postures of the human body (such as gestures, joint position changes, and minute dynamic movements). In addition, in a multi - target environment, the radar is easily interfered by other moving objects (such as the vibration of home appliances, pet activities, etc.), resulting in an increase in data noise and a weak ability to distinguish the human body from other objects. The radar may also misjudge some fast but non - abnormal behaviors (such as sitting down, bending over) as falls, and may also miss real abnormal behaviors, so the overall detection accuracy and reliability need to be improved.

[0004] To make up for the deficiencies of the radar, there are also solutions in the prior art that combine radar and camera for human body detection. By fusing the millimeter - wave radar and camera image data, the judgment of abnormal postures such as human body falls is realized. However, there are significant differences in the data types and acquisition frequencies between the camera and the radar: the camera mainly provides two - dimensional image information, while the radar outputs three - dimensional spatial point cloud data. During the data fusion process, problems such as inconsistent time synchronization and spatial registration error are likely to occur, affecting the accuracy and real - time performance of human body posture recognition. In addition, ordinary cameras directly collect clear images of users. Especially in sensitive areas such as bedrooms and bathrooms, they are likely to expose privacy information such as faces, body features, and environmental details. Once the data is stored, transmitted, or attacked, there is a serious risk of privacy leakage and abuse, and it is difficult to meet the privacy compliance requirements in scenarios such as home, medical care, and elderly care.

[0005] Designing a human body posture detection method based on a privacy - protected camera and radar to address the above - mentioned problems in the prior art is the objective of the research of the present invention. Summary of the Invention

[0006] In view of this, the objective of the present invention is to propose a human body posture detection method based on a privacy - protected camera and radar, which can solve the above problems.

[0007] The present invention provides a human body posture detection method based on a privacy - protected camera and radar, which is applied to a human body posture detection system based on a privacy - protected camera and radar, and includes: A radar for acquiring radar three - dimensional data and detecting abnormal human body postures through the radar three - dimensional data; A privacy - protected camera disposed in the same monitoring area as the radar for acquiring a blurred space image and detecting abnormal human body postures through the blurred space image; The method includes: Collecting spatial radar three - dimensional data through the radar and calculating radar human body posture features through the spatial radar three - dimensional data; Collecting a spatial blurred image through the privacy - protected camera and identifying image human body posture features through the spatial blurred image; Judging whether the current human body posture is abnormal according to the radar human body posture features and the image human body posture features. If so, an alarm is issued.

[0008] Further, the privacy - protected camera includes a camera and a blurring structure, and the blurring structure is disposed on the shooting path of the camera for physically blurring the image acquired by the camera.

[0009] Further, the blurring structure is any one of a static blurring light - transmitting member, a modulation light - transmitting element, and a dynamic blurring structure.

[0010] Further, the step of collecting spatial radar three - dimensional data through the radar and calculating radar human body posture features through the spatial radar three - dimensional data includes: Obtaining the radar digital signal of the reflection signal in the current space through the radar, performing two - dimensional Fourier transform on the radar digital signal to obtain the velocity, distance, and angle information of the target point, calculating the three - dimensional space coordinates of the target point according to the velocity, distance, and angle information of the target point, and forming three - dimensional point cloud data (radar three - dimensional data); Performing multi - frame fusion on the three - dimensional point cloud data (radar three - dimensional data), performing human body point cloud clustering segmentation and target tracking on the fused point cloud data to obtain human body target point cloud data; Calculating radar human body posture features according to the clustered human body target point cloud data, and the radar human body posture features include any one or more of: the height of the point cloud centroid, the main axis direction and inclination angle, the size of the point cloud bounding box, the spatial distribution density, the projected area, the volume, the three - dimensional coordinates of key points, the dynamic change rate of the centroid or the main axis, and the stationary duration.

