Household elderly remote monitoring alarm system with voice alarm and activity detection functions

Abstract

The invention provides a household elderly remote monitoring alarm system with voice alarm and activity detection, which comprises an infrared monitoring module, a preprocessing module, a risk assessment module, a tumble detection module, an alarm module and a mobile control terminal, and is characterized in that whether the elderly tumble or not in the activity process is found and detected by analyzing the tumble factors of the elderly; a model is trained through a deep learning algorithm according to falling risk factors of a plurality of falling factors and motion change characteristics related to falling, and the falling risk of the current target is comprehensively evaluated by analyzing the surrounding environment where the target is located and the motion characteristics of the target, so that the accuracy of judging the falling condition of the old people is improved to a great extent; the problem of inaccurate detection results caused by no comprehensive consideration is effectively solved, so that the purpose of improving the accuracy of judging the falling condition of the old people is achieved, the problem of low motion detection precision in the prior art is effectively solved, and the purpose of improving the target detection accuracy is achieved.

Description

Technical Field

[0001] The invention relates to the technical field of health monitoring for the elderly, and in particular to a household elderly remote monitoring alarm system with voice alarm and activity detection. Background Art

[0002] With the intensification of aging social problems and the miniaturization of family structures, the safety of the elderly has attracted widespread attention. The decline of physiological functions in the elderly will lead to health problems such as physical disability and cognitive impairment, thereby reducing the mobility and self-care ability of the elderly. In addition, elderly people living alone or at an advanced age find it difficult to solve emergencies by themselves, thus falling into dangerous situations. In order to enable the elderly to be discovered and rescued in time after an accidental fall at home, a lot of research has been conducted on the elderly fall detection and rescue system. The detection and evaluation of the fall status of the elderly is an important part of the effectiveness of the entire system. Due to the different physical health conditions of the elderly, the risks of complications after falls of elderly people with different physical conditions are different. Therefore, for the application scenarios of the elderly fall rescue system, it is necessary to classify the fall status of the elderly and the severity of falls in different fall categories. Therefore, a method for monitoring abnormal activities of elderly people falling is proposed. Through this method, the elderly's abnormal activities of falling and wrestling are monitored in real time, and when the abnormal activities of falling of the elderly are detected, an alarm signal is issued in time to protect the health of the elderly.

[0003] The prior art utilizes the point cloud data of a 3D depth camera and uses human body features to build a model. Based on the human body modeling, the joints of the human body are detected. The depth information of the human body is collected to calculate the height information of the human body and perform fall detection.

[0004] Disadvantages of the existing technology 1: 3D depth cameras are expensive and difficult to deploy in general home environments. In addition, the amount of calculation of point cloud data is huge, making it difficult to run offline in general embedded devices. Cloud deployment and other methods must be adopted, resulting in the outflow of private data.

[0005] The second existing technology uses an ordinary RGB camera to collect human RGB and depth thermal imaging data, uses a wearable device to collect acceleration sensor information of human joints, and analyzes the human portrait based on the joints of the human body model to obtain the human joints, thereby realizing the judgment of falls.

[0006] The second disadvantage of the existing technology: wearable devices cause inconvenience to users' normal lives, are difficult to wear for a long time, are not user-friendly, and require a lot of calculations for thermal imaging data analysis, making them difficult to deploy in the cloud, resulting in secondary leakage of user data.

[0007] The third disadvantage of the existing technology is that it needs to collect thermal imaging data and sound information, among which the sound information is easily disturbed by the environment, and it is difficult to detect the help request issued by the elderly when the elderly's voice is small and the background noise is loud.

[0008] To this end, we propose a home elderly remote monitoring alarm system with voice alarm and activity detection. Summary of the invention

[0009] The purpose of the present invention is to provide a home elderly remote monitoring alarm system with voice alarm and activity detection to solve the problems raised in the above background technology.

