Millimeter-Wave Radar-Based Fall Detection Method

Abstract

The present invention discloses a fall detection method based on millimeter-wave radar, which relates to the technical field of fall detection and solves the technical problem in the prior art that it is impossible to detect the actions of a fallen person after determining the passage space according to risk assessment. Specifically, it includes layout detection and annotation, layout detection of the activity area of the fall detection person according to millimeter-wave radar technology, and obtaining the passage space within the activity area where the person is located according to the layout detection; action recognition and early warning, after determining the passage space, detecting the behavioral actions of the fall detection person, and evaluating the real-time fall risk of the fall detection person according to the behavioral action detection; spatial layout analysis, obtaining the set object at the position where the fall detection person makes a falling action according to the spatial layout, and connecting and evaluating the body movements of the person with the set object when the fall detection person makes a falling action, and inferring whether the fall detection person needs to obtain balance through the set object when falling.

Description

Technical Field

[0001] The present invention relates to the technical field of fall detection, and specifically to a fall detection method based on millimeter-wave radar. Background Technique

[0002] Fall detection is a technology for detecting whether a human body has a fall behavior, and has a wide range of applications in fields such as smart home and healthcare; fall detection based on millimeter-wave radar is a method for monitoring and judging the fall behavior of a human body by using millimeter-wave radar technology, and has the advantages of non-contact, high precision, and being unaffected by light and occlusion.

[0003] However, in the prior art, when performing fall detection by millimeter-wave radar, it is impossible to perform risk assessment on the activity area of the fall detection personnel, nor can it determine the passage space according to the risk assessment and then perform fall personnel action detection. It is impossible to improve the fall detection efficiency according to the action acquisition points, reducing the accuracy of fall risk detection. In addition, it is impossible to re-plan and control the object layout of the activity area according to the fall action detection.

[0004] In view of the above technical defects, a solution is now proposed. Summary of the Invention

[0005] The purpose of the present invention is to solve the above-mentioned problems and propose a fall detection method based on millimeter-wave radar.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A fall detection method based on millimeter-wave radar, and the process of the fall detection method is as follows:

[0008] Layout detection and annotation, perform layout detection on the activity area of the fall detection personnel according to millimeter-wave radar technology, and obtain the passage space in the activity area according to the layout detection;

[0009] Action recognition and early warning, after determining the passage space, perform behavior action detection on the fall detection personnel, and evaluate the real-time fall risk of the fall detection personnel according to the behavior action detection;

[0010] Spatial layout analysis, obtain the set objects at the position where the fall detection personnel have a fall action according to the spatial layout, and connect and evaluate the personnel limb actions with the set objects when the fall detection personnel have a fall action, and infer whether the fall detection personnel need to obtain balance through the set objects when falling.

[0011] As a preferred embodiment of the present invention, the process of layout detection and annotation is as follows:

[0012] Set the signal wave emission point, expand the signal wave according to the millimeter-wave radar to cover the activity area where the fall-detection personnel are located, determine the set objects in the activity area according to the continuous emission and reception of the signal wave from the signal wave emission point, and obtain the shape information of each set object according to the change of the signal wave; after obtaining the shape information of each set object, obtain the space in the activity area except for the space occupied by the set objects, and mark it as the passage space; obtain the movement trajectory space of the fall-detection personnel in the passage space and the corresponding connected passage destinations according to the continuity of the passage space.

[0013] Collect the area ratio of the movement trajectory space of the fall-detection personnel to the passage space, and at the same time collect the ratio of the movement trajectory distance where the area of the movement trajectory space of the fall-detection personnel is higher than the area required for the movement of the fall-detection personnel.

[0014] As a preferred embodiment of the present invention, if the area ratio of the movement trajectory space to the passage space exceeds the space area ratio threshold, or the ratio of the movement trajectory distance where the area of the movement trajectory space of the fall-detection personnel is higher than the area required for the movement of the fall-detection personnel does not exceed the movement trajectory distance ratio threshold, a high-risk signal is generated; if the area ratio of the movement trajectory space to the passage space does not exceed the space area ratio threshold, and the ratio of the movement trajectory distance where the area of the movement trajectory space of the fall-detection personnel is higher than the area required for the movement of the fall-detection personnel exceeds the movement trajectory distance ratio threshold, a low-risk signal is generated and sent to the guardian's terminal.