[0011] Further, the step of collecting a spatial blurred image through the privacy - protected camera and identifying image human body posture features through the spatial blurred image includes: Extract the feature points of the spatially blurred image using a feature detection algorithm, match the feature points with those in the reference space, and obtain the corresponding relationship of the feature points between the two; According to the corresponding relationship of the feature points, solve through a spatial geometry algorithm to obtain the three-dimensional coordinates of the human body posture in the image, and calculate the skeleton structure features based on the three-dimensional coordinates of the human body posture in the image.

[0012] Furthermore, judging whether the current human body posture is abnormal based on the radar human body posture features and the image human body posture features, and if so, giving an alarm includes: Judge whether the radar human body posture features are of an abnormal human body posture type. If so, obtain the image human body posture features at the same time point, and judge whether there is a human body posture in the image human body posture features at the same time point; If there is a human body posture in the spatially blurred image at the same time point, analyze the correlation between the image human body posture features and the radar human body posture features. If the two are similar, output the abnormal human body posture and give an early warning according to the abnormal human body posture.

[0013] Furthermore, analyzing the correlation between the image human body posture features and the radar human body posture features, and if the two are similar, outputting the abnormal human body posture includes: Convert the image human body posture features and the radar human body posture features to the same coordinate system, and calculate the mean square error of the three-dimensional coordinates of the image human body posture and the three-dimensional coordinates of the radar human body posture; If the mean square error of the three-dimensional coordinates of the image human body posture and the three-dimensional coordinates of the radar human body posture is less than the error threshold, the two are similar, and output the abnormal human body posture and its type.

[0014] Furthermore, analyzing the correlation between the image human body posture features and the radar human body posture features, and if the two are similar, outputting the abnormal human body posture includes: Input the image human body posture features into a pre-trained image posture classification model to obtain the image posture type; Judge whether the image posture type belongs to the abnormal human body posture type. If so, compare it with the abnormal human body posture type obtained from the radar human body posture features. If the two types are the same, output the abnormal human body posture and its type.

[0015] Furthermore, judging whether the current human body posture is abnormal based on the radar human body posture features and the image human body posture features, and if so, giving an alarm includes: Judge whether the image human body posture features are of an abnormal human body posture type. If so, obtain the radar human body posture features at the same time point, and judge whether there is a human body posture in the radar human body posture features at the same time point; If there is a human body posture in the radar human body posture features at the same time point, analyze the correlation between the image human body posture features and the radar human body posture features. If the two are similar, output the abnormal human body posture and give an early warning based on the abnormal human body posture.

[0016] Furthermore, judging whether the current human body posture is abnormal according to the radar human body posture features and the image human body posture features. If so, the alarm includes: Respectively judge whether the image human body posture features and the radar human body posture features are abnormal human body posture types. If both are abnormal human body posture types, analyze the correlation between the image human body posture features and the radar human body posture features; If the two are similar, output the abnormal human body posture and give an early warning based on the abnormal human body posture.

[0017] Advantages of the present invention: First, physical image blurring is collected through a privacy protection camera to ensure that no clear privacy images will be generated from the source. Even if the algorithm is maliciously tampered with or hacked, the original picture cannot be reconstructed, taking into account both the artificial intelligence function and user privacy protection. The radar and the privacy protection camera have their own advantages and disadvantages in perceiving the human body posture. Complementary to each other, the combination of the two can support diverse posture anomaly detection.

[0018] Second, accurate human body three-dimensional space data can be obtained through the radar, and human body two-dimensional contour information can be obtained through the privacy protection camera. Further, human body posture recognition and abnormal posture detection are carried out through the human body three-dimensional space data and the human body two-dimensional contour information.

[0019] Third, by first analyzing the correlation between the radar data and the privacy protection camera data and then judging the posture abnormality, compared with directly fusing all modal data without the need for consistency discrimination, it has higher robustness and false alarm suppression ability. Through the consistency analysis of the two types of data, unilateral noise and misjudgment can be effectively eliminated, preventing overall false alarms caused by local abnormalities, and significantly improving the system practicability and accuracy. This method is especially applicable to scenarios with multi-source data noise superposition, high privacy protection requirements, and the need for fine alarm in the actual environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0021] Figure 1 It is the system module diagram of Embodiment 1.