[0010] To achieve the above-mentioned object, the present invention provides the following technical solutions: a home-use elderly remote monitoring alarm system with voice alarm and activity detection, comprising an infrared monitoring module, a preprocessing module, a risk assessment module, a fall detection module, an alarm module and a mobile control terminal; Among them, the infrared monitoring module is connected to the preprocessing module to obtain the thermal imaging data of the elderly activity scene; the preprocessing module is respectively connected to the risk assessment module and the fall detection module to process the thermal imaging data of the elderly activity scene to obtain the preprocessed thermal imaging data; the risk assessment module is used to assess the current risk of falling of the elderly based on the preprocessed thermal imaging data, the fall detection module is used to determine whether the elderly has fallen based on the target human body of the preprocessed thermal imaging data, the alarm module is respectively connected to the risk assessment module and the fall detection module, and the mobile control terminal is used to receive the elderly status information and issue a warning or help based on the elderly status information; The preprocessing module includes a data acquisition unit, a data collection system, a data upload system, and a data storage warehouse; The risk assessment module includes a data analysis system, a result determination system, and a threshold import system; The fall detection module includes an abnormal alarm system, a data retrieval system, a central control terminal, and a remote connection system.

[0011] Preferably, the data acquisition unit is connected to the data collection system and is used to collect and gather data of the elderly on site; the data acquisition unit includes a first data acquisition module, which is connected to the data collection system and is used to obtain thermal imaging data of the elderly's daily activities; the data acquisition unit includes a second data acquisition module, which is connected to the data collection system and is used to obtain thermal imaging data of the elderly's daily activities; The data collection system is connected to the data upload system to organize the collected data of the elderly and transmit it to the data upload system; The data uploading system is connected to the data storage warehouse and is used to upload the elderly data to the data storage warehouse; The data storage warehouse is connected to the data retrieval system and is used to store the elderly data input by the data retrieval system; The data storage warehouse is connected to the data analysis system, the data analysis system is connected to the result determination system, and the data analysis system performs remote monitoring through data analysis; The data analysis system is connected to the threshold import system and the abnormal alarm system respectively, and is used to compare the data analysis with the database threshold and issue an alarm; The data analysis system is connected to the central control terminal to control and process the central data information; The central control terminal is connected to the remote connection system for remote monitoring and processing of data information analyzed by the data analysis system; The data analysis system is connected to the mobile control terminal and is used to send the data information analyzed by the data analysis system to family members through the mobile control terminal.

[0012] Preferably, the fall detection module is used to detect whether the elderly falls during activities, and the detection method is: Step A: receiving thermal imaging data, subtracting two consecutive adjacent frames of thermal imaging data from each other, and obtaining a frame difference image corresponding to the two adjacent frames of thermal imaging data, wherein the non-zero pixel points in the frame difference image are suspected moving target pixel points; Step B: Intersection processing is performed on the suspected moving target pixel points in the frame difference image and the preset moving area. If there is no intersection, it is determined that a moving target is detected; Step C: Using the optical flow algorithm to calculate the moving direction and speed of the moving target in the thermal imaging data; Step D: Determine the motion region based on the frame difference map and non-zero pixel points, mark and connect the motion regions in the continuous frame differences by using the connected domain method, and obtain the centroid of the motion region; Step E: Track the centroid of the motion area and obtain the pixel distance between the two centroids. If the pixel distance is greater than a preset threshold, it is determined that the moving target has fallen. The fall detection module is used to detect and send out an alarm signal when the risk assessment module determines that the elderly has fallen.

[0013] Preferably, the fall detection module further comprises a difference analysis algorithm based on continuous frames to obtain the frame difference image, and the difference analysis algorithm formula based on continuous frames is: D(xc,yc)=|It(xc,yc)-I(ta,xc,yc)|; Among them, It(xc,yc) and I(ta,xc,yc) are the pixel values ​​of the position in the two consecutive frames before and after (assuming t is the current frame and ta is the previous frame), and D(xc,yc) represents the pixel value at the position (xc,yc) in the frame difference map, which is equal to the absolute value of the difference between the pixel values ​​at the same position in the two frames before and after.

[0014] Preferably, the risk assessment module includes a fall risk factor acquisition module, which is used to acquire risk factors related to falls. For falls of the elderly, the falling behavior and environmental factors are taken into account. The fall risk factor acquisition module is based on the thermal imaging data of the infrared monitoring module, and counts the motion target characteristics within 30 seconds before the elderly fall over a period of time to acquire a set of risk factors related to falls.