[0015] As a preferred embodiment of the present invention, the process of action recognition and warning is as follows:

[0016] Obtain the central point position of the person according to the body shape of the fall-detection personnel, and mark the center of gravity to obtain the center of gravity position; obtain the movement point position according to the body shape and movement action of the fall-detection personnel, and mark it as the limb point position according to the movement point position and the movement position of the fall-detection personnel; count the position where the center of gravity point is located when the fall-detection personnel move, and when there is no fall, the center of gravity points at each moment are linear in the activity area space, and collect the reciprocating cycle of the beating trajectory of the limb point position of the fall-detection personnel when there is no fall, and mark it as the pendulum action.

[0017] As a preferred embodiment of the present invention, the action of the fall-detection personnel is recognized through the signal wave, and the peak value of the instantaneous decrease speed of the height of the center of gravity point where the straight line shape of the fall-detection personnel changes is collected, and the increase span value of the swing amplitude of the limb point position after the fall-detection personnel change the pendulum action of the corresponding limb point:

[0018] If the peak value of the instantaneous reduction speed of the height of the center-of-gravity position of the person detecting a fall, which changes the linear center-of-gravity position, exceeds the reduction speed peak threshold, but the center-of-gravity position returns to the height of the linear type at the adjacent moment; if the limb position of the person detecting a fall exceeds the amplitude growth span threshold after changing the pendulum motion of the corresponding limb point, but the swing amplitude of the corresponding pendulum motion drops to zero and then returns to the swing motion period at the adjacent moment;

[0019] When the above two situations occur simultaneously, the real-time motion state of the person detecting a fall is marked as having a fall action and stabilizing the balance state, and the motion state is sent to the caregiver;

[0020] If the peak value of the instantaneous reduction speed of the height of the center-of-gravity position of the person detecting a fall, which changes the linear center-of-gravity position, exceeds the reduction speed peak threshold, and the center-of-gravity position does not return to the height of the linear type; if the increase span value of the swing amplitude of the limb position of the person detecting a fall exceeds the amplitude growth span threshold after changing the pendulum motion of the corresponding limb position, but the swing amplitude of the corresponding pendulum motion drops to zero and does not return to the swing motion period at the adjacent moment;

[0021] When any of the above two situations occurs, the real-time motion state of the person detecting a fall is marked as having a fall action and not stabilizing the balance state, and the motion state is sent to the caregiver.

[0022] As a preferred embodiment of the present invention, the process of spatial layout analysis is as follows:

[0023] When in the state of having a fall action and stabilizing the balance state, collect the overlapping moment of the limb position of the person detecting a fall and the position of the set object around the position where the fall action occurs, as well as the moment when the person detecting a fall stabilizes the balance state, and obtain the interval duration according to the moment comparison, and set it as the object-assisted stabilization duration;

[0024] When in the state of having a fall action and not stabilizing the balance state, collect the position interval distance value between the limb position of the person detecting a fall and the set object around the position where the fall action occurs, and mark the position interval distance value between the limb position of the person detecting a fall and the set object around the position where the fall action occurs as the object-assisted interval distance value.

[0025] As a preferred embodiment of the present invention, analyze the object-assisted stabilization duration and the object-assisted interval distance:

[0026] If the object-assisted stabilization duration exceeds the stabilization duration threshold, the current set object position is marked as the low-effect object position; if the object-assisted stabilization duration does not exceed the stabilization duration threshold, the current set object position is marked as the high-effect object position;

[0027] If the object-assisted interval exceeds the set interval distance threshold, and there is a tendency for the limb position of the fall detection person to move towards the set object when a fall action occurs, then mark the current set object position as the long-distance object position;

[0028] If the object-assisted interval does not exceed the set interval distance threshold, and there is a tendency for the limb position of the fall detection person to move towards the set object when a fall action occurs, then mark the current set object position as the low-efficiency object position for assistance;

