JP2023026123A - Human body attitude detection method and device, and data processing device - Google Patents

Abstract

To provide a human body attitude detection method and device, and a data processing device.SOLUTION: A method includes: determining the position where a human body is based upon change of angle FFT data of a radar; collecting statistics of height information on the human body at the position where the human body is based upon a difference in angle FFT data of the radar; and detecting the attitude of the human body based upon the height information on the human body.SELECTED DRAWING: Figure 1

Description

The present invention relates to the field of information technology, and more particularly to a human body pose detection method, device and data processing device.

Posture detection is the basis for recognizing people's activities and behaviors, and is widely used in fields such as surveillance systems, smart homes, and elderly care. With the development of technologies such as computer vision and deep learning, video-based pose detection has reached fairly high recognition accuracy and is the current mainstream pose detection scheme. However, because of its environmental sensitivity, video technology cannot be used in dark environments, and the detection accuracy can be degraded in complex background environments. In addition, videos do not protect people's privacy well and cannot be used in private places such as bathrooms, bedrooms, and so on.

The inventor discovered the following. Radar-based attitude detection technology is not affected by light or background environment, does not invade people's privacy, and can complement video, so it can be applied to scenes where video cannot be used. be able to. However, existing radar-based attitude detection methods mainly rely on point clouds for attitude detection. A point cloud is formed by capturing motion characteristics of a human body with radar, and includes position, speed information, and the like. The radar point cloud reflects the instantaneous motion characteristics of the human body, and the results are very sensitive and susceptible to the relative position, relative velocity, etc. between the radar and the target. The same motion pose can represent different point cloud data at different times, which can lead to poor generalization performance of the algorithm. For postures with no apparent movement, such as lying down or sitting, the radar cannot capture the point cloud information. Therefore, in such a case, posture detection cannot be performed using the point cloud.

To solve the above-mentioned problems or other similar problems, the present invention at least aims to provide a human body pose detection method that does not use a point cloud but uses signal fluctuations caused by the motion of the human body for pose detection. Make it an issue.

According to one aspect of an embodiment of the present invention, there is provided a human body posture detection method, the method comprising:
Determining the position of the human body based on the variation of the radar angle FFT data;
performing statistics on the height information of the human body at the position where the human body is located based on the difference of the radar angle FFT data; and detecting the posture of the human body based on the height information of the human body.

According to another aspect of an embodiment of the present invention, there is provided a human body posture detection device, the device comprising:
A first determination unit for determining the position of the human body based on the variation of the radar angle FFT data;
a statistical unit for performing statistics on the height information of the human body at the position where the human body is located based on the difference of the angle FFT data of the radar; and a detection unit for detecting the posture of the human body based on the height information of the human body. .

According to another aspect of an embodiment of the present invention, there is provided a data processing apparatus, the data processing apparatus including a processor and a memory, the memory storing a computer program, the processor comprising: By executing the computer program, it is configured to realize the following human body posture detection method, namely,
Determining the position of the human body based on the variation of the radar angle FFT data;
A method comprising: performing statistics on the height information of the human body at the position where the human body is located based on the difference of the radar angle FFT data; and detecting the posture of the human body based on the height information of the human body. .

The advantageous effect of the present invention is as follows: the method in the embodiment of the present invention does not use the point cloud, but can use the signal fluctuations caused by the motion of the human body to perform pose detection. Therefore, it is suitable for detecting postures for which a point cloud cannot be formed, such as standing, sitting on a stool, sitting on the ground, and lying down. It has good optimization performance, low complexity, and high detection accuracy.

It is a figure which shows the human body posture detection method in the Example of this invention. FIG. 4 is a diagram showing an example of determining the position of a human body based on variations in radar angle FFT data; FIG. 4 is a diagram showing a method of performing statistics on height information of a human body at a position where the human body is located based on a difference in angle FFT data; FIG. 4 is a diagram showing a method of detecting the posture of a human body based on height information of the human body; FIG. 5 is a diagram showing another method of detecting the posture of a human body based on height information of the human body; 4 is a diagram showing a human body height distribution _PH generated based on human body height information H. FIG. FIG. 10 is a diagram showing another human body height distribution _PH generated based on human body height information H; It is a figure which shows the human body posture detection apparatus in the Example of this invention. It is a figure which shows an example of a 1st determination unit. FIG. 4 is a diagram showing an example of a statistical unit; FIG. 4 is a diagram showing an example of a detection unit; FIG. 10 is a diagram showing another example of a detection unit; It is a figure which shows the data processing apparatus in the Example of this invention.

Preferred embodiments for carrying out the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that such an embodiment is merely an example and does not limit the present invention.

In an embodiment of the present invention, all microwave radars are FMCW (Frequency Modulated Continuous Wave) based radars. Radar periodically transmits radio signals into space, receives reflected signals, and then mixes the transmitted and received signals to produce an intermediate frequency signal. By analyzing and processing the data of the intermediate frequency signal, the radar can obtain information such as the distance, velocity and angle of objects in space. FMCW-based radar transmits multiple linear frequency modulated pulses in each period, referred to as a set of linear frequency modulated pulse frames. A range FFT (Fast Fourier Transform) processes the reflected signal of one linear frequency modulated pulse and separates the overlapping signals by range, whose phase is the distance between the object and the radar at this range. reflects slight changes in Velocity FFT processes multiple linear frequency modulated pulses continuously reflected at a distance to obtain Doppler velocity information of the object at that distance. Angle FFT processes the reflected signals of the same pulse received by multiple antennas at a certain distance or the signals of the same Doppler frequency points (frequency points) to obtain the angle information of the object at that distance. By combining the above distance information and angle information, the spatial coordinates of the object can be further calculated. Since the distance FFT, velocity FFT, and angle FFT are general-purpose techniques for radar signal processing, detailed description thereof will be omitted here.