[0022] Figure 2 It is the flowchart of the method in Embodiment 2. Specific implementation manners

[0023] For the convenience of those skilled in the art to understand, the embodiments will be further described in detail in conjunction with the accompanying drawings for the structure of the present invention. It should be understood that the steps mentioned in this embodiment, unless specifically stating their order, can be adjusted in their front and back order according to actual needs, and even can be executed simultaneously or partially simultaneously.

[0024] Embodiment 1 As Figure 1 shown, Embodiment 1 provides a human body posture detection system based on a privacy protection camera and a radar, including: A radar, configured to obtain radar three-dimensional data and detect abnormal human body postures through the radar three-dimensional data; A privacy protection camera, disposed in the same monitoring area as the radar, configured to obtain a blurred space image and detect abnormal human body postures through the blurred space image; Specifically, the privacy protection camera includes a camera and a blur structure, and the blur structure is disposed on the shooting path of the camera for physically blurring the image obtained by the camera.

[0025] Further, the blur structure is any one of a static blur light-transmitting member, a modulated light-transmitting element, and a dynamic blur structure.

[0026] In this embodiment, traditional camera monitoring will involve privacy images such as faces, body features, and environmental details. Once the images are stored or transmitted, there is a risk of leakage or abuse, resulting in privacy security problems. By using the blur structure, the image can only present the general dynamics and contours of the monitored space, preventing the leakage of personal privacy. It is ensured that no clear privacy images will be generated from the source, and even if the algorithm is maliciously tampered with or attacked by hackers, the original picture cannot be reconstructed, taking into account both the artificial intelligence function and user privacy protection, and meeting the privacy compliance requirements of multiple scenarios such as home / medical / elderly care.

[0027] The radar and the privacy protection camera each have advantages and disadvantages in the perception of human body postures, and can complement each other after combination. For example, the radar is not sensitive to environmental light, while the privacy protection camera is better in distinguishing humanoid structures and spatial positioning. The two can be cross-checked to avoid missed detection or false detection caused by the failure of a single perception, and improve the overall recognition accuracy. In addition, the combination of the two can support diversified posture anomaly detection. For example, the radar can accurately detect the spatial features of postures such as "falling to the ground" and "getting up", and the privacy protection camera can detect "the details of abnormal movements and joint forms". After fusion, it can judge both large-scale movement situations and capture more detailed anomalies, such as twitching and abnormal waving.

[0028] The main objective of the static blurring light-transmitting component is to allow light to pass through while scattering it sufficiently so that the camera cannot image clear details. Such as frosted glass, frosted acrylic sheets, frosted polycarbonate sheets, atomized PET films, nano light-scattering coatings, etc. These components are fixedly arranged on the shooting path of the camera and achieve image blurring through their own light-transmitting and scattering properties.

[0029] Modulated light-transmitting components, including but not limited to components whose light-transmitting state or blurring degree can be controlled by means such as heating, electro-optic modulation, mechanical vibration, liquid flow, acoustic disturbance, etc. For example: heating-type modulation components (such as heating atomization films), electrochromic or electro-scattering intelligent dimming films, mechanical vibration or micro-perturbation films, liquid-modulated light-transmitting components, acoustic-modulated light-transmitting components, etc. For example, materials such as PET films and glass that can be surface-heated utilize physical state changes (such as atomization, water vapor condensation, optical disturbance, etc.) caused by heating to achieve dynamic blurring of images.

[0030] Dynamic blurring structures, such as high-speed mechanical brush pieces arranged on the shooting path of the camera, driving the brush pieces to reciprocate or rotate at a specific frequency, forming continuous optical trailing or occlusion effects through dynamic disturbances, making the captured images unrecognizable both in time and space.