[0015] Preferably, the data analysis system includes a target detection module, a target tracking module and a target motion statistics module, wherein the target detection module extracts the characteristic area of ​​the preprocessed thermal imaging data through a target detection algorithm, the target tracking module is used to track according to the characteristic area, and the target motion statistics module is used to analyze the motion change characteristics of the characteristic area within a period of time; the target tracking module tracks the characteristic area obtained by the target detection module, and extracts the motion characteristics of the target area during the tracking process, wherein the motion characteristics include moving speed, moving distance and moving area; the target motion statistics module counts the motion characteristics of the target area within a certain period of time, and the analysis includes the moving speed characteristic score, the moving distance characteristic score and the moving area characteristic score, and then comprehensively scores the comprehensive motion change characteristics of the target area, and the comprehensive scoring formula is as follows: S = Σ(Si × Wi); Among them: S is the comprehensive score, which represents the overall evaluation of the comprehensive motion change characteristics of the target area; Si is the score of each feature, representing the performance of the target area in a specific motion change feature; i is the number of features, indicating the total number of features involved in the comprehensive scoring; Wi is the weight of each feature, indicating the importance of the feature in the comprehensive score.

[0016] Preferably, the data analysis system further includes a fall risk factor processing module, which combines the fall risk factor with the motion change characteristics obtained by the target motion statistics module to obtain comprehensive information of the motion change characteristics and different fall risk factors, and calculates the fall risk score of the elderly based on the comprehensive information. The fall risk score formula is as follows: A=aθ+bV+cL+dArea; Among them: A fall risk score is a comprehensive assessment result used to quantify the risk of falls; θ Elderly posture characteristic score, which reflects factors such as posture stability and coordination of the elderly at a specific time point; V is the moving speed feature score; L is the motion distance feature score; Area is the motion area feature score.

[0017] Preferably, the data acquisition unit simultaneously collects the posture, behavior and multiple continuous thermal imaging data of the elderly's activities, and the posture feature score includes scores corresponding to different fall risk factors. Corresponding to the posture feature score is the factor feature of the corresponding fall risk factor obtained by the fall risk factor acquisition module, and the factor feature includes arm posture feature, leg posture feature, body inclination feature and footstep size feature. Combined with the factor features, the posture feature score related to falls is obtained through comparison and comprehensive calculation. The posture feature score formula related to falls is as follows: θ_total=θ1×k1+θ2×k2+θ3×k3+θ4×k4; Among them: θ1 is the arm posture feature score, which reflects the impact of the arm posture or movement state on the risk of falling; θ2 is the characteristic score of leg posture, which reflects the relationship between leg posture or movement and fall risk; θ3 is the body tilt feature score, which evaluates the effect of body tilt angle or posture change on the risk of falling; θ4 is the step feature score, which involves factors such as step stability and step size; k1, k2, k3, and k4 are weight factors of the corresponding features, which determine the relative importance of each feature in the overall score.

[0018] Preferably, the alarm module includes a voice alarm system, which is connected to the alarm module, and the voice alarm system sends a voice alarm signal according to the fall risk score level and the movement change characteristics, and the voice alarm system includes: A voice playback module is used to match the sound features according to the fall risk score level and the movement change characteristics to obtain a voice alarm signal; A voice receiving module is used to collect voice information and transmit the voice information to a voice server; The voice judgment module is used to collect environmental noise and sound intensity to determine whether the current environment is suitable for voice alarm; A voice optimization module, used to optimize the acquired voice information; A speech coding module, used for compressing and coding the acquired speech information; A speech decoding module is used to decode and restore speech information; The speech synthesis module is used to synthesize speech according to the acquired speech alarm signal and in combination with the user's speech data.

[0019] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention detects whether the elderly have fallen during activities by analyzing the factors that cause them to fall, and comprehensively evaluates the falling risk of the current target by analyzing the surrounding environment and movement characteristics of the target. This greatly improves the accuracy of judging the elderly's falls, and effectively solves the problem of inaccurate detection results due to lack of all-round consideration, thereby achieving the purpose of improving the accuracy of judging the elderly's falls.

[0020] 2. The present invention trains the model through the fall risk factors of multiple fall factors and the motion change characteristics related to falls through a deep learning algorithm, thereby making the model more adaptable in predicting the fall problem of the elderly, and effectively solving the problem of biased model prediction caused by only using motion change characteristics in the prior art, thereby achieving the purpose of increasing the accuracy of model output.

[0021] 3. The present invention outputs a comprehensive score and a corresponding score for the fall risk factor through a model, which is then compared with a set threshold to determine the comprehensive risk score of the elderly, thereby improving the accuracy of the output results. The thresholds of the fall risk factor and the comprehensive motion change characteristics are set by the user through the system, and the comprehensive risk score is divided into several levels. When the score level exceeds the set threshold, the voice alarm system is triggered to issue a voice alarm to the target.

