Fall detection method, apparatus and radar

Abstract

The application provides a fall detection method, device and radar, comprising: obtaining a track of a target, and obtaining an associated point cloud associated with the track of the target for each frame; determining a height mean of a highest point of the target, a height variation degree of the highest point of the target and a position variation degree of the target according to the associated point cloud of a first preset number of continuous frames, wherein the highest point is used to represent a point cloud with the maximum height value in each frame of the associated point cloud; judging a scene type in which the target is located according to the height mean of the highest point of the target, the height variation degree of the highest point of the target and the position variation degree of the target; after the scene type is determined, determining an energy height distribution of each frame of the associated point cloud, wherein the energy height distribution is used to represent a point cloud energy value of each height interval, and a total height of the detected target comprises a plurality of continuous but non-overlapping height intervals; and determining whether the target falls according to the scene type and the energy height distribution of each frame of the associated point cloud. The application can improve fall discrimination accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of radar detection, and in particular to a fall detection method, device and radar. BACKGROUND

[0002] With the aging phenomenon, the safety monitoring problem of the elderly is concerned. Among the potential risk factors in the life scene of the elderly, accidental falling accounts for a high proportion. Among the many falling cases, bathroom falling accounts for a high proportion.

[0003] At present, there are several types of fall detection methods: the first type is a video processing scheme based on visible light, which obtains a human behavior sequence through an optical sensor, and judges a fall event through image analysis. Due to the privacy problem, the video scheme is not suitable for fall detection in the shower scene; the second type is a fall detection based on sound signals. Due to the interference of water sound and the like in the shower process, the fall detection scheme based on sound signals is also not suitable for the shower scene; the third type is a wearable device, which obtains posture and position information through a micro sensor such as an acceleration sensor and a gyroscope, and judges whether to fall after processing the information. In the shower scene, the wearable device is not easy to accept; the fourth type is based on environmental perception, which mainly uses sensors placed in the bathroom to capture the influence of human behavior on signals, and analyzes signal echo data to make a fall judgment. This type of method has the characteristics of non-contact, not affecting normal life, privacy, and the like. The existing environmental perception scheme such as ultrasonic sensor has a high false negative rate and low reliability; the infrared device has low reliability due to water vapor interference and the like, and at the same time, high-performance infrared sensors also have the risk of privacy leakage; the fall detection based on millimeter wave radar is a non-contact detection means, which has strong applicability, does not invade privacy, and can be continuously monitored for a long time.

[0004] However, the current fall detection based on radar is mainly based on a sample set to train a model. Due to the fusion of people and water point clouds in the shower scene, the signal of the person is submerged, resulting in low detection accuracy. SUMMARY

[0005] Therefore, the present application provides a fall detection method, device and radar, which can solve the problem of low detection accuracy of the shower scene based on radar.

[0006] In a first aspect, an embodiment of the present application provides a fall detection method, comprising:

[0007] obtaining a track of a target, and obtaining an associated point cloud associated with the track of the target for each frame;

[0008] determine a height mean of the highest points, a height variation degree of the highest points, and a position variation degree of the target according to the continuous first preset number of the associated point clouds, the highest points being used to represent point clouds with the maximum height value in each frame of the associated point clouds;

[0009] determine a scene type in which the target is located according to the height mean of the highest points, the height variation degree of the highest points, and the position variation degree of the target;

[0010] after determining the scene type, determine an energy height distribution of each frame of the associated point clouds, the energy height distribution being used to represent point cloud energy values in each height interval, and a total height of the detected target including a plurality of continuous but non-overlapping height intervals;

[0011] determine whether the target falls down according to the scene type and the energy height distribution of each frame of the associated point clouds.

[0012] In a possible implementation, the determining the scene type in which the target is located according to the height mean of the highest points, the height variation degree of the highest points, and the position variation degree of the target includes:

[0013] if the height mean of the highest points is greater than or equal to a first preset threshold value, the height variation degree of the highest points is less than a second preset threshold value, and the position variation degree of the target is less than a third preset threshold value, it is determined that the scene type in which the target is located is a shower scene;

[0014] the determining whether the target falls down according to the scene type and the energy height distribution of each frame of the associated point clouds includes:

[0015] after determining that the scene type is the shower scene, performing target state determination once according to the energy height distribution of the associated point clouds every second preset number of frames to determine a state of the target, the state of the target including that the target is in an active state and that the target is in an inactive state;

[0016] if the state of the target changes from the active state to the inactive state, it is determined that the target falls down.

[0017] In a possible implementation, the determining the energy height distribution of each frame of the associated point clouds includes:

[0018] for each height interval, obtaining point clouds in which all height values in the frame of the associated point clouds belong to the height interval;

[0019] calculating an energy value of each point cloud according to an amplitude value of the point cloud;

[0020] According to the energy sum value of all point clouds in the height interval, a point cloud energy value of the height interval is obtained;

[0021] According to the point cloud energy value of each height interval, an energy height distribution of the associated point cloud of the frame is obtained.