[0029] If the object-assisted interval exceeds the set interval distance threshold, and there is no tendency for the limb position of the fall detection person to move towards the set object when a fall action occurs, then it is considered that the corresponding set object is too far away to provide assistance for stability, and mark the current set object position as the unconnected object position;

[0030] If the object-assisted interval does not exceed the set interval distance threshold, and there is no tendency for the limb position of the fall detection person to move towards the set object when a fall action occurs, then it is considered that the corresponding set object does not meet the requirement for assistance stability, and mark the current set object position as the ineffective object position for assistance.

[0031] Compared with the prior art, the beneficial effects of the present invention are:

[0032] 1. In the present invention, the layout of the activity area of the fall detection person is detected according to the millimeter-wave radar technology, and the passage space within the activity area obtained from the layout detection is used to preliminarily predict the fall of the fall detection person according to the passage space. At the same time, according to the trajectory change of the passage space, the accuracy of the fall warning of the fall detection person can also be improved, avoiding the increase in the fall risk of the fall detection person caused by the change of the passage space, and not warning in time but warning after the fall action occurs, which cannot ensure the passage safety of the fall detection person.

[0033] 2. In the present invention, after determining the passage space, the behavior actions of the fall detection person are detected, and the real-time fall risk of the fall detection person is evaluated according to the behavior action detection. A warning is given in time when a fall risk action occurs, so that the fall detection person can be rescued in time after falling, and the impact of the fall of the fall detection person cannot be reduced.

[0034] 3. In the present invention, the set object at the position where the fall detection person has a fall action is obtained according to the space layout, and the limb action of the person is connected and evaluated with the set object when the fall detection person has a fall action to infer whether the fall detection person needs to obtain balance through the set object when falling. Thus, the space layout plan is carried out according to the actual role played by the set object, and at the same time, the space occupied by the set object at the current position is reset. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] For the convenience of those skilled in the art to understand, the present invention will be further described below in conjunction with the accompanying drawings.

[0036] Figure 1 is the overall method flow chart of the present invention;

[0037] Figure 2 is the process flow chart of the layout detection and annotation in the present invention;

[0038] Figure 3 is the process flow chart of the action recognition and early warning in the present invention. Detailed implementation manners

[0039] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present invention.

[0040] The mention of "embodiment" in this article means that the specific features, structures or characteristics described in conjunction with the embodiment can be included in at least one embodiment of the present invention. The phrase appears in various positions in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0041] Please refer to Figure 1 As shown, the fall detection method based on millimeter-wave radar, and the specific fall detection method process is as follows:

[0042] Perform layout detection and annotation through a millimeter-wave radar. According to the millimeter-wave radar technology, perform layout detection on the activity area of the fall detection personnel, and obtain the passage space within the activity area according to the layout detection, so as to perform a preliminary prediction of the fall of the fall detection personnel according to the passage space. At the same time, according to the trajectory change of the passage space, the accuracy of the fall early warning of the fall detection personnel can also be improved, avoiding the increase in the fall risk of the fall detection personnel caused by the change of the passage space, and not giving an early warning in time but giving an early warning after the fall action occurs, which cannot guarantee the passage safety of the fall detection personnel;

[0043] Action recognition and early warning for fall detection personnel. After determining the passage space, conduct behavioral action detection on the fall detection personnel, and evaluate the real-time fall risk of the fall detection personnel based on the behavioral action detection. Provide early warning in a timely manner when fall risk actions occur, so that timely assistance can be obtained after the fall detection personnel fall, and avoid the inability to reduce the impact of the fall of the fall detection personnel;

[0044] Analysis of fall actions and spatial layout. Based on the spatial layout, obtain the set objects at the positions where the fall detection personnel perform fall actions. When the fall detection personnel perform fall actions, connect the personnel's limb actions with the set objects for evaluation, and infer whether the fall detection personnel need to obtain balance through the set objects when falling. Thus, plan the spatial layout according to the actual role played by the set objects, and at the same time, re-set the space occupied by the set objects at the current position;