[0031] Embodiment 2 Based on Embodiment 1, this embodiment provides a human body pose detection system based on a privacy protection camera and a radar, as Figure 2 shown. Embodiment 2 provides a human body pose detection method based on a privacy protection camera and a radar, including: S1 Collect spatial radar three-dimensional data through the radar, and calculate the radar human body pose characteristics from the spatial radar three-dimensional data; S101 Obtain the radar digital signal of the reflection signal in the current space through the radar, perform two-dimensional Fourier transform on the radar digital signal to obtain the velocity, distance, and angle information of the target point, calculate the three-dimensional space coordinates of the target point according to the velocity, distance, and angle information of the target point, and form radar three-dimensional data; S102 Perform multi-frame fusion on the radar three-dimensional data, perform human body point cloud clustering segmentation and target tracking on the fused data to obtain human body target point cloud data; S103 Calculate the radar human body pose characteristics according to the clustered human body target point cloud data. The radar human body pose characteristics include any one or more of: the height of the point cloud centroid, the main axis direction and inclination angle, the size of the point cloud bounding box, the spatial distribution density, the projected area, the volume, the three-dimensional coordinates of the key points, the dynamic change rate of the centroid or the main axis, and the stationary duration.

[0032] In this step, multi-frame fusion can compensate for the sparsity of the point cloud. Clustering segmentation is a standard step in point cloud object detection, and object tracking facilitates the recognition of the continuous postures of the same human body. The spatial feature parameters of the human object are structured data descriptions of the distribution, shape, scale, direction, etc. of the human body point cloud object obtained through clustering and tracking in the three-dimensional space. Since different human postures (such as standing, sitting, falling to the ground, etc.) show obvious differences in terms of point cloud distribution, main axis direction, and spatial occupancy area, etc., it is possible to based on these features, use rule judgment or model reasoning to subsequently realize the recognition and classification of the target human body posture.

[0033] The radar human body posture features include but are not limited to: the height of the point cloud centroid, the main axis direction and inclination angle, the dimensions (length, width, height) of the point cloud bounding box, the spatial distribution density, the projected area, the volume, the three-dimensional coordinates of the key points, the dynamic change rate of the centroid or the main axis, the static duration, etc.

[0034] S2 collects spatially blurred images through a privacy-protected camera and identifies the human body posture features in the images through the spatially blurred images; S201 uses a feature detection algorithm to extract the feature points of the spatially blurred image, matches the feature points with the feature points in the reference space, and obtains the corresponding relationship of the feature points between the two; S202 solves according to the corresponding relationship of the feature points through a spatial geometric algorithm to obtain the three-dimensional coordinates of the human body posture in the image, and calculates the skeleton structure features through the three-dimensional coordinates of the human body posture in the image.

[0035] In addition, the spatially blurred image can also be input into a pre-trained posture recognition model to obtain the human body posture features in the image.

[0036] In this step, by using a spatial geometric algorithm (such as PnP) in combination with the known reference space coordinates, the three-dimensional human body posture features of the spatially blurred picture can be obtained. In addition, the three-dimensional human body posture features can also be automatically output through a pre-trained deep learning model, such as: spatial coordinates, to improve the accuracy of the human body posture information of the blurred image.

[0037] The human body posture features in the image include but are not limited to: the three-dimensional coordinate information of the key points of the human skeleton, and the skeleton structure features calculated therefrom (such as the distance, angle, main axis direction of the trunk, centroid position, movement speed, etc.) of the skeleton joint points.

[0038] S3 determines whether the current human body posture is abnormal based on the radar human body posture features and the human body posture features in the image, and issues an alarm if it is.

[0039] S301 determines whether the radar human body posture features are of an abnormal human body posture type. If so, it obtains the human body posture features in the image at the same time point and determines whether there is a human body posture in the human body posture features in the image at the same time point; In this step, the obtained posture types include: standing, walking, sitting, squatting, kneeling, bending, raising hands, lying down, falling, prone, supine, stationary, etc.; among them, posture types such as falling, lying down, prone, supine, and long-term stationary are usually regarded as abnormal human posture types.