[0022] In a possible implementation, after determining that the scene type is a shower scene, target state determination is performed once according to the energy height distribution of the associated point cloud every second preset frame number, and the state of the target is determined by:

[0023] All height intervals with a height greater than or equal to a first preset height are obtained as first target height intervals;

[0024] A first proportion is obtained by counting a proportion of the number of frames that meet a first determination condition in the second preset frame number of associated point clouds, the first determination condition being that a ratio of a point cloud energy total value of the first target height interval to a point cloud energy total value of all height intervals is greater than a fourth preset threshold value;

[0025] If the first proportion is greater than a first preset proportion, it is determined that the state of the target is an inactive state.

[0026] In a possible implementation, after determining that the scene type is a shower scene, target state determination is performed once according to the energy height distribution of the associated point cloud every second preset frame number, and the state of the target is determined by:

[0027] All height intervals with a height less than a second preset height are obtained as second target height intervals;

[0028] A second proportion is obtained by counting a proportion of the number of frames that meet a second determination condition in the second preset frame number of associated point clouds, the second determination condition being that a ratio of a point cloud energy total value of the second target height interval to a point cloud energy total value of all height intervals is greater than a fifth preset threshold value;

[0029] If the second proportion is greater than a second preset proportion, it is determined that the state of the target is an active state.

[0030] In a possible implementation, the height mean value of the highest point of the target, the height variation degree, and the position variation degree of the target are determined according to the associated point clouds of a continuous first preset frame number by:

[0031] A height array is obtained by obtaining a height value of the highest point of each frame of associated point cloud in the first preset frame number of associated point clouds;

[0032] The height mean value of the highest point of the target and the height variation degree are determined according to the height array.

[0033] obtain radial distance of each frame of the associated point cloud in the first preset frame number of associated point clouds, to obtain a radial distance array;

[0034] According to the radial distance array, determine the degree of change of the position of the target.

[0035] In a possible implementation, the obtaining of the radial distance of each frame of the associated point cloud in the first preset frame number of associated point clouds comprises:

[0036] For each frame of the associated point cloud in the first preset frame number of associated point clouds, according to the coordinate value of each point cloud in the frame of the associated point cloud after being projected to the geodetic coordinate system, the average value of the x-axis and the average value of the y-axis of all point clouds are calculated, to obtain a target coordinate point;

[0037] The distance between the target coordinate point and the origin of the geodetic coordinate system is calculated, to obtain the radial distance of the frame of the associated point cloud.

[0038] In a second aspect, the embodiments of the present application provide a fall detection device, comprising: an acquisition module, a first determination module, a first judgment module, a second determination module and a second judgment module;

[0039] The acquisition module is configured to acquire a track of a target, and acquire an associated point cloud associated with the track of the target for each frame;

[0040] The first determination module is configured to determine the average height of the highest point of the target, the degree of change of the height of the highest point of the target and the degree of change of the position of the target according to the associated point cloud of the first preset frame number of continuous frames, the highest point being used to represent the point cloud with the maximum height value in each frame of the associated point cloud;

[0041] The first judgment module is configured to determine the type of the scene where the target is located according to the average height of the highest point of the target, the degree of change of the height of the highest point of the target and the degree of change of the position of the target;

[0042] The second determination module is configured to determine the energy height distribution of each frame of the associated point cloud after determining the type of the scene, the energy height distribution being used to represent the point cloud energy value of each height interval, the total height of the target to be detected comprising a plurality of continuous but non-overlapping height intervals;

[0043] The second judgment module is configured to determine whether the target falls according to the type of the scene and the energy height distribution of each frame of the associated point cloud.

[0044] In a third aspect, an embodiment of the present application provides a radar, comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor implements steps of the method according to the first aspect or any possible implementation manner of the first aspect when running the computer program.

[0045] In a possible implementation manner, the radar is a millimeter wave radar.

[0046] Compared with the prior art, the embodiment of the present application has the beneficial effects that:

[0047] The embodiment of the present application determines the height mean value, the height change degree and the target change degree of the target through analyzing the associated point cloud of the target track and through multiple frames of continuous associated point cloud, judges the scene type based on this, and after determining the scene, acquires the energy height distribution of each frame of associated point cloud, judges whether the target falls based on the scene type and the multiple frames of energy height distribution, and the precision of fall judgment is improved. BRIEF DESCRIPTION OF DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0049] Figure 1 is an implementation flowchart of a fall detection method provided by the embodiment of the present application;

[0050] Figure 2 is a structural schematic diagram of a fall detection device provided by the embodiment of the present application;

[0051] Figure 3 is a schematic diagram of a radar provided by the embodiment of the present application. DETAILED DESCRIPTION

[0052] In the following description, specific details are set forth in order to provide a thorough understanding of the embodiments of the present application. However, persons skilled in the art will understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary details.

[0053] In order to make the objects, technical solutions and advantages of the present application clearer, the following will combine the specific implementation

[0054] Reference Figure 1Fig. 1 shows a flowchart of a method for detecting a fall according to an embodiment of the present application, which is described in detail as follows:

[0055] In step 101, a track of the target is obtained, and an associated point cloud of each frame associated with the track of the target is obtained.

[0056] In an embodiment of the present application, the point cloud data of the target is obtained by a radar, such as a millimeter wave radar, and the track of the target is obtained by target tracking.