[0022] 4. The present invention is based on a difference analysis algorithm of continuous frames, including motion detection and optical flow method. By comprehensively applying motion detection and optical flow method, the accuracy of motion detection is improved, and the problem of low motion detection accuracy in the prior art is effectively solved, thereby achieving the purpose of improving the accuracy of target detection. DETAILED DESCRIPTION

[0023] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0024] This embodiment provides a technical solution: a home elderly remote monitoring alarm system with voice alarm and activity detection, including an infrared monitoring module, a preprocessing module, a risk assessment module, a fall detection module, an alarm module and a mobile control terminal; Among them, the infrared monitoring module is connected with the preprocessing module, and is used to obtain thermal imaging data of the elderly's activity scene. The infrared thermal imaging technology can accurately capture the location of the heat source and judge the thermal properties of the object. Compared with the traditional manual inspection method, infrared detection can greatly improve the detection efficiency and accuracy, and avoid missed detection and false detection; the preprocessing module is respectively connected with the risk assessment module and the fall detection module, and is used to process the thermal imaging data of the elderly's activity scene to obtain preprocessed thermal imaging data; the risk assessment module is used to assess the current risk of falling of the elderly based on the preprocessed thermal imaging data, and the fall detection module is used to judge whether the elderly has fallen based on the target human body of the preprocessed thermal imaging data. The alarm module is respectively connected with the risk assessment module and the fall detection module, and the mobile control terminal is used to receive the elderly's status information and issue a warning or help based on the elderly's status information; The preprocessing module includes a data acquisition unit, a data collection system, a data upload system, and a data storage warehouse; The risk assessment module includes a data analysis system, a result determination system, and a threshold import system; The fall detection module includes an abnormal alarm system, a data retrieval system, a central control terminal, and a remote connection system.

[0025] The data acquisition unit is connected to the data collection system and is used to collect and gather data of the elderly on site; the data acquisition unit includes a first data acquisition module, which is connected to the data collection system and is used to obtain thermal imaging data of the elderly's daily activities; the data acquisition unit includes a second data acquisition module, which is connected to the data collection system and is used to obtain thermal imaging data of the elderly's daily activities; The data collection system is connected to the data upload system to organize the collected data of the elderly and transmit it to the data upload system; The data uploading system is connected to the data storage warehouse and is used to upload the elderly data to the data storage warehouse; The data storage warehouse is connected to the data retrieval system and is used to store the elderly data input by the data retrieval system; The data storage warehouse is connected to the data analysis system, the data analysis system is connected to the result determination system, and the data analysis system performs remote monitoring through data analysis; The data analysis system is connected to the threshold import system and the abnormal alarm system respectively, and is used to compare the data analysis with the database threshold and issue an alarm; The data analysis system is connected to the central control terminal to control and process the central data information; The central control terminal is connected to the remote connection system for remote monitoring and processing of data information analyzed by the data analysis system; The data analysis system is connected to the mobile control terminal and is used to send the data information analyzed by the data analysis system to family members through the mobile control terminal.

[0026] The fall detection module is used to detect whether the elderly fall during activities. The detection method is: Step A: receiving thermal imaging data, subtracting two consecutive adjacent frames of thermal imaging data from each other, and obtaining a frame difference image corresponding to the two adjacent frames of thermal imaging data, wherein the non-zero pixel points in the frame difference image are suspected moving target pixel points; Step B: Intersection processing is performed on the suspected moving target pixel points in the frame difference image and the preset moving area. If there is no intersection, it is determined that a moving target is detected; Step C: Using the optical flow algorithm to calculate the moving direction and speed of the moving target in the thermal imaging data; Step D: Determine the motion region based on the frame difference map and non-zero pixel points, mark and connect the motion regions in the continuous frame differences by using the connected domain method, and obtain the centroid of the motion region; Step E: Track the centroid of the motion area and obtain the pixel distance between the two centroids. If the pixel distance is greater than a preset threshold, it is determined that the moving target has fallen. The fall detection module is used to detect and send out an alarm signal when the risk assessment module determines that the elderly has fallen.