[0045] Please refer to Figure 2 As shown in the figure, the layout detection and annotation process is as follows:

[0046] Set the signal wave emission point. Expand the signal wave according to the millimeter-wave radar to cover the activity area where the fall detection personnel are located. Determine the set objects in the activity area according to the continuous emission and reception of the signal wave by the signal wave emission point. Obtain the shape information of each set object according to the change of the signal wave, specifically data such as shape, height, and length;

[0047] After obtaining the shape information of each set object, obtain the space in the activity area except for the space occupied by the set objects and mark it as the passage space; Obtain the movement trajectory space of the fall detection personnel in the passage space and the corresponding connected passage destinations according to the continuity of the passage space; And conduct all collection and annotation;

[0048] Collect the area ratio of the movement trajectory space of the fall detection personnel to the passage space, and at the same time collect the ratio of the movement trajectory distance where the area of the movement trajectory space of the fall detection personnel is higher than the area required for the movement of the fall detection personnel; It should be noted that the area of the space is calculated based on the floor area within the corresponding space; The area required for the movement of the fall detection personnel is calculated according to the steps of the fall detection personnel and the lateral step spacing; And compare the area ratio of the movement trajectory space of the fall detection personnel to the passage space and the ratio of the movement trajectory distance where the area of the movement trajectory space of the fall detection personnel is higher than the area required for the movement of the fall detection personnel with the area ratio threshold and the movement trajectory distance ratio threshold respectively: The threshold is obtained by analyzing the fall process of historical personnel of the same type during actual monitoring by personnel in this field;

[0049] If the area ratio of the movement trajectory space of the fall detection person to the passage space exceeds the space area ratio threshold, or the ratio of the movement trajectory distance where the area of the movement trajectory space of the fall detection person is higher than the area required for the movement of the fall detection person does not exceed the movement trajectory distance ratio threshold, it is inferred that the fall risk of the current activity area of the fall detection person is high, a high-risk signal is generated and sent to the guardian terminal. After receiving the high-risk signal, the guardian terminal conducts a layout planning of the set objects in the activity area where the fall detection person is located, and expands the passage space according to the requirements of the actual destination;

[0050] If the area ratio of the movement trajectory space of the fall detection person to the passage space does not exceed the space area ratio threshold, and the ratio of the movement trajectory distance where the area of the movement trajectory space of the fall detection person is higher than the area required for the movement of the fall detection person exceeds the movement trajectory distance ratio threshold, it is inferred that the fall risk of the current activity area of the fall detection person is low, a low-risk signal is generated and sent to the guardian terminal;

[0051] Please refer to Figure 3 As shown below, the action recognition and warning process is as follows:

[0052] When determining the low fall risk in the space area where the fall detection person is located, the movement actions of the fall detection person are continuously and real-time recognized and detected. The center point of the person is obtained according to the body shape of the fall detection person, and the center of gravity is marked according to whether the position of the center point floats when the fall detection person moves, that is, if the position of the center point floats, the point is re-acquired, and if the position of the center point does not float, the center point is marked as the center of gravity point;

[0053] The movement points are obtained according to the body shape and movement actions of the fall detection person, and are marked as limb points according to the movement points and the movement positions of the fall detection person;

[0054] When the fall detection person moves, the position where the center of gravity point is located is statistically analyzed. When no fall occurs, the center of gravity points at each moment are in a straight line in the activity area space. It should be noted that when the floating of the center of gravity point is within the set range, the floating is not considered; and the reciprocating cycle of the swing trajectory of the limb points of the fall detection person when no fall occurs is collected and marked as a pendulum action;

[0055] The action of the fall detection person is recognized through a signal wave. The instantaneous reduction speed peak value of the height of the center of gravity point that changes from a straight line of the fall detection person is collected. At the same time, the increase span value of the swing amplitude of the limb point after the corresponding limb point of the fall detection person changes the pendulum action is collected. The instantaneous reduction speed peak value of the height of the center of gravity point that changes from a straight line of the fall detection person and the increase span value of the swing amplitude of the limb point after the corresponding limb point of the fall detection person changes the pendulum action are respectively compared with the reduction speed peak value threshold and the amplitude increase span threshold:

[0056] If the peak value of the instantaneous reduction speed of the height at which the center of gravity point of the person detecting a fall changes the linear shape exceeds the reduction speed peak threshold, but the center of gravity point returns to the height of the linear shape at the adjacent moment;

[0057] If the limb position of the person detecting a fall exceeds the amplitude growth span threshold after changing the pendulum motion, but the swing amplitude of the corresponding pendulum motion drops to zero at the adjacent moment and then returns to the swing motion period;

[0058] When the above two situations occur simultaneously, mark the real-time motion state of the person detecting a fall as having a fall action and stabilizing the balance state, and send the motion state to the caregiver;

[0059] If the peak value of the instantaneous reduction speed of the height at which the center of gravity point of the person detecting a fall changes the linear shape exceeds the reduction speed peak threshold, and the center of gravity point does not return to the height of the linear shape; if the increase span value of the swing amplitude of the limb position of the person detecting a fall exceeds the amplitude growth span threshold after changing the pendulum motion, but the swing amplitude of the corresponding pendulum motion drops to zero at the adjacent moment and does not return to the swing motion period;

[0060] When any of the above two situations occurs, mark the real-time motion state of the person detecting a fall as having a fall action and not stabilizing the balance state, and send the motion state to the caregiver;

[0061] If the peak value of the instantaneous reduction speed of the height at which the center of gravity point of the person detecting a fall changes the linear shape does not exceed the reduction speed peak threshold, and the increase span value of the swing amplitude of the limb position of the person detecting a fall exceeds the amplitude growth span threshold but does not exceed the amplitude growth span threshold after changing the pendulum motion, then it is inferred that the real-time motion state of the person detecting a fall is marked as not having a fall action, and the motion state is sent to the caregiver;

[0062] The caregiver gives an early warning to the person detecting a fall; and rescues the person detecting a fall in a timely manner;

[0063] The analysis process of the fall action and the spatial layout is as follows:

[0064] When in the state of having a fall action and stabilizing the balance state, collect the overlapping moment of the limb position of the person detecting a fall and the positions of the set objects around the position where the fall action occurs, as well as the moment when the person detecting a fall stabilizes the balance state, and obtain the interval duration according to the moment comparison, and set it as the object-assisted stabilization duration;

[0065] When a falling action occurs and the balance is not stabilized, collect the position interval distance value between the body points of the fall detection person and the set objects around the position where the falling action occurs, and mark the position interval distance value between the body points of the fall detection person and the set objects around the position where the falling action occurs as the object-assisted interval distance value;

[0066] Analyze the object-assisted stability duration and the object-assisted interval distance:

[0067] If the object-assisted stability duration exceeds the stability duration threshold, it is inferred that the auxiliary stability effect of the current set object is low, and the current set object position is marked as the low-effect object position;

[0068] If the object-assisted stability duration does not exceed the stability duration threshold, it is inferred that the auxiliary stability effect of the current set object is high, and the current set object position is marked as the high-effect object position;

[0069] If the object-assisted interval exceeds the set interval distance threshold and there is a tendency for the body points of the fall detection person to move towards the set object when the falling action occurs, it is inferred that the distance of the corresponding set object does not meet the auxiliary stability requirement, and the current set object position is marked as the long-distance object position;

[0070] If the object-assisted interval does not exceed the set interval distance threshold and there is a tendency for the body points of the fall detection person to move towards the set object when the falling action occurs, it is inferred that the corresponding set object does not meet the auxiliary stability requirement, and the current set object position is marked as the auxiliary low-efficiency object position;

[0071] If the object-assisted interval exceeds the set interval distance threshold and there is no tendency for the body points of the fall detection person to move towards the set object when the falling action occurs, it is inferred that the distance of the corresponding set object is too far to provide auxiliary stability, and the current set object position is marked as the non-related object position;