For example, the radar angle FFT result S _t finally obtained through the processing of distance FFT, velocity FFT, and angle FFT at time t can be expressed by the following formula (1).

where r is the range frequency point of the radar, 1≤r≤N _r , N _r is the maximum number of range frequency points, a is the horizontal angle frequency point of the radar, 1≤a≤N _a , N _a is the maximum number of horizontal angular frequency points, e is the vertical angular frequency point of the radar, 1 ≤ e ≤ N _e , N _e is the maximum number of vertical angular frequency points, s _t ^{r , a and e} are the angle FFT results of the range frequency point r, the horizontal angle frequency point a, and the vertical angle frequency point b corresponding to the radar at time t. s _t ^r,a,e is one complex number and can be expressed by the following formula (2).

where A _tr ^,a,e and φ _tr ^,a,e are the amplitude and phase of s _tr ^,a,e respectively.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. It should be noted that these examples are merely illustrative and do not limit the present invention.

<First aspect of the embodiment>
This embodiment provides a human body posture detection method. FIG. 1 is a diagram showing an example of a human body posture detection method according to an embodiment of the present invention. As shown in FIG. 1, the method includes the following steps.

101: Determine the position of the human body based on the variation of the radar angle FFT data;
102: Perform statistics on the height information of the human body at the position where the human body is located based on the difference in the radar angle FFT data;
103: Detect the posture of the human body based on the height information of the human body.

Since the method in the embodiments of the present invention can perform pose detection using signal fluctuations caused by the motion of the human body without using a point cloud, it can detect poses for which a point cloud cannot be formed, such as standing. It is suitable for detecting postures such as standing, sitting on a stool, sitting on the ground, lying down, etc. The algorithm has good generalization performance, low complexity, and high detection accuracy.

Although FIG. 1 illustrates an embodiment of the present invention, the present invention is not limited to this. For example, the execution order between operations can be adjusted appropriately, or some operations can be increased or decreased. That is, those skilled in the art are not limited to the description of FIG. 1 above, and can appropriately modify the above contents.

It is more difficult for the human body to stand still than on a fixed object such as a wall or table. Therefore, radar signals reflected from fixed objects change less over time, and angle FFT data fluctuate less. On the other hand, the radar signal reflected from the human body changes greatly over time, and the angular FFT data fluctuates greatly. In an embodiment of the present invention, 101 can determine the position of the human body based on the variation of the angle FFT data.

FIG. 2 is a diagram showing an example of determining the position of a human body based on fluctuations in radar angle FFT data. As shown in FIG. 2, the method includes the following steps.

201: Calculate the variation of the angular FFT data at the distance frequency points and the horizontal angular frequency points;
202: searching for a local maximum of variation of the angular FFT data at said range frequency points and horizontal angle frequency points;
203: Determining whether there is a human body at the location of the local maximum based on the local maximum.

In an embodiment of the present invention, the angular FFT data variation is one three-dimensional value corresponding to the distance frequency points, the horizontal angular frequency points and the vertical angular frequency points. At 201, the amount of computation can be reduced by compressing the variation of the angle FFT data from three dimensions to two dimensions.

For example, for each distance frequency point and horizontal angle frequency point, calculate the variation of the angle FFT data at all vertical angle frequency points corresponding to the above distance frequency point and horizontal angle frequency point, and at all vertical angle frequency points The maximum value or average value of the variation of the angle FFT data can be the variation of the angle FFT data of the current distance frequency point and horizontal angle frequency point.

In an embodiment of the present invention, within a given time period T, the variation of the angle FFT data can be represented by V, where V={ _Vr,a,e }, where _Vr,a,e is the range frequency point r , horizontal angular frequency point a, and vertical angular frequency point e, or V _r,a,e is the angular FFT data variation corresponding to frequency point (r,a,e) where r is the range frequency point, a is the horizontal angular frequency point, and e is the vertical angular frequency point.

In some embodiments, the variation in the radar angle FFT data can be represented by the mean value of the angle FFT data differences. For variation in angle FFT data at each vertical angle frequency point corresponding to each distance frequency point and horizontal angle frequency point, the difference in angle FFT data within a predetermined period of the distance frequency point, horizontal angle frequency point and vertical angle frequency point can be calculated.

For example, the average value of the difference of the angle FFT data within a predetermined period of the distance frequency point, the horizontal angle frequency point, and the vertical angle frequency point is calculated, and the average value of the difference of the angle FFT data is calculated as the distance frequency point, It can be the variation of the angular FFT data of the horizontal angular frequency points and the vertical angular frequency points.

In the embodiment of the present invention, the average value of the difference of the angle FFT data is the average value of the difference of the angle FFT data within the predetermined period T of the same distance frequency point, horizontal angle frequency point and vertical angle frequency point. It can be obtained by calculating using formula (3).

Here, _Dtr ^,a,e is the angle FFT data difference of the distance frequency point r, the horizontal angle frequency point a, and the vertical angle frequency point e at time t. The angle FFT data difference can be obtained by calculating using formula (4) or (5) as follows.

In the embodiment of the present invention, the radar periodically transmits radio signals, and for each frame of radar data, angle FFT information as shown in formula (1) can be obtained. The difference between the angle FFT data at different times is the difference between the angle FFT value at the current time and the angle FFT value at the previous time, or the angle FFT value at the current time and the angle FFT value at some previous time It is the difference from the average value of angle FFT values. For example, the difference in angle FFT data at time t is the difference between the angle FFT value at time t and the angle FFT value at time t-1, or the angle FFT value at time t and some number before time t. It is the difference from the average value of the angle FFT values at the time.

In some embodiments, the difference in angle FFT data within a predetermined period of time for the same distance frequency point, horizontal angle frequency point and vertical angle frequency point is the same distance frequency point, horizontal angle frequency point and vertical angle frequency point for a predetermined period of time. It is the absolute value of the angle FFT amplitude difference at different times in the period.

