Correlation detection method for human body movement by utilizing Doppler characteristics of human body movement
By setting the correlation calculation of threshold values and continuous frequency modulation signals, the noise and human motion signals are distinguished, and the problems of misjudgment and high cost in the prior art are solved, thereby achieving more efficient and accurate human motion detection.
Patent Information
- Application Number
- CN202410894508.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-04
- Publication Date
- 2024-11-22
AI Technical Summary
When detecting human movements, it is difficult to effectively distinguish between noise and human movement signals, especially in outdoor environments, misjudgment is prone to occur, and algorithm resources are high and costly.
By setting a preset threshold D, the correlation calculation is performed using continuous frequency modulation signals to distinguish noise from human motion signals, and the frequency concentration characteristics of human motion signals are used to reduce algorithm complexity and improve identification efficiency.
The effective difference between noise and human motion signals is achieved, the misjudgment rate is reduced, the algorithm resource requirements are simplified, the cost is saved, and the detection accuracy and real-timeness are improved.
Smart Images

Figure CN119001706A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of human body movement detection, and in particular to a human body movement related detection method using human body movement Doppler characteristics. Background Art
[0002] Energy conservation and emission reduction is the theme of today's social development, and technical means such as smart lighting, smart air conditioning, and smart access control have become an effective way and important means to promote energy conservation and emission reduction and mitigate climate change.
[0003] Currently, many lighting lamps, ultraviolet lamps, and access control systems on the market are still manually controlled and are non-automated. Manual control will lead to a poor user experience. If the lamps and access control systems are not turned off in time due to human negligence, it will not only cause energy loss, but disinfection lamps such as ultraviolet lamps will also cause harm to human health.
[0004] At present, the detection of human motion in the market mainly relies on infrared sensors, cameras, and continuous wave radar sensors. Infrared sensors use lens arrays to divide the environment into light and dark areas. When a person moves between these light and dark areas, the infrared radiation from the human body will stimulate voltage pulses on the sensor, and the sensor can amplify and identify the pulses to achieve the perception of the moving human body. The principle of camera perception is to recognize people and their movements through image recognition technology. The advantages of infrared are low power consumption and the use of human infrared characteristics. The disadvantages are that the detection principle based on infrared is easily affected by changes in ambient temperature, and the lens faces problems such as decreased sensitivity due to dust coverage when used for a long time. The advantage of the camera is that it can obtain good detailed features of human motion, but the disadvantages are insufficient privacy protection, high power consumption, and high cost.
[0005] Radar was mainly used in the military field in the early days to detect aircraft. With the development of electronic technology and information processing technology, the application field of radar has been greatly expanded, from the military field to the civilian field, from detecting high-speed moving targets such as aircraft to low, small and slow targets such as pedestrians, and its application is becoming more and more extensive. Radar can be divided into pulse radar and continuous wave (CW) radar according to the transmission waveform. Among them, pulse radar transmits pulse-modulated high-frequency signals and calculates the target distance by measuring the round-trip delay of the pulse. It is generally used for long-distance motion detection and is not suitable for short-distance ranging. Continuous wave radar continuously transmits high-frequency signals. According to whether the continuous wave signal is modulated, continuous wave radar is divided into FMCW radar, PMCW radar, and single-tone Doppler continuous wave radar. It is generally used for close-range motion detection. Radar has the advantages of easy operation, low hardware cost, no weather and light constraints, and wall penetration. Continuous wave radar can be used for active human motion detection and is suitable for a variety of complex scenes.
[0006] The hardware structure of the single-tone Doppler continuous wave radar used for motion detection includes an oscillator, an amplifier, a downconverter, an antenna, and an MCU. The oscillator generates a single-tone RF continuous wave signal, which is amplified by the amplifier and radiated to the irradiation area through the antenna. The reflected signal of the moving target in the area is received by the antenna and then outputs an intermediate frequency signal through the downconverter. The MCU samples and detects the intermediate frequency output signal of the mixer, and finally determines whether there is a moving target in the radar irradiation area based on a specific algorithm. According to the Doppler principle, a moving object will change the frequency of the reflected electromagnetic wave signal, so the mixer will output a signal with a frequency corresponding to the moving speed.
[0007] Since radar has a wide range of application scenarios and a huge shipment volume, it is extremely sensitive to cost. The algorithm's signal processing is often implemented through an extremely cheap MCU, so computing resources are extremely limited, which also requires the algorithm to be sufficiently concise.
[0008] Radar sensors use the Doppler principle of electromagnetic wave propagation, but the output signal contains two parts: noise and useful signal. The noise is divided into device noise and environmental noise. The device output noise is only related to the working conditions and ambient temperature of the device itself and has nothing to do with the moving object. It fluctuates irregularly with a certain amplitude and its spectrum is relatively dispersed. When the radar sensor is placed in an indoor environment with no human motion, the noise floor of the intermediate frequency signal will increase relative to the device noise, and occasionally a spike will appear. The frequency of the signal is also irregular. This is because there are some environmental interferences in the indoor environment, such as electromagnetic wave signals radiated by other devices and irregular small-amplitude object movements in the environment. When the module is placed in an outdoor environment with no human motion, the peak of the intermediate frequency signal noise floor is relatively sharp. When the wind blows, the intermediate frequency signal fluctuates greatly. This is because there are many large-amplitude irregular movements in the environment, such as the irregular shaking of flowers and plants blown by the wind. When there is human motion in the radar irradiation area, whether indoors or outdoors, the intermediate frequency signal is similar to a sine wave, and has time domain and frequency domain characteristics that are significantly different from the noise signal. By comparing and analyzing the noise signal and the human motion signal, it can be found that whether it is the background noise in the dark room or the background noise when no one is moving indoors or outdoors, their common point is that they are all signals similar to white noise, and the only difference is the amplitude size. The human motion signal is very regular, and the signal power is concentrated at a certain frequency or several frequencies from the frequency domain.