[0040] The abnormal types of radar human posture features can be judged by the abnormal thresholds of radar human posture features. For example, if the centroid height is lower than 0.3 meters, the main axis inclination angle is less than 20 degrees, and the projected area is greater than 1.0 square meters, it is determined as the 'lying / falling' posture; if the centroid height is higher than 1.2 meters and the main axis inclination angle is greater than 60 degrees, it is determined as the 'normal standing'. Within a specific time interval, if the change rate of the centroid or the main axis direction and height exceeds the preset threshold and is accompanied by a sudden change in height, the system can determine it as a 'falling' behavior. This part of the judgment method is the prior art and will not be elaborated in this application.

[0041] It can also be judged through a big data model. By using a deep learning model to construct a radar posture classification model, the radar human posture features and their calibrated human posture types can be used as training samples to train the abnormal judgment model.

[0042] S302 If there is a human posture in the spatial blurred image at the same time point, analyze the correlation between the image human posture features and the radar human posture features. If the two are similar, output the abnormal human posture and give an early warning according to the abnormal human posture.

[0043] S3021 Convert the image human posture features and the radar human posture features into the same coordinate system, and calculate the mean square error of the three-dimensional coordinates of the image human posture and the three-dimensional coordinates of the radar human posture; S3022 If the mean square error of the three-dimensional coordinates of the image human posture and the three-dimensional coordinates of the radar human posture is less than the error threshold, the two are similar, and output the abnormal human posture and its type.

[0044] In this step, the image human posture features are input into the image posture classification model after preprocessing to identify the specific posture type of the current human body. Through coordinate system conversion, the human posture features in different modalities are unified to the same spatial reference. Through the alignment of three-dimensional spatial features and the comparison of the mean square error (MSE), only when the two results are highly consistent will the abnormal result be output, effectively filtering out false alarms caused by single detection misdetection. If direct data fusion judgment is carried out, the two sets of data may judge different postures, resulting in inaccurate postures finally judged by fusion.

[0045] In addition, the correlation between the image human posture features and the radar human posture features can also be analyzed through a big data model. Specifically: Input the human body pose features of the image into a pre-trained image pose classification model to obtain the image pose type; Determine whether the image pose type belongs to the abnormal human body pose type. If so, compare it with the abnormal human body pose type obtained from the radar human body pose features. If the two types are the same, output the abnormal human body pose and its type.

[0046] In this step, the pose types corresponding to the radar human body pose features and the pose types corresponding to the image human body pose features are respectively obtained through a pre-trained big data model. Only when the two types are the same, the abnormal pose and its type are output, and warnings are issued according to the type, reducing the false alarm and missed alarm risks of single modality, and enhancing the anti-interference ability and device self-check ability of the system. Compared with the mean square error method, the big data model can monitor the development trend of actions, give early warnings, and make dynamic judgments.

[0047] Embodiment III Based on Embodiment I, this embodiment provides a human body pose detection system based on a privacy protection camera and a radar. The difference between this embodiment and Embodiment II is that it first determines whether there is an abnormality in the image human body pose features, and if so, then performs an abnormality judgment on the radar human body pose features. Specifically: Determine whether the image human body pose features are of an abnormal human body pose type. If so, obtain the radar human body pose features at the same time point, and determine whether there is a human body pose in the radar human body pose features at the same time point; If there is a human body pose in the radar human body pose features at the same time point, analyze the correlation between the image human body pose features and the radar human body pose features. If the two are similar, output the abnormal human body pose and give a warning according to the abnormal human body pose.

[0048] In this embodiment, the specific method for determining the human body pose type and the abnormal human body pose type through the image human body pose features is the same as that in Embodiment II, the specific method for determining the human body pose type and the abnormal human body pose type through the radar human body pose features is the same as that in Embodiment II, and the correlation analysis method between the image human body pose features and the radar human body pose features is the same as that in Embodiment II.