For example, at time t, the absolute value of the angular FFT amplitude difference is denoted by D _t , where D _t ={D _t ^r,a,e }, where D _t ^r,a,e is the range frequency point r, The absolute value of the difference between the angular FFT amplitudes corresponding to horizontal angular frequency point a and vertical angular frequency point e.

In other words, the angle FFT amplitude difference _Dtr ^,a,e is the absolute value of the angle FFT amplitude difference at the same distance frequency point, horizontal angle frequency point and vertical angle frequency point at time t.

In some embodiments, the difference in angle FFT data within a predetermined period of time for the same distance frequency point, horizontal angle frequency point and vertical angle frequency point is the same distance frequency point, horizontal angle frequency point and vertical angle frequency point for a predetermined period of time. is the modulus of the angular FFT complex difference at different times in the period.

For example, at time t, the angular FFT complex difference modulus is denoted by D' _t , where D' _t = {D' _t ^r,a,e }, where D' _t ^r,a,e is the range frequency Modulus of the angular FFT complex difference corresponding to point r, horizontal angular frequency point a, and vertical angular frequency point e.

In other words, the modulus D′ _t ^r,a,e of the difference of the angle FFT complex numbers is the modulus of the difference of the angle FFT complex numbers at time t of the same range frequency point, horizontal angle frequency point and vertical angle frequency point.

In some embodiments, the variation of the radar angle FFT data is expressed using the standard deviation of the angle FFT amplitudes. Standard deviation of angle FFT amplitude within a predetermined period of distance frequency point, horizontal angle frequency point and vertical angle frequency point for variation of angle FFT data at each vertical angle frequency point corresponding to each distance frequency point and horizontal frequency point can be calculated.

For example, the standard deviation of the angle FFT amplitude within a predetermined period of the distance frequency point, the horizontal angle frequency point and the vertical angle frequency point is calculated, and the standard deviation of the angle FFT amplitude is calculated as the distance frequency point and the horizontal angle frequency point. and the variation of the angular FFT data of the vertical angular frequency points.

In an embodiment of the present invention, the standard deviation of the angular FFT amplitudes is the standard deviation of the amplitudes of the same range frequency point, horizontal angle frequency point and vertical angle frequency point within a given time period, using formula (6) below: can be obtained by calculating

Here, E(A _r,a,e ) is the average value of the angle FFT amplitudes within a predetermined period T of the distance frequency point r, the horizontal angle frequency point a, and the vertical angle frequency point e. It can be obtained by averaging the sum of angle FFT amplitudes at each time.

In some embodiments, the angle FFT amplitude within a predetermined period of a distance frequency point, a horizontal angle frequency point and a vertical angle frequency point is different within a predetermined period of the distance frequency point, the horizontal angle frequency point and the vertical angle frequency point. Amplitude of angle FFT at time.

In an embodiment of the present invention, at 202, local maximum refers to the maximum value of variation of the angle FFT data at range frequency points and horizontal angle frequency points within a preset range. The present invention does not limit the preset range, and the preset range refers to the range in which the human body may exist, which may be empirically set, and the human body pose detection location or It may be set based on the range of human body pose detection, but it depends on the specific implementation.

In an embodiment of the present invention, at 203, (r _h , a _h ) is used to represent the location of one local maximum, where r _h is the distance frequency point corresponding to the local maximum, and a _h is the horizontal angular frequency point corresponding to the local maximum. If the variation value of the angle FFT data corresponding to the distance frequency point and the horizontal angle frequency point at the position (r _h , a _h ) where the local maximum is located is greater than a preset threshold (referred to as the first threshold), It is determined that there is a human body at the location where the local maximum is located, otherwise it is determined that there is no human body at the location where the local maximum is located.

In the embodiment of the present invention, after obtaining the position (r _h , _ah ) of the human body, at 102, statistically calculate the height information of the human body at the position of the human body based on the difference of the angle FFT data. It can be performed.

FIG. 3 is a diagram showing a method of performing statistics on the height information of the human body at the position where the human body is located based on the difference in angle FFT data. As shown in FIG. 3, the method includes the following steps.

301: determining a range of the human body based on the position of the human body;
302: calculating the difference of the angular FFT data of all frequency points within the range for each time point in the predetermined period;
303: if the maximum value of the differences in the angle FFT data is greater than a preset threshold (referred to as a second threshold), the height information of the vertical angle frequency point corresponding to the maximum value is transferred to the human body at the current time; By using height information, height information of the human body within the predetermined period is obtained.

In 301, if the position where the human body is located is (r _h , _ah ), r _h is the distance frequency point, and _ah is the horizontal angle frequency point, then the range where the human body is located is the distance frequency The point is greater than r _h -L _r and less than r _h +U _r , the horizontal angle frequency point is greater than a _h -L _a and less than a _h +U _a , and the vertical angle frequency point is It can be a region larger than Le and smaller than _{Ue .} where L _r and U _r are preset distance frequency point thresholds, L _a and U _a are preset horizontal angular frequency point thresholds, and L _e and U _e are preset vertical angular frequency point thresholds. is. The present invention does not limit the setting method and specific values of the distance frequency point threshold, the horizontal angle frequency point threshold, and the vertical angle frequency point threshold. can be

In the above, the method for determining the range occupied by the human body based on the position of the human body has been described as an example, but the present invention is not limited to this, and another method is used to determine the range occupied by the human body based on the position of the human body. A range can also be determined.

In 302 and 303, one distance frequency point, horizontal angle frequency point and vertical angle frequency point are referred to as one frequency point (frequency point), the present invention based on the difference information of the angle FFT data, within T period For each time, determine the height belonging to the human body.