[0027] The fall detection module also includes a difference analysis algorithm based on continuous frames to obtain the frame difference map. The difference analysis algorithm formula based on continuous frames is: D(xc,yc)=|It(xc,yc)-I(ta,xc,yc)|; Among them, It(xc,yc) and I(ta,xc,yc) are the pixel values ​​of the position in the two consecutive frames before and after (assuming t is the current frame and ta is the previous frame), and D(xc,yc) represents the pixel value at the position (xc,yc) in the frame difference map, which is equal to the absolute value of the difference between the pixel values ​​at the same position in the two frames before and after.

[0028] The risk assessment module includes a fall risk factor acquisition module, which is used to obtain risk factors related to falls. For falls of the elderly, the falling behavior and environmental factors are taken into account. The fall risk factor acquisition module is based on the thermal imaging data of the infrared monitoring module, and counts the motion target characteristics within 30 seconds before the elderly fall over a period of time to obtain a set of risk factors related to falls.

[0029] The data analysis system includes a target detection module, a target tracking module and a target motion statistics module, wherein the target detection module extracts the characteristic area of ​​the preprocessed thermal imaging data through a target detection algorithm, the target tracking module is used to track according to the characteristic area, and the target motion statistics module is used to analyze the motion change characteristics of the characteristic area within a period of time; the target tracking module tracks the characteristic area obtained by the target detection module, and extracts the motion characteristics of the target area during the tracking process, wherein the motion characteristics include moving speed, moving distance and moving area, and the target motion statistics module counts the motion characteristics of the target area within a certain period of time, and analyzes the moving speed characteristic score, the moving distance characteristic score and the moving area characteristic score, and then comprehensively scores the comprehensive motion change characteristics of the target area, and the comprehensive scoring formula is as follows: S = Σ(Si × Wi); Among them: S is the comprehensive score, which represents the overall evaluation of the comprehensive motion change characteristics of the target area; Si is the score of each feature, representing the performance of the target area in a specific motion change feature; i is the number of features, indicating the total number of features involved in the comprehensive scoring; Wi is the weight of each feature, indicating the importance of the feature in the comprehensive score.

[0030] The data analysis system also includes a fall risk factor processing module, which combines the fall risk factor with the motion change characteristics obtained by the target motion statistics module to obtain comprehensive information of the motion change characteristics and different fall risk factors, and calculates the fall risk score of the elderly based on the comprehensive information. The fall risk score formula is as follows: A=aθ+bV+cL+dArea; Among them: A fall risk score is a comprehensive assessment result used to quantify the risk of falls; θ Elderly posture characteristic score, which reflects factors such as posture stability and coordination of the elderly at a specific time point; V is the moving speed feature score; L is the motion distance feature score; Area is the motion area feature score.

[0031] The data acquisition unit simultaneously collects the posture, behavior and multiple continuous thermal imaging data of the elderly's activities. The posture feature score includes scores corresponding to different fall risk factors. Corresponding to the posture feature score is the factor feature of the corresponding fall risk factor obtained by the fall risk factor acquisition module. The factor feature includes arm posture feature, leg posture feature, body tilt feature and footstep size feature. Combined with the factor features, the posture feature score related to falls is obtained through comparison and comprehensive calculation. The posture feature score formula related to falls is as follows: θ_total=θ1×k1+θ2×k2+θ3×k3+θ4×k4; Among them: θ1 is the arm posture feature score, which reflects the impact of the arm posture or movement state on the risk of falling; θ2 is the characteristic score of leg posture, which reflects the relationship between leg posture or movement and fall risk; θ3 is the body tilt feature score, which evaluates the effect of body tilt angle or posture change on the risk of falling; θ4 is the step feature score, which involves factors such as step stability and step size; k1, k2, k3, and k4 are weight factors of the corresponding features, which determine the relative importance of each feature in the overall score.

[0032] The alarm module includes a voice alarm system, which is connected to the alarm module. The voice alarm system sends a voice alarm signal according to the fall risk score level and the movement change characteristics, and the voice alarm system includes: A voice playback module is used to match the sound features according to the fall risk score level and the movement change characteristics to obtain a voice alarm signal; A voice receiving module is used to collect voice information and transmit the voice information to a voice server; The voice judgment module is used to collect environmental noise and sound intensity to determine whether the current environment is suitable for voice alarm; A voice optimization module, used to optimize the acquired voice information; A speech coding module, used for compressing and coding the acquired speech information; A speech decoding module is used to decode and restore speech information; The speech synthesis module is used to synthesize speech according to the acquired speech alarm signal and in combination with the user's speech data.