[0057] In different application scenarios, the number of targets can be different. Taking a shower scenario as an example, due to the privacy of the shower scenario, if a fall event occurs when multiple people are present, timely assistance or alarm will be provided. Optionally, only a single person shower scenario is considered at this time, and the target referred to in the embodiment of the present application is a single target, and the track in the bathroom is a single track that stably exists. The associated point cloud of each frame is the point cloud associated with the single track.

[0058] In step 102, the height mean of the highest point of the target, the height variation degree, and the position variation degree of the target are determined according to the associated point cloud of the first preset number of continuous frames.

[0059] The highest point is used to represent the point cloud with the maximum height value in each frame of associated point cloud.

[0060] In an embodiment of the present application, the height variation degree can be the standard deviation, variance, etc. of the height variation of the highest point, and the position variation degree can be the standard deviation, variance, etc. of the position. The mathematical quantity used to represent the height variation degree and the position variation degree can be determined according to a specific application scenario, which is not limited in the embodiment of the present application.

[0061] The position variation degree is used to represent the position jitter degree of the target. For example, in a shower scenario, a person usually stands under a shower, and the position jitter degree is low. The centroid of the target can be determined from each frame of associated point cloud, and the position variation degree can be determined from the variation degree of the centroid. The radial distance of the target can also be determined from each frame of associated point cloud, and the position variation degree can be determined from the radial distance, or the position jitter degree can be calculated in other ways, which is not limited in the embodiment of the present application.

[0062] In an optional implementation, the height value of the highest point of each frame of associated point cloud in the first preset number of associated point clouds is obtained to obtain a height array; the height mean of the highest point of the target and the height variation degree are determined according to the height array; the radial distance of each frame of associated point cloud in the first preset number of associated point clouds is obtained to obtain a radial distance array; and the position variation degree of the target is determined according to the radial distance array.

[0063] In the embodiment of the present application, optionally, the geodetic coordinate system is adopted to obtain the coordinates of each point cloud in the geodetic coordinate system, and the value of the z axis in the coordinates is the height value of each point cloud. Each frame of associated point cloud has a point cloud with the maximum z value, that is, the highest point. Through the associated point cloud of the first preset number of continuous frames, the variation rule of the height value of the highest point can be obtained, and the height mean and the height variation degree are obtained.

[0064] The method provided by the embodiment of the present application can also adopt the Cartesian three-dimensional coordinate system as the world coordinate system to represent the position of the point cloud in the real world. Other coordinate systems can also be adopted, and the embodiment of the present application does not limit them.

[0065] Optionally, the highest point coordinates of each frame of associated point cloud are extracted, the single-frame highest point splicing can constitute a time sequence height curve, and the coordinates of each highest point constitute the highest point historical trajectory information.

[0066] In the embodiment of the present application, the fixed length can be calculated, for example, the first preset number of frames is 100 frames, and the height array can be obtained through the associated point cloud of the first preset number of continuous frames. Optionally, the height array is represented by h. Optionally, the height standard deviation is calculated by the first formula, the height variation degree is represented by the height standard deviation, and the first formula is:

[0067] stdH=std(h)

[0068] Wherein, std(h) is the standard deviation of the data in the height array h, and stdH is the height variation degree calculated by the height array.

[0069] The calculation method of the height mean of the highest point can have various implementation forms, for example, the height mean of the highest point is directly obtained by averaging the data in the height array, or the height mean is obtained by preprocessing the data in the height array and removing the abnormal point data. The embodiment of the present application does not limit the determination method of the height mean.

[0070] Optionally, the radial distance array is obtained through the associated point cloud of the first preset number of continuous frames, and the radial distance array is represented by rlist. Optionally, the distance standard deviation is calculated by the second formula, the position variation degree is represented by the distance standard deviation, and the second formula is:

[0071] stdR=std(rlist)

[0072] Wherein, std(rlist) is the standard deviation of the data in the radial distance array rlist, and stdR is the position variation degree calculated by the radial distance array.

[0073] Optionally, for each frame of the associated point cloud, the radial distance obtained from the frame of the associated point cloud can be directly obtained from the distance between the coordinate point obtained after Kalman filtering and the coordinate origin;

[0074] Optionally, for each frame of the associated point cloud in the first preset number of frames of the associated point cloud, the x-axis average value and the y-axis average value of all point clouds are calculated according to the coordinate value of each point cloud in the frame of the associated point cloud after being projected into the geodetic coordinate system, to obtain a target coordinate point; the distance between the target coordinate point and the origin of the geodetic coordinate system is calculated to obtain the radial distance of the frame of the associated point cloud.

[0075] For example, for a frame of the associated point cloud, the frame of the associated point cloud includes 20 point clouds, and the coordinates of each point cloud can be directly obtained. The x-axis average value of the 20 point clouds is calculated to obtain xcenter, and the y-axis average value of the 20 point clouds is calculated to obtain ycenter. The coordinate value of the target coordinate point is (xcenter, ycenter). Optionally, the radial distance of the frame of the associated point cloud is calculated by the third formula, and the third formula is:

[0076]

[0077] Wherein, r is used to represent the radial distance of the frame of the associated point cloud. The radial distance of the first preset number of frames, such as 100 frames of the associated point cloud, is calculated to obtain the radial distance array rlist.

[0078] In step 103, according to the height average of the highest point of the target, the height variation degree, and the position variation degree of the target, the scene type where the target is located is judged.