Suppose H is a data list that stores the heights belonging to the human body at each time point within T period. At 302 and 303, calculate the maximum difference of the angle FFT data at all frequency points within the range occupied by the human body at time t, and determine whether the maximum difference of the angle FFT data is greater than a second threshold. and if the maximum difference of the angle FFT data is greater than a second threshold, determine that the position corresponding to the maximum difference of the angle FFT data belongs to the human body, and determine the height corresponding to the vertical angle frequency point. H is added to the height of the human body at time t, and if the maximum difference of the angle FFT data is less than or equal to the second threshold, determine that the position corresponding to the maximum difference of the angle FFT data does not belong to the human body. . Thus, the height information belonging to the human body within T period can be obtained, H = {h _i |1 ≤ i ≤ N _h }, where N _h is the number of heights belonging to the human body be.

In some embodiments, the difference in angle FFT data for each frequency point at the current time is the absolute value of the angle FFT amplitude difference between the current time and the previous time for that frequency point. In some embodiments, the angle FFT data difference for each frequency point at the current time is the modulus of the angle FFT complex difference between the frequency point's current time and the previous time. Since the method of calculating the difference between the angle FFT data has already been described, detailed description thereof will be omitted here.

In an embodiment of the present invention, after obtaining the height information of the human body, at 103, the posture of the human body can be detected based on the height information of the human body.

FIG. 4 is a diagram showing a method of detecting the posture of a human body based on height information of the human body. As shown in FIG. 4, the method includes the following steps.

401: Calculate the average value (h) of the height information of the human body;
402: Calculate the ratio (r) (number ratio) of the height information of the human body that is higher than a preset threshold (referred to as a third threshold);
403: Height average value thresholds (L _H and U _H ) and ratio thresholds (L _r and U _r ) whose height is higher than the third threshold corresponding to each preset posture, and the height of the human body The posture of the human body is determined based on the ratio (r) of the average value (h) of the height information and the height information of the human body based on the ratio (r) in which the height is higher than the third threshold.

In 401, the average value h of height information of the human body can be obtained by calculating using the following formula (7).

Here, h _i is the height belonging to the human body, and h _i εH, and H is the aforementioned data list, which stores the height belonging to the human body at each time within the period T.

At 402, the ratio r of heights higher than the third threshold among the height information of the human body can be calculated by the following formula (8).

where C is the number of heights in H above the third threshold, and N _h is the number of all heights in H.

At 403, for each pose, define thresholds L _H and U _H for the average height value, and thresholds L _r and U _r for the ratio of height above the threshold. If the actually calculated mean height value h is greater than L _H and less than U _H and the ratio r of heights higher than the third threshold is greater than L _r and less than U _r , then The pose of the human body can be determined to be the pose corresponding to the threshold of the mean height value and the threshold of the ratio of heights higher than the threshold.

FIG. 5 is a diagram showing another method of detecting the posture of the human body based on height information of the human body. As shown in FIG. 5, the method includes the following steps.

501: Determine the human body height distribution _PH based on the human body height information;
502: Determine the pose of the human body based on the human body height distribution {P _j } corresponding to different preset poses.

Still taking the above as an example, suppose that the height information H={h _i |1≤i≤N _h } belonging to the human body within T period, and N _h is the number of all heights in H do. At 501, based on the height information H of the human body, a human body height distribution _PH can be generated. FIG. 6 shows one of the human body height distributions P _H generated based on the human body height information H, and FIG. 7 shows the human body height distribution P H generated based on the human body height information H. FIG. 11 shows another one of _H. FIG.

In 502, preset the body height distributions {P _j } corresponding to different poses, and then calculate the distance between the body height distributions P _H and the body height distributions {P _j } corresponding to each pose as The pose corresponding to the minimum distance is determined as the pose of the human body.

For example, it can be determined that the posture of the human body is "standing" based on _{the human body height distribution PH in} FIG. It can be determined that

The present invention does not limit the above distance calculation method, in some embodiments, the above distance may be the KL divergence (Kullback-Leibler divergence), and is calculated by the following formula (9): can be obtained by

Among them, P _j is the height distribution of the given posture j, PH is the above-mentioned human body height distribution _, L _h and U _h are the minimum and maximum human body heights, and P _j (h ) is the probability of height h in the height distribution of posture j, and P _H (h) is the probability of height h in the above-described human body height distribution.

Although each operation or process related to the present invention has been described above, the present invention is not limited to this. The method may further include other operations or processes, and reference can be made to the prior art for the specific content of these operations or processes.

Since the method in the embodiments of the present invention does not use a point cloud, but uses the signal fluctuations caused by the motion of the human body to perform pose detection, it can detect poses that cannot form a point cloud, such as standing. It is suitable for detecting postures such as standing, sitting on a stool, sitting on the ground, lying down, etc. The algorithm has good generalization performance, low complexity, and high detection accuracy.

<Second aspect of the embodiment>
This embodiment provides a human body posture detection device. Since the human body posture detection device corresponds to the human body posture detection method in the first aspect of the embodiment, redundant description of the same content will be omitted here.

FIG. 8 is a diagram showing a human body posture detection device according to an embodiment of the present invention. As shown in FIG. 8, the human body posture detection device 800 in an embodiment of the present invention includes a first determination unit 801, a statistics unit 802 and a detection unit 803. As shown in FIG. The first determination unit 801 is used to determine the position of the human body based on the variation of the radar angle FFT data, and the statistical unit 802 is used to determine the position of the human body based on the difference of the radar angle FFT data. It is used to make statistics on the height information of the human body, and the detecting unit 803 is used to detect the posture of the human body according to the height information of the human body.

FIG. 9 is an example diagram of a first determining unit 801. As shown in FIG. 9, in some embodiments, the first determining unit 801 includes a first computing unit 901, a searching unit 902 and a second determining unit. Including 903. The first calculation unit 901 is used to calculate the variation of the angle FFT data at the distance frequency points and the horizontal angle frequency points, and the search unit 902 is used to locally calculate the variation of the angle FFT data at the distance frequency points and the horizontal angle frequency points. It is used to search for the maximum value, and the second determining unit 903 is used to determine whether there is a human body at the location of the local maximum based on the local maximum.