[0072] If the object-assisted interval does not exceed the set interval distance threshold and there is no tendency for the body points of the fall detection person to move towards the set object when the falling action occurs, it is inferred that the corresponding set object does not meet the auxiliary stability requirement, and the current set object position is marked as the auxiliary ineffective object position;

[0073] Send the location where the fall action of the real-time fall detection person occurs, the locations of high-action objects, low-action objects, long-distance objects, auxiliary low-efficiency objects, and auxiliary ineffective objects to the guardian's terminal. After receiving the information, the guardian's terminal re-layouts the activity area where the fall detection person is located, that is, adjusts the set objects according to the set object position types around the location where the fall action occurs. For example, for the location of high-action objects, the current set objects are retained; for the location of low-action objects, the object types are replaced or components for assisting stability, such as handles, are added; for the location of long-distance objects, the set object positions are adjusted; for the locations of auxiliary low-efficiency objects and auxiliary ineffective objects, the positions are adjusted according to the actual layout, and components are added when adjustment is not possible.

[0074] When the present invention is in use, layout detection and annotation are carried out. The activity area of the fall detection person is detected according to the millimeter-wave radar technology, and the passage space within the activity area is obtained according to the layout detection; action recognition and early warning are carried out. After determining the passage space, the behavior actions of the fall detection person are detected, and the real-time fall risk of the fall detection person is evaluated according to the behavior action detection; space layout analysis is carried out. According to the space layout, the set objects at the location where the fall action of the fall detection person occurs are obtained, and the connection between the person's limb actions and the set objects is evaluated when the fall detection person makes a fall action, so as to infer whether the fall detection person needs to obtain balance through the set objects when falling.

[0075] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the present invention to the specific implementation manners only. Obviously, many modifications and changes can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the relevant technical fields can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A fall detection method based on a millimeter-wave radar, characterized in that, The process of the fall detection method is as follows: Layout detection and annotation: Detect the layout of the activity area of the fall detection personnel according to the millimeter-wave radar technology, and obtain the passage space within the activity area according to the layout detection; Action recognition and early warning: After determining the passage space, detect the behavior actions of the fall detection personnel, and evaluate the real-time fall risk of the fall detection personnel according to the behavior action detection; Spatial layout analysis: Obtain the set object at the position where the fall action of the fall detection personnel occurs according to the spatial layout, and connect and evaluate the personnel limb actions with the set object when the fall detection personnel have a fall action, and infer whether the fall detection personnel need to obtain balance through the set object when falling; The process of spatial layout analysis is as follows: When in the state of having a fall action and stabilizing the balance, collect the overlapping moment of the limb points of the fall detection personnel and the points of the set object around the position where the fall action occurs, as well as the moment when the fall detection personnel stabilize the balance state, and obtain the interval duration according to the moment comparison, and set it as the object-assisted stabilization duration; When in the state of having a fall action and not stabilizing the balance, collect the position interval distance value between the limb points of the fall detection personnel and the set object around the position where the fall action occurs, and mark the position interval distance value between the limb points of the fall detection personnel and the set object around the position where the fall action occurs as the object-assisted interval distance value; Analyze the object-assisted stabilization duration and the object-assisted interval distance: If the object-assisted stabilization duration exceeds the stabilization duration threshold, mark the current set object position as the low-effect object position; If the object-assisted stabilization duration does not exceed the stabilization duration threshold, mark the current set object position as the high-effect object position; If the object-assisted interval exceeds the set interval distance threshold and there is a moving trend of the limb points of the fall detection personnel towards the set object when a fall action occurs, mark the current set object position as the long-distance object position; If the object-assisted interval does not exceed the set interval distance threshold and there is a moving trend of the limb points of the fall detection personnel towards the set object when a fall action occurs, mark the current set object position as the low-efficiency object position; If the object-assisted interval exceeds the set interval distance threshold and there is no moving trend of the limb points of the fall detection personnel towards the set object when a fall action occurs, it means that the corresponding set object is too far away to play an auxiliary stabilizing role, and mark the current set object position as the non-related object position; If the object-assisted interval does not exceed the set interval distance threshold and there is no moving trend of the limb points of the fall detection personnel towards the set object when a fall action occurs, it means that the corresponding set object does not meet the auxiliary stabilization requirement, and mark the current set object position as the ineffective object position; 2. The fall detection method based on millimeter wave radar according to claim 1, wherein, The process of layout detection and annotation is as follows: Set the signal wave emission point, expand the signal wave according to the millimeter-wave radar to cover the activity area where the fall-detection personnel are located, determine the set objects in the activity area according to the continuous emission and reception of the signal wave at the signal wave emission point, and obtain the shape information of each set object according to the change of the signal wave; after obtaining the shape information of each set object, obtain the space in the activity area except for the space occupied by the set objects, and mark it as the passage space; Obtain the movement trajectory space of the fall-detection personnel in the passage space and the corresponding connected passage destinations according to the continuity of the passage space; Collect the area ratio of the movement trajectory space of the fall-detection personnel to the passage space, and at the same time collect the ratio of the movement trajectory distance where the area of the movement trajectory space of the fall-detection personnel is higher than the area required for the movement of the fall-detection personnel.