[0079] In the embodiment of the application, the scene where the target is located can be divided into a shower scene and a non-shower scene. If the scene where the target is located is a shower scene, the point cloud of water and the point cloud of a person are usually difficult to distinguish in the shower scene, and the detected target includes water and a person. Based on this, in the embodiment of the application, the basis for determining whether it is a shower scene is that the highest point of the target in the shower scene is the position of the shower head. Since the shower head is fixedly installed, the position of the highest point of the target is relatively high, the shaking degree of the highest point is low, and the person is usually located below the shower head during the shower process, that is, the position variation degree of the target is low.

[0080] Therefore, if the height average of the highest point of the target is high, the height variation degree is low, and the position variation degree is low, it is determined to be a shower scene. If the characteristics are not met, it is confirmed that the scene type where the target is located is a non-shower scene.

[0081] In an optional implementation, the judgment condition of the shower scene is that if the average height of the highest points is greater than or equal to a first preset threshold value, the height variation degree of the highest points is less than a second preset threshold value, and the position variation degree of the target is less than a third preset threshold value, the target is determined to be in the shower scene.

[0082] The first preset threshold value, the second preset threshold value and the third preset threshold value are not limited in specific values, and the values can be set according to specific application scenarios.

[0083] Due to the influence of sensor errors and the like, there may be a single frame without effective point clouds, that is, there is no effective point cloud in a frame of associated point clouds. In the embodiment of the application, in order to improve the judgment accuracy, each data in the height array and the radial distance array further includes flag data for indicating whether there is an effective point cloud in the current frame of associated point clouds. If there is an effective point cloud, the flag is set to 1, and if there is no effective point cloud, the flag is set to 0. Alternatively, based on the same inventive concept, the height array and the radial distance array correspond to a flag array, and the flag array is used to save information about whether there is an effective point cloud in each frame of associated point clouds in the first preset number of frames. The embodiment of the application does not limit the specific implementation form.

[0084] In combination with the flag array, if there is a single frame of associated point clouds without effective point clouds, in the process of determining the shower scene in this step, the proportion of the effective frame number in the first preset number of frames of associated point clouds should also be determined. When the proportion of the effective frame number is higher than a preset value, such as 95%, and the above-mentioned judgment condition is met, it can be determined to be a shower scene. The effective frame number refers to the number of frames with effective point clouds. When the proportion of the effective frame number is lower than the preset value, even if the above-mentioned judgment condition is met, it cannot be determined to be a shower scene, thereby improving the judgment accuracy of whether it is a shower scene. For example, the proportion of the number of flags with a value of 1 in the flag array is greater than 95%, and the above-mentioned judgment condition of the shower scene is met, and it is determined to be a shower scene at this time. Through the method provided in the embodiment of the application, the judgment error caused by too many invalid frame associated point clouds can be reduced, and the judgment accuracy is improved. The invalid frame associated point cloud is used to indicate that there is no effective point cloud in a single frame of associated point clouds.

[0085] In step 104, after determining the scene type, for each frame of associated point clouds, the energy height distribution of the frame of associated point clouds is determined.

[0086] The energy height distribution is used to represent the point cloud energy value of each height interval, and the total height of the detected target includes a plurality of continuous but non-overlapping height intervals.

[0087] The total height of the detected target is divided into a plurality of continuous but non-overlapping height intervals. For each frame of associated point cloud, according to the height values of the point clouds in the frame of associated point cloud, the height distribution of the frame of associated point cloud is counted, that is, it is counted into which height interval each point cloud falls, that is, it is counted which point clouds are included in each height interval.

[0088] For each frame of associated point cloud, the number of the frame of associated point cloud is the point cloud obtained by associating the tracked target track. The total number of point clouds in each frame of associated point cloud can be the same or different.

[0089] For example, the height of a general bathroom shower is set to about 2 meters. At this time, the total height of the detected target is 2 meters. The 2 meters is divided into 20 height intervals, each height interval is 10 cm, and the ground height corresponds to 0 meters. Therefore, the height from the ground to the shower is divided into 20 height intervals.

[0090] For a frame of associated point cloud, the frame of associated point cloud has 40 point clouds. Taking one height interval as an example, such as the height interval corresponding to 80 cm to 90 cm, there are 5 point clouds in the 40 point clouds whose height values fall into the height interval, that is, the height interval corresponding to 80 cm to 90 cm. By calculating the energy value represented by the 5 point clouds, the point cloud energy value of the height interval can be obtained. Using the same method, the point cloud energy value of each height interval can be obtained, and the energy height distribution of the frame of associated point cloud can be obtained.

[0091] In an optional implementation, for each height interval, all point clouds in the frame of associated point cloud whose height values belong to the height interval are obtained. According to the amplitude value of each point cloud, the energy value of the point cloud is calculated. According to the energy sum value of all point clouds in the height interval, the point cloud energy value of the height interval is obtained. According to the point cloud energy value of each height interval, the energy height distribution of the frame of associated point cloud is obtained.

[0092] Optionally, a height interval array HeightBin is established, the total length of the height interval array is HeightBinSize, the lowest height interval is MINH, the highest height interval is MAXH, and the total height is MAXH-MINH. The total height is also the total height of the detected target in this step.