In some embodiments, calculating the variation of the angle FFT data at the distance frequency points and the horizontal angle frequency points by the first calculation unit 901 is performed by calculating all the vertical angles corresponding to each distance frequency point and the horizontal angle frequency points. calculating the variation of the angle FFT data at the frequency points, and taking the maximum value or average value of the variation of the angle FFT data at all the vertical angle frequency points as the variation of the angle FFT data at the distance frequency points and the horizontal angle frequency points; including.

In some embodiments, calculating the variation of the angle FFT data at each vertical angle frequency point corresponding to each distance frequency point and horizontal angle frequency point by the first calculation unit 901 is performed by calculating the distance frequency point, the horizontal angle calculating an average value of differences in angle FFT data within a predetermined period of the frequency points and vertical angle frequency points, and calculating the average value of the differences in the angle FFT data as angles of the distance frequency points, the horizontal angle frequency points and the vertical angle frequency points; Including as variation of FFT data.

In some embodiments, the difference in angle FFT data within a predetermined time period of the distance frequency points, the horizontal angle frequency points and the vertical angle frequency points is:
absolute values of differences in angle FFT amplitudes at different times within said predetermined time period of said distance frequency points, horizontal angle frequency points and vertical angle frequency points; or said predetermined of said distance frequency points, horizontal angle frequency points and vertical angle frequency points. is the modulus of the angular FFT complex difference at different times in the period.

In some embodiments, calculating the variation of the angle FFT data at each vertical angle frequency point corresponding to each distance frequency point and horizontal angle frequency point by the first calculation unit 901 is performed by calculating the distance frequency point, the horizontal angle calculating the standard deviation of the angle FFT amplitudes within a predetermined period of the frequency points and the vertical angle frequency points, and calculating the standard deviation of the angle FFT amplitudes from the variation of the angle FFT data of the distance frequency points, the horizontal angle frequency points and the vertical angle frequency points; including

In some embodiments, the angular FFT amplitude within a predetermined period of the distance frequency points, the horizontal angle frequency points and the vertical angle frequency points is the angle FFT amplitude within the predetermined period of the distance frequency points, the horizontal angle frequency points and the vertical angle frequency points. It is the amplitude of the angle FFT at different times.

In some embodiments, the local maximum is the maximum variation of the angle FFT data at the range frequency points and horizontal angle frequency points within a preset range.

In some embodiments, when the variation of the angle FFT data of the distance frequency points and the horizontal angle frequency points at the location of the local maximum is greater than a first threshold, the second determining unit determines the local It is determined that there is a human body at the location of the local maximum, otherwise the second determining unit determines that there is no human body at the location of the local maximum.

FIG. 10 is a diagram showing an example of the statistics unit 802. As shown in FIG. As shown in FIG. 10, in some embodiments, the statistics unit 802 includes a third determination unit 1001, a second calculation unit 1002 and a fourth determination unit 1003. As shown in FIG. A third determining unit 1001 determines a range in which the human body is located based on the position in which the human body is located, and for each time of a predetermined period of time, a second calculating unit 1002 calculates all distance frequency points within the range, horizontal calculating the difference between the angle FFT data of the angle frequency point and the vertical angle frequency point, and if the maximum value of the angle FFT data difference is greater than a second threshold, the fourth determining unit 1003 corresponds to the maximum value; By using the height information of the vertical angular frequency point to be the height information of the human body at the current time, it is used to acquire the height information of the human body within the predetermined period.

In some embodiments, if the location of the human body is (r _h , _ah ), where r _h is the distance frequency point and _ah is the horizontal angle frequency point, the third determining unit 1001 may: The range in which the human body is located has a distance frequency point greater than r _h -L _r and less than r _h +U _r , and a horizontal angle frequency point greater than a _h -L _a and less than a _h +U _a . and determining that the vertical angle frequency point is a region greater than L _e and less than U _e , wherein L _r and U _r are preset distance frequency point thresholds, and L _a and U _a are preset is the preset horizontal angular frequency point threshold, and L _e and U _e are the preset vertical angular frequency point thresholds.

In some embodiments, the difference between the angle FFT data of each distance frequency point, horizontal angle frequency point and vertical angle frequency point at the current time is the current time of said distance frequency point, horizontal angle frequency point and vertical angle frequency point. It is the absolute value of the angle FFT amplitude difference between the time and the previous time.

In some embodiments, the difference between the angle FFT data of each distance frequency point, horizontal angle frequency point and vertical angle frequency point at the current time is the current time of said distance frequency point, horizontal angle frequency point and vertical angle frequency point. It is the modulus of the angle FFT complex difference between the time and the previous time.

FIG. 11 is a diagram showing an example of the detection unit 803. As shown in FIG. As shown in FIG. 11, in some embodiments, the detection unit 803 includes a third computing unit 1101, a fourth computing unit 1102 and a fifth determining unit 1103. As shown in FIG. A third calculation unit 1101 calculates an average value h of the height information of the human body, and a fourth calculation unit 1102 calculates a ratio (number ratio) r of the height information of the human body whose height is greater than a third threshold. , and the fifth determination unit 1103 sets thresholds L _H and U _H for the average height value and thresholds L _r and U _r for the ratio of heights higher than the third threshold corresponding to each preset pose, and It is used to determine the posture of the human body based on the average value h of the height information of the human body and the ratio r of the height information of the human body whose height is higher than the third threshold.

In some embodiments, the average value h of the height information is greater than L _H and less than U _H , and the ratio r of heights higher than the third threshold is greater than L _r and less than U _r . is also smaller, a fifth determining unit 1103 is used to determine that the pose of the human body is the pose corresponding to the L _H and U _H and L _r and U _r .