[0033] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A home elderly remote monitoring alarm system with voice alarm and activity detection, characterized by: It includes infrared monitoring module, pre-processing module, risk assessment module, fall detection module, alarm module and mobile control terminal; Among them, the infrared monitoring module is connected to the preprocessing module to obtain the thermal imaging data of the elderly activity scene; the preprocessing module is respectively connected to the risk assessment module and the fall detection module to process the thermal imaging data of the elderly activity scene to obtain the preprocessed thermal imaging data; the risk assessment module is used to assess the current risk of falling of the elderly based on the preprocessed thermal imaging data, the fall detection module is used to determine whether the elderly has fallen based on the target human body of the preprocessed thermal imaging data, the alarm module is respectively connected to the risk assessment module and the fall detection module, and the mobile control terminal is used to receive the elderly status information and issue a warning or help based on the elderly status information; The preprocessing module includes a data acquisition unit, a data collection system, a data upload system, and a data storage warehouse; The risk assessment module includes a data analysis system, a result determination system, and a threshold import system; The fall detection module includes an abnormal alarm system, a data retrieval system, a central control terminal, and a remote connection system.

2. The home elderly remote monitoring alarm system with voice alarm and activity detection according to claim 1 is characterized by: The data acquisition unit is connected to the data collection system and is used to collect and gather data of the elderly on site; the data acquisition unit includes a first data acquisition module, which is connected to the data collection system and is used to obtain thermal imaging data of the elderly's daily activities; the data acquisition unit includes a second data acquisition module, which is connected to the data collection system and is used to obtain thermal imaging data of the elderly's daily activities; The data collection system is connected to the data upload system to organize the collected data of the elderly and transmit it to the data upload system; The data uploading system is connected to the data storage warehouse and is used to upload the elderly data to the data storage warehouse; The data storage warehouse is connected to the data retrieval system and is used to store the elderly data input by the data retrieval system; The data storage warehouse is connected to the data analysis system, the data analysis system is connected to the result determination system, and the data analysis system performs remote monitoring through data analysis; The data analysis system is connected to the threshold import system and the abnormal alarm system respectively, and is used to compare the data analysis with the database threshold and issue an alarm; The data analysis system is connected to the central control terminal to control and process the central data information; The central control terminal is connected to the remote connection system for remote monitoring and processing of data information analyzed by the data analysis system; The data analysis system is connected to the mobile control terminal and is used to send the data information analyzed by the data analysis system to family members through the mobile control terminal.

3. The home elderly remote monitoring alarm system with voice alarm and activity detection according to claim 1 is characterized by: The fall detection module is used to detect whether the elderly fall during activities. The detection method is: Step A: receiving thermal imaging data, subtracting two consecutive adjacent frames of thermal imaging data from each other, and obtaining a frame difference image corresponding to the two adjacent frames of thermal imaging data, wherein the non-zero pixel points in the frame difference image are suspected moving target pixel points; Step B: Intersection processing is performed on the suspected moving target pixel points in the frame difference image and the preset moving area. If there is no intersection, it is determined that a moving target is detected; Step C: Using the optical flow algorithm to calculate the moving direction and speed of the moving target in the thermal imaging data; Step D: Determine the motion region based on the frame difference map and non-zero pixel points, mark and connect the motion regions in the continuous frame differences by using the connected domain method, and obtain the centroid of the motion region; Step E: Track the centroid of the motion area and obtain the pixel distance between the two centroids. If the pixel distance is greater than a preset threshold, it is determined that the moving target has fallen. The fall detection module is used to detect and send out an alarm signal when the risk assessment module determines that the elderly has fallen.

4. The home elderly remote monitoring alarm system with voice alarm and activity detection according to claim 3 is characterized by: The fall detection module also includes a difference analysis algorithm based on continuous frames to obtain the frame difference map. The difference analysis algorithm formula based on continuous frames is: D(xc,yc)=|It(xc,yc)-I(ta,xc,yc)|; Among them, It(xc,yc) and I(ta,xc,yc) are the pixel values ​​of the position in the two consecutive frames before and after (assuming t is the current frame and ta is the previous frame), and D(xc,yc) represents the pixel value at the position (xc,yc) in the frame difference map, which is equal to the absolute value of the difference between the pixel values ​​at the same position in the two frames before and after.