[0093] The interval size of each height interval is binsize, and

[0094] For example, MAXH=200, MINH=0cm, binsize=10cm, then HeightBinSize=20, that is, the total height of the detected target is divided into 20 continuous but non-overlapping height intervals. Then the index number corresponding to each height interval in low to high order can be 1, 2, 3…20.

[0095] For each frame of associated point cloud, all height values in the frame of associated point cloud can be obtained by the following method:

[0096] All point clouds in the frame of associated point cloud are traversed, and for point cloud i, the index number of the height interval corresponding to the point cloud is calculated by the fourth formula, and the fourth formula is:

[0097]

[0098] Wherein, h((i) is the height value of point cloud i, ceil() is used to represent the minimum positive integer obtained in the parentheses, for example, if the value obtained in the parentheses is 5.6, then ceil(5.6)=6, and ind is the index number of the height interval corresponding to point cloud i.

[0099] By this method, the point cloud distribution of each height interval is obtained.

[0100] In the embodiment of the application, optionally, since the square of the amplitude of the point cloud is linearly related to the point cloud energy, the amplitude of the point cloud can be used to represent the point cloud energy.

[0101] Optionally, the amplitude of the point cloud is represented by amp, and for any point cloud i, the point cloud energy value can be represented by k(point(i).amp) 2 , wherein k is a preset coefficient, and point(i).amp is the amplitude of point cloud i. In the embodiment of the application, the point cloud data is obtained by the millimeter wave radar, and the point cloud amplitude is the data directly obtained in the process of generating the point cloud data by the millimeter wave radar.

[0102] For any index number corresponding to the height interval, the point cloud energy value of the height interval is obtained by adding the energy values of all point clouds in the height interval. Optionally, the point cloud energy value of each index number representing the height interval is represented by HeightBin(Ind).

[0103] In step 105, according to the scene type and the energy height distribution of each frame of associated point cloud, it is judged whether the target falls down.

[0104] Optionally, after judging that the scene type is the shower scene, a target state judgment is performed according to the energy height distribution of the associated point cloud every second preset frame number, to determine the state of the target, the state of the target including that the target is in an active state and that the target is in an inactive state; if the state of the target changes from the active state to the inactive state, it is judged that the target falls down.

[0105] Optionally, for each frame of associated point cloud, the energy proportion of each height interval is calculated in a normalized manner, specifically, for any height interval, the point cloud energy value of the height interval is divided by the sum of the point cloud energy values of all height intervals, to obtain the energy proportion of the height interval.

[0106] In the embodiment of the application, the energy height distribution of the associated point cloud of the continuous second preset frame number can also be spliced to generate a height heat map, and the state of the target is determined through the height heat degree. The state of the target can also be directly determined through the energy height distribution of the associated point cloud of the continuous second preset frame number.

[0107] For example, in the shower scene, when a person is taking a shower and the person is obviously active, the point cloud energy of the person is large, and the point cloud energy of the water is small, that is, when a person is taking a shower under the shower and the person is obviously active, the main energy of the frame of associated point cloud is in a low height place. The obvious activity includes actions such as washing and shampooing, and also includes the relatively obvious action of falling down. Based on this, in the embodiment of the application, the target in the active state in the shower scene includes a person under the shower and the person has obvious actions.

[0108] In the shower scene, after a person falls down under the water, the signal is generally weak, at this time, the energy of the water is dominant, that is, the energy proportion of the position with high height is high. In the embodiment of the application, since the state of no person and the state of weak life signal are similar, the target in the inactive state also includes the state of weak life signal.

[0109] For example, the energy height distribution of a single frame of associated point cloud is analyzed, and a height threshold is set, for example, 70 cm from the ground is set as the height threshold, if the point cloud energy above 70 cm is dominant in the frame of associated point cloud, it indicates that the state of the target corresponding to the frame of associated point cloud is the inactive state. Or, 140 cm from the ground is set as the height threshold, if the point cloud energy below 140 cm is dominant in the frame of associated point cloud, it is judged that the state of the target corresponding to the frame of associated point cloud is the active state.

[0110] In one optional implementation, all height intervals with heights greater than or equal to a first preset height are obtained as the first target height interval; the ratio of the number of frames satisfying the first discrimination condition to the number of frames in the associated point cloud of a second preset number of frames is calculated to obtain the first proportion, wherein the first discrimination condition is that the ratio of the total energy value of the point cloud in the first target height interval to the total energy value of the point cloud in all height intervals is greater than a fourth preset threshold; if the first proportion is greater than the first preset ratio, the target is determined to be inactive.

[0111] In one optional implementation, all height intervals with heights less than a second preset height are obtained as the second target height interval; the ratio of the number of frames satisfying the second discrimination condition to the number of frames in the associated point cloud of the second preset number of frames is calculated to obtain the second proportion, and the second discrimination condition is that the ratio of the total energy value of the point cloud in the second target height interval to the total energy value of the point cloud in all height intervals is greater than a fifth preset threshold; if the second proportion is greater than the second preset ratio, the target is determined to be in an active state.