[0049] Embodiment IV Based on Embodiment I, this embodiment provides a human body pose detection system based on a privacy protection camera and a radar. The differences between this embodiment and Embodiment II and Embodiment III are that it respectively determines whether there is an abnormality in the image human body pose features and the radar human body pose features, and then combines the judgments to determine whether there is a pose abnormality. Specifically: Respectively determine whether the image human body pose features and the radar human body pose features are of an abnormal human body pose type. If both are of an abnormal human body pose type, analyze the correlation between the image human body pose features and the radar human body pose features; If the two are similar, an abnormal human posture is output, and a warning is given according to the abnormal human posture.

[0050] In this embodiment, the specific method for judging the human posture type and the abnormal human posture type through the image human posture feature is the same as that in the second embodiment. The specific method for judging the human posture type and the abnormal human posture type through the radar human posture feature is the same as that in the second embodiment. The correlation analysis method for the image human posture feature and the radar human posture feature is the same as that in the second embodiment.

[0051] Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0052] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram can be realized by computer program instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the functions specified in one Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0053] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured product including an instruction device, and the instruction device realizes the functions specified in one Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0054] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in one Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0055] It should be noted that in the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in the claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention can be implemented by means of hardware including several different elements and by means of a suitably programmed computer. In a unit claim listing several devices, several of these devices can be embodied by the same item of hardware. The use of the words first, second, and third, etc. does not denote any order. These words can be interpreted as names.

[0056] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made by those skilled in the art once they learn of the basic inventive concept. Therefore, the appended claims are intended to be construed to cover the preferred embodiments as well as all changes and modifications falling within the scope of the present invention.

[0057] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these changes and modifications.

[0058] In the present invention, unless otherwise clearly defined and limited, the terms "mounted", "connected", "coupled", "fixed", etc. shall be construed in a broad sense. For example, it can be a fixed connection, a detachable connection, or integrated; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the internal communication of two elements or the interaction relationship between two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0059] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms should not be understood as necessarily referring to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

Claims

1. A human body posture detection method based on a privacy-protected camera and radar, characterized in that, Applied to a human body pose detection system based on a privacy - protected camera and radar, including: A radar for acquiring radar three - dimensional data and detecting abnormal human body poses through the radar three - dimensional data; A privacy - protected camera disposed in the same monitoring area as the radar for acquiring blurred space images and detecting abnormal human body poses through the blurred space images; The method includes: Collecting spatial radar three - dimensional data through the radar and calculating radar human body pose features from the spatial radar three - dimensional data; Collecting spatial blurred images through the privacy - protected camera and identifying image human body pose features from the spatial blurred images; Judging whether the current human body pose is abnormal based on the radar human body pose features and the image human body pose features. If so, an alarm is issued.

2. The human body posture detection method based on a privacy protection camera and radar according to claim 1, wherein, The privacy - protected camera includes a camera and a blur structure. The blur structure is disposed on the shooting path of the camera for physically blurring the images acquired by the camera.

3. The human body posture detection method based on a privacy protection camera and a radar according to claim 2, characterized in that, The blur structure is any one of a static blur light - transmitting member, a modulated light - transmitting element, and a dynamic blur structure.

4. A human body posture detection method based on a privacy protection camera and radar according to claim 1, characterized in that, The step of collecting spatial radar three - dimensional data through the radar and calculating radar human body pose features from the spatial radar three - dimensional data includes: Obtaining the radar digital signal of the reflected signal in the current space through the radar, performing two - dimensional Fourier transform on the radar digital signal to obtain the velocity, distance, and angle information of the target point, calculating the three - dimensional space coordinates of the target point based on the velocity, distance, and angle information of the target point, and forming three - dimensional point cloud data (radar three - dimensional data); Performing multi - frame fusion on the three - dimensional point cloud data (radar three - dimensional data), performing human point cloud clustering segmentation and target tracking on the fused point cloud data to obtain human target point cloud data; Calculating radar human body pose features based on the clustered human target point cloud data. The radar human body pose features include any one or more of: the height of the point cloud centroid, the main axis direction and inclination angle, the size of the point cloud bounding box, the spatial distribution density, the projected area, the volume, the three - dimensional coordinates of key points, the dynamic change rate of the centroid or the main axis, and the stationary duration.