FIG. 12 is a diagram showing another example of the detection unit 803. As shown in FIG. In some embodiments, the detection unit 803 includes a sixth determination unit 1201 and a seventh determination unit 1208, as shown in FIG. A sixth determining unit 1201 determines a human body height distribution P _H based on the human body height information, and a seventh determining unit 1202 based on the human body height distribution {P _j } corresponding to different preset postures. Used to determine the posture of the human body.

In some embodiments, determining the pose of the human body by the seventh determining unit 1208 based on the preset human body height distribution {P _j } corresponding to different poses includes combining the human body height distribution _PH and each calculating the distance between the body height distributions {P _j } corresponding to the poses, and determining that the body pose is the pose corresponding to the minimum said distance.

Although each component or module related to the present invention has been described above, the present invention is not limited to this. In addition, the human body posture detection apparatus 800 may further include other components or modules, and the related art can be referred to for specific contents of these components or modules.

The human body posture detection apparatus according to the embodiment of the present invention can perform posture detection using signal fluctuations caused by the movement of the human body without using a point cloud. For example, it is suitable for detecting postures such as standing, sitting on a stool, sitting on the ground, and lying down.In addition, the algorithm has good generalization performance, low complexity, and detection accuracy. is high.

<Third aspect of the embodiment>
An embodiment of the present invention provides a data processing apparatus. The data processing device may be, for example, a computer, a server, a workstation, a desktop computer, a smart phone, etc., but embodiments of the present invention are not limited thereto.

FIG. 13 is a diagram showing a data processing device in an embodiment of the present invention. As shown in FIG. 13, a data processing device 1300 includes at least one interface (not shown), a processor (e.g., a central processor ( CPU)) 1301 and a memory 1302 , the memory 1302 being connected to the processor 1301 . Among them, the storage device 1302 can store various data, can also store a human body posture detection program 1303, and executes the program 1303 under the control of the processor 1301 to store various data, such as Various preset thresholds, conditions, etc. can be stored.

In some embodiments, the functions of the human body posture detection device 800 described in the second aspect of the embodiment are integrated in the processor 1301 to realize the human body posture detection method described in the first aspect of the embodiment. be able to. For example, the processor 1301 may be configured as follows:
Determining the position of the human body based on the variation of the radar angle FFT data;
performing statistics on the height information of the human body at the position where the human body is located based on the difference of the radar angle FFT data; and detecting the posture of the human body based on the height information of the human body.

In some embodiments, the human body pose detection device 800 described in the second aspect of the embodiment may be arranged separately from the processor 1301 . For example, the human body posture detection device 800 can be configured as a chip connected to the processor 1301, and the functions of the human body posture detection device 800 can be realized under the control of the processor 1301. FIG.

It should be noted that data processor 1300 may further include display 1305 and I/O device 1304, or need not include all the components shown in FIG. It may further include to obtain the input image frame. In addition, the data processing device 1300 may further include components not shown in FIG. 13, for which reference can be made to prior art.

In embodiments of the present invention, processor 1301, which may be referred to as a controller or operational controller and may include a microprocessor or other processing and/or logic device, receives input The operation of each component of the data processing device 1300 can be controlled by using the

In embodiments of the invention, storage 1302 may be, for example, one or more of a buffer, flash memory, HDD, removable media, volatile storage, non-volatile storage, or other suitable device. information and information processing programs. By executing the program stored in the storage device 1302, the processor 1301 can implement information storage, processing, and the like. Since the functions of other parts are similar to those of the conventional one, detailed description thereof is omitted here. Each component of data processing apparatus 1300 may be implemented in dedicated hardware, firmware, software, or a combination thereof, all within the scope of the present invention.

The data processing device in the embodiment of the present invention can perform pose detection using signal fluctuations caused by the motion of the human body without using a point cloud. , standing, sitting on a stool, sitting on the ground, lying down, etc. The algorithm has good generalization performance, low complexity, and high detection accuracy. expensive.

An embodiment of the present invention further provides a computer-readable program, wherein, when the program is implemented in a data processing device, the program causes the data processing device to: Run the detection method.

An embodiment of the present invention further provides a storage medium storing a computer-readable program, wherein the computer-readable program causes a data processing device to execute the human body posture detection method according to the first aspect of the embodiment. used for

The above apparatus and method of the present invention may be implemented by hardware, or by a combination of hardware and software. The invention further relates to a computer readable program which, when executed by a logic component, causes the logic component to implement the device or component described above, or which causes the logic component to: Various methods or steps described above may be implemented. The present invention further relates to storage media capable of storing such programs, such as hard disks, magnetic disks, optical disks, DVDs, flash memories, and the like.

The object of the present invention may also be realized in the following manner: directly or indirectly providing a storage medium storing the above-described executable program code to a system or apparatus; Or a computer or central processing unit (CPU) in the device reads and executes the above program code. At this time, as long as the system or device has a function of executing a program, the implementation method of the present invention is not limited to the program, and the program can be executed in any form, such as an object-oriented program or an interpreter. It may be a program provided by the operating system, a script program provided by the operating system, or the like.

These machine-readable storage media may include, but are not limited to, various memory and storage units, semiconductor devices, magnetic disks such as optical, magnetic, magneto-optical disks, and other media suitable for storing information. do not have.

In addition, the computer connects to the corresponding website on the Internet, downloads and installs the computer program code according to the present invention into the computer, and then executes the program to implement the technical solution of the present invention. can also

Additionally, the present invention further provides a program product comprising machine-readable instruction code. Such instruction code, when read and executed by a machine, is capable of performing the methods in the embodiments of the invention described above. Correspondingly, to carry such program products, e.g. magnetic disks (including floppy disks), optical disks (including CD-ROMs and DVDs), magneto-optical disks (MD ), and various storage media such as semiconductor memory devices are also included in the present invention.

The above storage medium may include, for example, a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory device, etc., but is not limited to these.

Each operation (process) in the above-described method can also be implemented in the form of a computer-executable program stored in various machine-readable storage media.