[0112] Alternatively, if the scene type is determined to be a non-shower scene, the target state can be determined based on the energy height distribution of the associated point cloud at the second preset frame number. The determination criterion is the same as described above: if the target changes from an active state to an inactive state, it is determined that the target has fallen. In this case, by analyzing the non-shower scene, the first preset height, first target height range, first proportion, fourth preset threshold, and first preset ratio corresponding to the non-shower scene are confirmed, and the target state in the non-shower scene is determined to be inactive using the same method as above. Furthermore, by analyzing the non-shower scene, the second preset height, second target height range, second proportion, fifth preset threshold, and second preset ratio corresponding to the non-shower scene are confirmed, and the target state in the non-shower scene is determined to be active using the same method as above.

[0113] For example, the second preset frame count is 120 frames. A judgment is made every 120 frames by associating the point cloud data. The first preset ratio is set to 95%. If more than 95% of the 120 frames are judged to be in an unmanned state, then the result of this judgment is determined to be an unmanned state.

[0114] Assuming the second preset ratio is set to 60%, if more than 60% of the frames in 120 frames are identified as character activity, then the result of this identification is determined to be character activity.

[0115] In this embodiment of the invention, a judgment is performed every second preset number of frames, and the target's state is obtained each time. If the target's state changes from active to inactive, it is determined that the target has fallen, and an alarm is triggered promptly.

[0116] In the embodiment of the present application, the target can include other states in addition to the active state and the inactive state, and the embodiment of the present application does not analyze the other states. The target is transformed from the active state to the inactive state, which can be two consecutive discrimination results, or can be after obtaining the discrimination result that the target is in the active state, and then obtaining the discrimination result that the target is in the inactive state, which also satisfies the fall alarm condition.

[0117] The method provided by the embodiment of the present application has at least the following advantages: first, it does not need to be worn, and is installed in the corner of the bathroom, so that the members in the bathroom can be real-time safety guarded; second, there is no privacy problem, the point cloud data is obtained by the millimeter wave radar, the millimeter wave radar is in the overlapping area of far infrared and microwave, and there is no privacy leakage problem; third, the method provided by the embodiment of the present application can greatly reduce the interference of water, and is convenient to operate, has strong portability, and has low cost; fourth, compared with the conventional millimeter wave scheme, the alarm accuracy is high in the shower scene, and the reliability is strong; fifth, based on the height processing scheme, the radar installation position is not disturbed.

[0118] The embodiment of the present application analyzes the associated point cloud of the target track, determines the height mean, the height change degree and the target change degree of the target through multiple frames of continuous associated point cloud, judges the scene type based on this, obtains the energy height distribution of each frame of associated point cloud after determining the scene, judges whether the target falls based on the scene type and the multiple frames of energy height distribution, and improves the precision of fall discrimination.

[0119] It should be understood that the size of the serial number of each step in the above embodiment does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiment of the present application.

[0120] The following is the device embodiment of the present application, and for the details not described in detail, reference can be made to the corresponding method embodiments described above.

[0121] Figure 2 The structure schematic diagram of the fall detection device provided by the embodiment of the present application is shown, only the part related to the embodiment of the present application is shown for the convenience of description, and the details are as follows:

[0122] As shown in Figure 2 The fall detection device 2 includes an acquisition module 21, a first determination module 22, a first judgment module 23, a second determination module 24 and a second judgment module 25;

[0123] The acquisition module 21 is used for acquiring the track of the target and acquiring the associated point cloud associated with the track of the target of each frame;

[0124] The first determining module 22 is used to determine the average height of the highest point of the target, the degree of height change, and the degree of position change of the target based on the associated point cloud of a consecutive first preset number of frames. The highest point is used to represent the point cloud with the largest height value in each frame of the associated point cloud.

[0125] The first judgment module 23 is used to determine the scene type of the target based on the average height of the highest point of the target, the degree of height change, and the degree of position change of the target;

[0126] The second determining module 24 is used to determine the energy height distribution of the associated point cloud for each frame after determining the scene type. The energy height distribution is used to represent the point cloud energy value of each height interval. The total height of the detected target includes multiple continuous but non-overlapping height intervals.

[0127] The second judgment module 25 is used to determine whether the target has fallen based on the scene type and the energy height distribution of the associated point cloud in each frame.

[0128] This invention analyzes the associated point cloud of the target trajectory, determines the target's average altitude, altitude variation, and target variation through multiple consecutive associated point clouds, and makes a scene type determination based on this. After determining the scene, the energy altitude distribution of each frame of associated point cloud is obtained, and the accuracy of fall detection is improved based on the scene type and the energy altitude distribution of multiple frames.

[0129] In one possible implementation, the first judgment module 23 is used for:

[0130] If the average height of the highest point is greater than or equal to the first preset threshold, the degree of change in the height of the highest point is less than the second preset threshold, and the degree of change in the position of the target is less than the third preset threshold, then the scene type of the target is determined to be a shower scene.

[0131] The second judgment module 25 is used for:

[0132] After determining that the scene type is a shower scene, the target state is determined once based on the energy height distribution of the associated point cloud every second preset frame number. The target state includes whether the target is in an active state or an inactive state.

[0133] If the target's state changes from active to inactive, then the target is judged to have fallen.