5. The home elderly remote monitoring alarm system with voice alarm and activity detection according to claim 1 is characterized by: The risk assessment module includes a fall risk factor acquisition module, which is used to obtain risk factors related to falls. For falls of the elderly, the falling behavior and environmental factors are taken into account. The fall risk factor acquisition module is based on the thermal imaging data of the infrared monitoring module, and counts the motion target characteristics within 30 seconds before the elderly fall over a period of time to obtain a set of risk factors related to falls.

6. The home elderly remote monitoring alarm system with voice alarm and activity detection according to claim 5 is characterized by: The data analysis system includes a target detection module, a target tracking module and a target motion statistics module, wherein the target detection module extracts the characteristic area of ​​the preprocessed thermal imaging data through a target detection algorithm, the target tracking module is used to track according to the characteristic area, and the target motion statistics module is used to analyze the motion change characteristics of the characteristic area within a period of time; the target tracking module tracks the characteristic area obtained by the target detection module, and extracts the motion characteristics of the target area during the tracking process, wherein the motion characteristics include moving speed, moving distance and moving area, and the target motion statistics module counts the motion characteristics of the target area within a certain period of time, and analyzes the moving speed characteristic score, the moving distance characteristic score and the moving area characteristic score, and then comprehensively scores the comprehensive motion change characteristics of the target area, and the comprehensive scoring formula is as follows: S = Σ(Si × Wi); Among them: S is the comprehensive score, which represents the overall evaluation of the comprehensive motion change characteristics of the target area; Si is the score of each feature, representing the performance of the target area in a specific motion change feature; i is the number of features, indicating the total number of features involved in the comprehensive scoring; Wi is the weight of each feature, indicating the importance of the feature in the comprehensive score.

7. The home elderly remote monitoring alarm system with voice alarm and activity detection according to claim 6 is characterized by: The data analysis system also includes a fall risk factor processing module, which combines the fall risk factor with the motion change characteristics obtained by the target motion statistics module to obtain comprehensive information of the motion change characteristics and different fall risk factors, and calculates the fall risk score of the elderly based on the comprehensive information. The fall risk score formula is as follows: A=aθ+bV+cL+dArea; Among them: A fall risk score is a comprehensive assessment result used to quantify the risk of falls; θ Elderly posture characteristic score, which reflects factors such as posture stability and coordination of the elderly at a specific time point; V is the moving speed feature score; L is the motion distance feature score; Area is the motion area feature score.

8. The home elderly remote monitoring alarm system with voice alarm and activity detection according to claim 7 is characterized by: The data acquisition unit simultaneously collects the posture, behavior and multiple continuous thermal imaging data of the elderly's activities. The posture feature score includes scores corresponding to different fall risk factors. Corresponding to the posture feature score is the factor feature of the corresponding fall risk factor obtained by the fall risk factor acquisition module. The factor feature includes arm posture feature, leg posture feature, body tilt feature and footstep size feature. Combined with the factor features, the posture feature score related to falls is obtained through comparison and comprehensive calculation. The posture feature score formula related to falls is as follows: θ_total=θ1×k1+θ2×k2+θ3×k3+θ4×k4; Among them: θ1 is the arm posture feature score, which reflects the impact of the arm posture or movement state on the risk of falling; θ2 is the characteristic score of leg posture, which reflects the relationship between leg posture or movement and fall risk; θ3 is the body tilt feature score, which evaluates the effect of body tilt angle or posture change on the risk of falling; θ4 is the step feature score, which involves factors such as step stability and step size; k1, k2, k3, and k4 are weight factors of the corresponding features, which determine the relative importance of each feature in the overall score.

9. The home elderly remote monitoring alarm system with voice alarm and activity detection according to claim 1 is characterized by: The alarm module includes a voice alarm system, which is connected to the alarm module. The voice alarm system sends a voice alarm signal according to the fall risk score level and the movement change characteristics, and the voice alarm system includes: A voice playback module is used to match the sound features according to the fall risk score level and the movement change characteristics to obtain a voice alarm signal; A voice receiving module is used to collect voice information and transmit the voice information to a voice server; The voice judgment module is used to collect environmental noise and sound intensity to determine whether the current environment is suitable for voice alarm; A voice optimization module, used to optimize the acquired voice information; A speech coding module, used for compressing and coding the acquired speech information; A speech decoding module is used to decode and restore speech information; The speech synthesis module is used to synthesize speech according to the acquired speech alarm signal and in combination with the user's speech data.