In addition, the above examples and the like are further disclosed as supplementary notes as follows.

(Appendix 1)
A human body posture detection method comprising:
Determining the position of the human body based on the variation of the radar angle FFT data;
A method, comprising: making statistics on the height information of the human body at the position where the human body is located based on the difference of the radar angle FFT data; and detecting the posture of the human body based on the height information of the human body.

(Appendix 2)
The method of Appendix 1, wherein
Determining the location of the human body based on the variation of the radar angle FFT data is
calculating the variation of the angular FFT data at the distance frequency points and the horizontal angular frequency points;
searching for a local maximum of variation of the angular FFT data at the distance frequency points and the horizontal angle frequency points; and determining, based on the local maximum, whether there is a human body at the location of the local maximum. A method comprising:

(Appendix 3)
The method of Appendix 2,
Computing the variation of the angular FFT data at the distance frequency points and the horizontal angular frequency points is
calculating the variation of the angle FFT data at all vertical angle frequency points corresponding to each distance frequency point and the horizontal angle frequency point; and calculating the maximum or average variation of the angle FFT data at all said vertical angle frequency points. as variation of angular FFT data of said distance frequency points and horizontal angular frequency points.

(Appendix 4)
The method of Appendix 3, wherein
Computing the variation of the angular FFT data at each vertical angular frequency point corresponding to each distance frequency point and horizontal angular frequency point
calculating the average difference of the angle FFT data within a predetermined period of the distance frequency point, the horizontal angle frequency point and the vertical angle frequency point; and calculating the average difference of the angle FFT data for the distance frequency point and the horizontal angle frequency. A method comprising as variation of angular FFT data for points and vertical angular frequency points.

(Appendix 4.1)
The method of Appendix 4,
A difference in angle FFT data within a predetermined period of the distance frequency point, the horizontal angle frequency point and the vertical angle frequency point is
is the absolute value of the difference of the angle FFT amplitudes at different times within the predetermined time period of the distance frequency points, the horizontal angle frequency points and the vertical angle frequency points.

(Appendix 4.2)
The method of Appendix 4,
A difference in angle FFT data within a predetermined period of the distance frequency point, the horizontal angle frequency point and the vertical angle frequency point is
is the modulus of the angular FFT complex difference at different times within the predetermined time period of the distance frequency points, the horizontal angular frequency points and the vertical angular frequency points.

(Appendix 5)
The method of Appendix 3, wherein
Computing the variation of the angular FFT data at each vertical angular frequency point corresponding to each distance frequency point and horizontal angular frequency point
calculating the standard deviation of angle FFT amplitudes within a predetermined time period of said distance frequency points, horizontal angle frequency points and vertical angle frequency points; and calculating the standard deviation of said angle FFT amplitudes of said distance frequency points, horizontal angle frequency points and vertical angle. A method comprising taking angular FFT data as a variation of frequency points.

(Appendix 5.1)
The method of Appendix 5,
Angle FFT amplitudes within a predetermined period of the distance frequency points, the horizontal angle frequency points and the vertical angle frequency points are:
is the amplitude of the angle FFT at different times within a predetermined time period of the distance frequency points, the horizontal angle frequency points and the vertical angle frequency points.

(Appendix 6)
The method of Appendix 2,
The method, wherein the local maximum is the maximum variation of the angular FFT data at the distance frequency points and horizontal angle frequency points within a preset range.

(Appendix 7)
The method of Appendix 2,
When the change value of the angle FFT data of the distance frequency point and the horizontal angle frequency point at the position where the local maximum is located is larger than a preset threshold (first threshold), the position where the local maximum is located A method of determining that there is a human body, otherwise determining that there is no human body at the location of the local maximum.

(Appendix 8)
The method of Appendix 1, wherein
Performing statistics on the height information of the human body based on the difference in the amplitude of the radar angle FFT is
determining a range of the human body based on the position of the human body;
calculating a difference in angle FFT data at all distance, horizontal and vertical angle frequency points within the range for each time in a predetermined time period; If it is larger than the set threshold (second threshold), the height information of the vertical angle frequency point corresponding to the maximum value is used as the height information of the human body at the current time, so that the height of the human body within the predetermined period method, including obtaining information about

(Appendix 9)
The method of Appendix 8, wherein
If the position of the human body is (r _h , _ah ), r _h is the distance frequency point, and _ah is the horizontal angle frequency point, then the range of the human body is the distance frequency point r _h − L is greater than _r and less than r _h +U _r , the horizontal angle frequency point is greater than a _h -L _a and less than a _h +U _a , and the vertical angle frequency point is greater than L _e and determined to be a region smaller than U _e ,
L _r and U _r are preset distance frequency point thresholds, L _a and U _a are preset horizontal angular frequency point thresholds, and L _e and U _e are preset vertical angular frequency point thresholds, Method.

(Appendix 10)
The method of Appendix 8, wherein
The difference between the angle FFT data of each distance frequency point, horizontal angle frequency point and vertical angle frequency point at the current time is
the absolute value of the difference in angle FFT amplitude between the current time and the previous time of said distance frequency point, horizontal angle frequency point and vertical angle frequency point; or said distance frequency point, horizontal angle frequency point and vertical angle frequency point method, which is the modulus of the angular FFT complex difference between the current time and the previous time in .

(Appendix 11)
The method of Appendix 1, wherein
Detecting the posture of the human body based on the height information of the human body includes:
calculating an average value (h) of height information of the human body;
calculating a ratio (r) of heights higher than a preset threshold (third threshold) among the height information of the human body; and calculating an average height value corresponding to each preset posture. Thresholds (L _H and U _H ) and ratio thresholds (L _r and U _r ) whose height is higher than the third threshold, and the average value (h) of the height information of the human body and the height information of the human body determining the pose of the human body based on the ratio (r) of which height is higher than a third threshold.