[0134] In one possible implementation, the second determining module 24 is used for:

[0135] For each height interval, obtain all point clouds in the associated point cloud of that frame whose height values ​​belong to that height interval;

[0136] According to the amplitude value of each point cloud, an energy value of the point cloud is calculated;

[0137] According to the energy sum value of all point clouds in the height interval, a point cloud energy value of the height interval is obtained.

[0138] According to the point cloud energy value of each height interval, an energy height distribution of the associated point cloud of the frame is obtained.

[0139] In a possible implementation, the second determining module 25 is configured to:

[0140] obtain all height intervals with a height greater than or equal to a first preset height as first target height intervals;

[0141] obtain a first proportion by counting a proportion of the number of frames satisfying a first determination condition in the second preset number of frames, the first determination condition being that a ratio of a total point cloud energy value of the first target height intervals to a total point cloud energy value of all height intervals is greater than a fourth preset threshold value;

[0142] if the first proportion is greater than a first preset proportion, determining that the state of the target is an inactive state.

[0143] In a possible implementation, the second determining module 25 is configured to:

[0144] obtain all height intervals with a height less than a second preset height as second target height intervals;

[0145] obtain a second proportion by counting a proportion of the number of frames satisfying a second determination condition in the second preset number of frames, the second determination condition being that a ratio of a total point cloud energy value of the second target height intervals to a total point cloud energy value of all height intervals is greater than a fifth preset threshold value;

[0146] if the second proportion is greater than a second preset proportion, determining that the state of the target is an active state.

[0147] In a possible implementation, the first determining module 22 is configured to:

[0148] obtain a height array by obtaining a height value of a highest point of each associated point cloud in the first preset number of frames;

[0149] determine a height mean value and a height variation degree of the highest point of the target according to the height array;

[0150] obtain a radial distance array by obtaining a radial distance of each associated point cloud in the first preset number of frames;

[0151] determine a position variation degree of the target according to the radial distance array.

[0152] In one possible implementation, the first determining module 22 is used for:

[0153] For each frame of the associated point cloud in the first preset number of frames, the average x-axis and average y-axis of all point clouds are calculated based on the coordinate values ​​of each point cloud in that frame after being projected onto the geodetic coordinate system, to obtain the target coordinate point.

[0154] Calculate the distance between the target coordinate point and the origin of the geodetic coordinate system to obtain the radial distance of the associated point cloud in that frame.

[0155] The fall detection device provided in this embodiment can be used to execute the fall detection method embodiment described above. Its implementation principle and technical effect are similar, and will not be described again here.

[0156] Figure 3 This is a schematic diagram of a radar provided according to an embodiment of the present invention. Figure 3 As shown, the radar 3 in this embodiment includes a processor 30, a memory 31, and a computer program 32 stored in the memory 31 and executable on the processor 30. When the processor 30 executes the computer program 32, it implements the steps described in the various fall detection method embodiments above, for example... Figure 1 Steps 101 to 105 are shown. Alternatively, when the processor 30 executes the computer program 32, it implements the functions of each module / unit in the above-described device embodiments, for example... Figure 2 The functions of modules 21 to 25 are shown.

[0157] For example, the computer program 32 can be divided into one or more modules / units, which are stored in the memory 31 and executed by the processor 30 to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program 32 in the radar 3.

[0158] The radar 3 can be a millimeter-wave radar. The radar 3 may include, but is not limited to, a processor 30 and a memory 31. Those skilled in the art will understand that... Figure 3 This is merely an example of radar 3 and does not constitute a limitation on radar 3. It may include more or fewer components than shown, or combine certain components, or different components. For example, the radar may also include input / output devices, network access devices, buses, etc.

[0159] The processor 30 can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0160] The memory 31 can be an internal storage unit of the radar 3, such as a hard disk or a memory of the radar 3. The memory 31 can also be an external storage device of the radar 3, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory 31 can also include both the internal storage unit and the external storage device of the radar 3. The memory 31 is used to store the computer program and other programs and data required by the radar. The memory 31 can also be used to temporarily store data that has been output or will be output.

[0161] It can be clearly understood by those skilled in the art that, for the convenience and brevity of description, only the division of the above functional units and modules is exemplified, and in actual application, the above functions can be completed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to complete all or part of the above-described functions. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit, and the integrated unit can be realized in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction, and do not limit the protection scope of the present application. The specific working process of the units and modules in the above system can refer to the corresponding process in the foregoing method embodiments, and will not be described here.

[0162] In the above embodiments, the description of each embodiment has its own emphasis, and the parts not described or recorded in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0163] Those skilled in the art can understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0164] In the embodiments provided by the present application, it should be understood that the disclosed apparatus / radar and method can be implemented in other ways. For example, the apparatus / radar embodiments described above are merely schematic, for example, the division of the modules or units is only a logical function division, and there can be another division manner in actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or in other forms.