5. A human body posture detection method based on a privacy protection camera and radar according to claim 1, characterized in that, The step of collecting spatial blurred images through the privacy - protected camera and identifying image human body pose features from the spatial blurred images includes: Using a feature detection algorithm to extract the feature points of the spatial blurred image, matching the feature points with the feature points in the reference space to obtain the corresponding relationship of the feature points between the two; Solving through a spatial geometric algorithm based on the corresponding relationship of the feature points to obtain the three - dimensional coordinates of the image human body pose, and calculating the skeleton structure features from the three - dimensional coordinates of the image human body pose.

6. A human body posture detection method based on a privacy protection camera and radar according to claim 1, characterized in that, The step of judging whether the current human body pose is abnormal based on the radar human body pose features and the image human body pose features. If so, an alarm is issued includes: Judging whether the radar human body pose features are of an abnormal human body pose type. If so, obtaining the image human body pose features at the same time point and judging whether there is a human body pose in the image human body pose features at the same time point; If there is a human body pose in the spatial blurred image at the same time point, analyzing the correlation between the image human body pose features and the radar human body pose features. If the two are similar, outputting the abnormal human body pose and giving an early warning according to the abnormal human body pose.

7. A human body posture detection method based on a privacy-protected camera and radar according to claim 6, characterized in that, Analyze the correlation between the human body posture features in the image and the human body posture features detected by the radar. If the two are similar, output the abnormal human body postures, including: Convert the human body posture features in the image and the human body posture features detected by the radar to the same coordinate system, and calculate the mean square error between the three-dimensional coordinates of the human body posture in the image and the three-dimensional coordinates of the human body posture detected by the radar. If the mean square error between the three-dimensional coordinates of the human body posture in the image and the three-dimensional coordinates of the human body posture detected by the radar is less than the error threshold, it means the two are similar, and output the abnormal human body posture and its type.

8. A human body posture detection method based on a privacy protection camera and radar according to claim 6, characterized in that, Analyze the correlation between the human body posture features in the image and the human body posture features detected by the radar. If the two are similar, output the abnormal human body postures, including: Input the human body posture features in the image into a pre-trained image posture classification model to obtain the image posture type. Judge whether the image posture type belongs to the abnormal human body posture type. If so, compare it with the abnormal human body posture type obtained from the human body posture features detected by the radar. If the two types are the same, output the abnormal human body posture and its type.

9. The human body posture detection method based on a privacy-protected camera and radar according to claim 1, characterized in that, Judge whether the current human body posture is abnormal based on the human body posture features detected by the radar and the human body posture features in the image. If so, issue an alarm, including: Judge whether the human body posture features in the image are of the abnormal human body posture type. If so, obtain the human body posture features detected by the radar at the same time point, and judge whether there is a human body posture in the human body posture features detected by the radar at the same time point. If there is a human body posture in the human body posture features detected by the radar at the same time point, analyze the correlation between the human body posture features in the image and the human body posture features detected by the radar. If the two are similar, output the abnormal human body posture and issue a warning based on the abnormal human body posture.

10. A human body posture detection method based on a privacy protection camera and radar according to claim 1, characterized in that, Judge whether the current human body posture is abnormal based on the human body posture features detected by the radar and the human body posture features in the image. If so, issue an alarm, including: Respectively judge whether the human body posture features in the image and the human body posture features detected by the radar are of the abnormal human body posture type. If both are of the abnormal human body posture type, analyze the correlation between the human body posture features in the image and the human body posture features detected by the radar. If the two are similar, output the abnormal human body posture and issue a warning based on the abnormal human body posture.

Images