(Appendix 11.1)
11. The method of Supplementary Note 11,
When the average value h of height information is larger than L _H and smaller than U _H , and the ratio r of heights higher than the third threshold is larger than L _r and smaller than U _r , the posture of the human body is A method of determining which poses correspond to L _H and U _H and L _r and U _r .

(Appendix 12)
The method of Appendix 1, wherein
Detecting the posture of the human body based on the height information of the human body includes:
determining a human body height distribution P _H based on the human body height information; and determining a human body posture based on a human body height distribution {P _j } corresponding to preset different postures. .

(Appendix 12.1)
12. The method of Appendix 12, wherein
Determining the pose of the human body based on the human body height distribution {P _j } corresponding to different preset poses is
calculating the distance between the human body height distribution P _H and the human body height distribution {P _j } corresponding to each pose; and determining that the human body pose is the pose corresponding to the minimum distance. A method, including

(Appendix 13)
A data processing device,
including a processor and a memory;
The storage stores a computer program, and the processor is configured to implement the following human body posture detection method by executing the computer program:
Determining the position of the human body based on the variation of the radar angle FFT data;
A method comprising: performing statistics on the height information of the human body at the position where the human body is located based on the difference of the radar angle FFT data; and detecting the posture of the human body based on the height information of the human body. , data processing equipment.

Although the preferred embodiment of the present invention has been described above, the present invention is not limited to this embodiment, and all modifications to the present invention fall within the technical scope of the present invention as long as they do not depart from the gist of the present invention.

Claims (10)

A human body posture detection device,
A first determination unit for determining the position of the human body based on the variation of the radar angle FFT data;
a statistical unit for performing statistics on the height information of the human body at the position where the human body is located based on the difference of the angle FFT data of the radar; and a detection unit for detecting the posture of the human body based on the height information of the human body. ,Device. A device according to claim 1, wherein
The first determination unit is
a first calculation unit for calculating the variation of the angular FFT data at range frequency points and horizontal angular frequency points;
a search unit for searching a local maximum of variation of the angular FFT data at the distance frequency points and the horizontal angle frequency points; and based on the local maximum whether there is a human body at the location of the local maximum including a second determination unit that determines the
The apparatus, wherein the local maximum is the maximum variation of the angular FFT data at the distance frequency points and horizontal angular frequency points within a preset range. A device according to claim 2, wherein
The first computing unit computing variations in angular FFT data at range frequency points and horizontal angular frequency points comprising:
calculating the variation of the angle FFT data at all vertical angle frequency points corresponding to each distance frequency point and the horizontal angle frequency point; and calculating the maximum or average variation of the angle FFT data at all said vertical angle frequency points. as variation of angular FFT data of said distance frequency points and horizontal angular frequency points. A device according to claim 2, wherein
If the variation value of the angle FFT data of the distance frequency points and the horizontal angle frequency points at the position where the local maximum is located is greater than a first threshold, the second determination unit determines that the local maximum is located. A device, determining that there is a human body at a position, otherwise said second determining unit determines that there is no human body at the position where said local maximum is located. A device according to claim 1, wherein
The statistical unit is
a third determination unit for determining the range of the human body based on the position of the human body;
a second computation unit for calculating, for each time instant of a predetermined time period, the difference in angle FFT data at all distance frequency points, horizontal angle frequency points, and vertical angle frequency points within said range; and a maximum of said angle FFT data differences. If the value is greater than the second threshold, the height information of the vertical angle frequency point corresponding to the maximum value is used as the height information of the human body at the current time, so that the height information of the human body within the predetermined period is obtained. an apparatus comprising a obtaining fourth determination unit. A device according to claim 5, wherein
When the position where the human body is located is (r _h , _ah ), where r _h is the distance frequency point and _ah is the horizontal angle frequency point, the third determination unit determines the range where the human body is located as follows: The distance frequency point is greater than r _h -L _r and less than r _h +U _r , the horizontal angle frequency point is greater than a _h -L _a and less than a _h +U _a , and the vertical angle frequency Determine a region where the point is larger than L _e and smaller than U _e ,
where L _r and U _r are preset distance frequency point thresholds, L _a and U _a are preset horizontal angular frequency point thresholds, and L _e and U _e are preset vertical angular frequency point thresholds. is the device. A device according to claim 5, wherein
The difference between the angle FFT data of each distance frequency point, horizontal angle frequency point and vertical angle frequency point at the current time is
is the absolute value of the difference in angle FFT amplitude between the current time and the previous time of said distance frequency point, horizontal angle frequency point and vertical angle frequency point; or said distance frequency point, horizontal angle frequency point and vertical angle A device that is the modulus of the angular FFT complex difference between the current time and the previous time of a frequency point. A device according to claim 1, wherein
The detection unit is
a third calculation unit for calculating an average value h of height information of the human body;
a fourth calculation unit for calculating a ratio _r of the number _of height information of the human body whose height is higher than a third threshold; Thresholds L _r and U _r for the ratio of the number of numbers whose height is higher than the third threshold, and the average value h of the height information of the human body and the number of the height information of the human body whose height is higher than the third threshold a fifth determination unit that determines the posture of the human body based on the ratio r of
If the average value h of the height information is greater than L _H and less than U _H and the ratio r of the number of heights higher than the third threshold is greater than L _r and less than U _r , the first 5. The apparatus, wherein a determination unit determines that the posture of the human body is a posture corresponding to the L _H and U _H and L _r and U _r . A device according to claim 1, wherein
The detection unit is
a sixth determining unit for determining a human body height distribution _PH based on the height information of _the human body; An apparatus containing seven determination units. A device according to claim 9, wherein
Determining the posture of the human body by the seventh determination unit based on the human body height distribution {P _j } corresponding to preset different postures,
Calculate the distance between the human body height distribution P _H and the human body height distribution {P _j } corresponding to each pose; and determine that the human body pose is the pose corresponding to the minimum distance. equipment, including

Images