3. The fall detection method based on millimeter-wave radar according to claim 2, characterized in that If the area ratio of the movement trajectory space to the passage space exceeds the space area ratio threshold, or the ratio of the movement trajectory distance where the area of the movement trajectory space of the fall-detection personnel is higher than the area required for the movement of the fall-detection personnel does not exceed the movement trajectory distance ratio threshold, a high-risk signal is generated; if the area ratio of the movement trajectory space to the passage space does not exceed the space area ratio threshold, and the ratio of the movement trajectory distance where the area of the movement trajectory space of the fall-detection personnel is higher than the area required for the movement of the fall-detection personnel exceeds the movement trajectory distance ratio threshold, a low-risk signal is generated and sent to the guardian's terminal.

4. The fall detection method based on millimeter-wave radar according to claim 1, characterized in that, The process of action recognition and early warning is as follows: Obtain the center point of the personnel according to the body shape of the fall-detection personnel, and mark the center of gravity to obtain the center of gravity point; obtain the movement point according to the body shape and movement action of the fall-detection personnel, and mark it as the limb point according to the movement point and the movement position of the fall-detection personnel; when the fall-detection personnel move, count the position where the center of gravity point is located, and when there is no fall, the center of gravity points at each moment are linear in the activity area space, and collect the reciprocating cycle of the beating trajectory of the limb points of the fall-detection personnel when there is no fall, and mark it as the pendulum action.

5. The fall detection method based on millimeter-wave radar according to claim 4, wherein Perform action recognition on the fall-detection personnel through the signal wave, and collect the peak value of the instantaneous reduction speed of the height of the center of gravity point where the fall-detection personnel change the linear shape, and the increase span value of the swing amplitude of the limb point after the fall-detection personnel change the pendulum action of the corresponding limb point: If the peak value of the instantaneous reduction speed of the height of the center of gravity point where the fall-detection personnel change the linear shape exceeds the reduction speed peak threshold, but the height of the center of gravity point returns to the linear shape at the adjacent moment; if the swing amplitude of the limb point after the fall-detection personnel change the pendulum action of the corresponding limb point exceeds the amplitude growth span threshold, but the swing amplitude of the corresponding pendulum action drops to zero and then returns to the swing action cycle at the adjacent moment; When the above two situations occur simultaneously, mark the real-time action state of the fall-detection personnel as having a fall action and stabilizing the balance state, and send the action state to the guardian; If the peak value of the instantaneous reduction speed of the center-of-gravity position of the fall-detection person, which changes the height of the linear center-of-gravity position, exceeds the reduction speed peak threshold, and the center-of-gravity position does not return to the linear height; if the increase span value of the swing amplitude of the limb position of the fall-detection person after the pendulum movement of the corresponding limb position change exceeds the amplitude growth span threshold, but the swing amplitude of the corresponding pendulum movement at adjacent moments drops to zero and does not return to the swing movement period; When either of the above two situations occurs, the real-time action state of the fall-detection person is marked as having a fall action and not stabilizing the balance state, and the action state is sent to the caregiver.

Images