[0165] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e. they can be located in one place or distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0166] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically independently, or two or more units can be integrated into one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0167] The integrated module / unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. When the computer program is executed by a processor, the steps of each of the above-mentioned fall detection method embodiments can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files or some intermediate forms. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0168] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A fall detection method, characterized by, The method comprises the following steps: acquiring a track of a target and acquiring an associated point cloud associated with the track of the target for each frame; determining a height mean value of a highest point of the target, a height variation degree of the highest point of the target and a position variation degree of the target according to the associated point cloud of a continuous first preset frame number, the highest point being used to represent a point cloud with the maximum height value in each frame of the associated point cloud; judging a scene type in which the target is located according to the height mean value of the highest point of the target, the height variation degree of the highest point of the target and the position variation degree of the target; after determining the scene type, determining an energy height distribution of each frame of the associated point cloud, the energy height distribution being used to represent a point cloud energy value of each height interval, and a total height of the detected target comprising a plurality of continuous but non-overlapping height intervals; judging whether the target falls down according to the scene type and the energy height distribution of each frame of the associated point cloud.

2. The method of claim 1, wherein, The step of judging the scene type in which the target is located according to the height mean value of the highest point of the target, the height variation degree of the highest point of the target and the position variation degree of the target comprises: if the height mean value of the highest point is greater than or equal to a first preset threshold value, the height variation degree of the highest point is less than a second preset threshold value, and the position variation degree of the target is less than a third preset threshold value, it is judged that the target is located in a shower scene. The step of judging whether the target falls down according to the scene type and the energy height distribution of each frame of the associated point cloud comprises: after judging that the scene type is the shower scene, performing a target state judgment according to the energy height distribution of the associated point cloud of every second preset frame number to determine a state of the target, the state of the target comprising an active state of the target and an inactive state of the target; if the state of the target changes from the active state to the inactive state, it is judged that the target falls down.

3. The method of claim 1, wherein, The step of determining the energy height distribution of each frame of the associated point cloud comprises: for each height interval, acquiring all point clouds in which the height value of each point cloud belongs to the height interval in the frame of the associated point cloud; calculating an energy value of each point cloud according to an amplitude value of the point cloud; obtaining a point cloud energy value of the height interval according to the energy sum value of all point clouds in the height interval; obtaining the energy height distribution of the frame of the associated point cloud according to the point cloud energy value of each height interval.

4. The method of claim 2, wherein, The step of determining the state of the target according to the energy height distribution of the associated point cloud of every second preset frame number after judging that the scene type is the shower scene comprises: acquiring all height intervals with a height greater than or equal to a first preset height as first target height intervals; obtaining a first proportion by counting a proportion of the number of frames satisfying a first discrimination condition in the second preset frame number of the associated point cloud, the first discrimination condition being that a ratio of a point cloud energy total value of the first target height interval to a point cloud energy total value of all height intervals is greater than a fourth preset threshold value; if the first proportion is greater than a first preset proportion, it is determined that the state of the target is the inactive state.

5. The method of claim 2, wherein, The target state determination is performed according to the energy height distribution of the associated point cloud every second preset frame number after the scene type is determined as the shower scene. All height intervals with a height less than a second preset height are obtained as second target height intervals. A second proportion is obtained by counting a proportion of a frame number of the associated point cloud of the second preset frame number that meets a second discrimination condition to the second preset frame number, the second discrimination condition being that a ratio of a total point cloud energy value of the second target height interval to a total point cloud energy value of all height intervals is greater than a fifth preset threshold value. If the second proportion is greater than a second preset proportion, it is determined that the state of the target is an active state.

6. The method according to any one of claims 1 to 5, characterized in that, The height mean value of the highest point of the target, the height variation degree and the position variation degree of the target are determined according to the associated point cloud of the continuous first preset frame number, including: A height array is obtained by obtaining a height value of the highest point of each frame of the associated point cloud of the first preset frame number. The height mean value of the highest point of the target and the height variation degree are determined according to the height array. A radial distance array is obtained by obtaining a radial distance of each frame of the associated point cloud of the first preset frame number. The position variation degree of the target is determined according to the radial distance array.

7. The method of claim 6, wherein, The radial distance of each frame of the associated point cloud of the first preset frame number is obtained, including: For each frame of the associated point cloud of the first preset frame number, the x-axis mean value and the y-axis mean value of all point clouds are calculated according to the coordinate value of each point cloud projected into the geodetic coordinate system in the frame of the associated point cloud, to obtain a target coordinate point; The radial distance of the frame of the associated point cloud is obtained by calculating the distance between the target coordinate point and the origin of the geodetic coordinate system.

8. A fall detection apparatus characterized by comprising: It includes: An obtaining module, a first determining module, a first judging module, a second determining module and a second judging module; The obtaining module is configured to obtain a track of a target and obtain an associated point cloud of each frame related to the track of the target; The first determining module is configured to determine a height mean value of a highest point of the target, a height variation degree and a position variation degree of the target according to the associated point cloud of the continuous first preset frame number, the highest point being a point cloud with the maximum height value in each frame of the associated point cloud; The first judging module is configured to determine a scene type in which the target is located according to the height mean value of the highest point of the target, the height variation degree and the position variation degree of the target; The second determining module is configured to determine, after the scene type is determined, the energy height distribution of each frame of the associated point cloud, the energy height distribution being used to represent a point cloud energy value of each height interval, and the total height of the detected target including a plurality of continuous but non-overlapping height intervals; The second judging module is configured to determine whether the target falls according to the scene type and the energy height distribution of each frame of the associated point cloud.

9. A radar comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the method of any one of claims 1 to 7.

10. The radar of claim 9, wherein, The radar is a millimeter wave radar.

Images