[0009] At present, the algorithm for distinguishing noise and human motion through intermediate frequency signals is the peak discrimination method: select a suitable sampling frequency, collect several points, then find the maximum and minimum values of these points, calculate the difference between the maximum and minimum values, and count this difference as the threshold. The method diagram is shown in the figure below: Figure 1 As shown, Figure 1Take 60 sampling points from it, find the maximum and minimum values of these points, calculate the difference between the maximum and minimum values, and record this difference as the threshold; if the fluctuation amplitude of the intermediate frequency signal exceeds this threshold and the number of exceedances is greater than the set number within a specified number of time windows, we consider that a moving object has appeared in the radar illumination area.
[0010] However, it is difficult to distinguish noise and human movement by the peak discrimination method. In an indoor environment, the effect is relatively good, but when applied outdoors, the effect is not good. The amplitude of random noise increases outdoors, especially the vibrations caused by natural wind blowing and moving objects, such as the shaking of leaves and the fluttering of flags, which are more likely to cause misjudgments. The time domain and frequency domain of the noise signal are as Figure 2 shown, and the time domain and frequency domain of the human movement signal are as Figure 3 shown. The peak discrimination method does not utilize the difference that the noise signal spectrum is dispersed and the human movement frequency is concentrated; the spectrum analysis method can distinguish the noise with a dispersed spectrum and the human movement signal with a concentrated spectrum, but the spectrum analysis method will use FFT, which has a relatively high requirement for computing power, and higher computing power requires higher costs. Summary of the Invention
[0011] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a human movement related detection method using the Doppler characteristics of human movement.
[0012] The purpose of the present invention is achieved through the following technical solutions: The present invention provides a human movement related detection method using the Doppler characteristics of human movement, including the following steps: Step 1: Place the device with a continuous wave radar in the measured environment, detect the environment, set a preset threshold D based on the values measured by the radar, and the signal collected by the device with the continuous wave radar is sampled by ADC to obtain the input signal x(t); Step 2: Set a correlation signal h(t), and the correlation signal h(t) is a continuously frequency modulated signal. The formula for the correlation signal h(t) is: ; where f represents frequency, pi represents pulse, and t represents signal period; Step 3: Calculate the correlation between the input signal x(t) and the correlation signal h(t) to obtain a correlation result S. Compare the threshold D in step S1 with the correlation result S. If S > D, it is determined that the input signal is a human movement signal; if S < D, it is determined that the input signal is a noise signal; the calculation method for the correlation between the input signal x(t) and the correlation signal h(t) is: ; Step 4: After the system determines that it is a human movement signal, trigger corresponding actions. If it is determined to be a noise signal, do not trigger corresponding actions.
[0013] Furthermore, in step 1, if the detection distance of the continuous wave radar device is longer, the preset threshold value D is smaller.
[0014] Furthermore, in step 2, when setting specific parameters of the signal h(t), the frequency component of the signal h(t) includes the frequency band range of the Doppler frequency shift corresponding to the moving person.
[0015] Furthermore, the value of the frequency f in step 2 needs to be determined according to the signal bandwidth that the user wants to detect.
[0016] Furthermore, in step 3, when the output signal x(t) is a noise signal and is correlated with h(t), the frequencies cancel each other out, and the output result S after correlation is small; when the output signal x(t) is a human motion signal, a signal with frequencies concentrated at several frequency points, when the human motion signal is correlated with h(t), there is no cancellation of other frequencies, and the output result S after correlation is large.
[0017] Furthermore, the corresponding actions include lighting, opening doors, taking photos with cameras, or waking up with low power consumption. The beneficial effects of the present invention are: 1) The present invention makes full use of the difference between noise and human motion signals in the frequency domain to achieve more effective recognition; 2) The algorithm used in the present invention does not require high resources, is simple to operate, and saves costs. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] Figure 1 It is a schematic diagram of a peak value discrimination method in the prior art; Figure 2 It is the spectrum diagram of the noise signal, where (a) is the frequency domain diagram and (b) is the time domain diagram; Figure 3 The spectrum diagram of human motion signal, where (a) is the frequency domain diagram and (b) is the time domain diagram; Figure 4 A flow chart of a method for detecting human body movement using Doppler characteristics of human body movement according to an embodiment of the present invention; Figure 5 The spectrum diagram of the human motion signal of the system with a sampling frequency of 250 Hz according to the embodiment of the present invention; Figure 6 This is a spectrum diagram of a noise signal of a system with a sampling frequency of 250 Hz according to an embodiment of the present invention; Figure 7 Spectrum diagram of the relevant signal h(t) of the system with a sampling frequency of 250 Hz according to an embodiment of the present invention; Figure 8A process block diagram of a formula of a method for detecting human movement related to human movement using Doppler characteristics of human movement according to an embodiment of the present invention; Fig. 9 This is a spectrum diagram of the output results of a simulated human motion signal and a noise signal using 5 Hz superimposed with 20% noise in an embodiment of the present invention; Fig.10 This is a spectrum diagram of the output results of a simulated human motion signal and a noise signal using 10 Hz superimposed with 20% noise in an embodiment of the present invention; Fig.11 This is a spectrum diagram of the output results of a simulated human motion signal and a noise signal using 15 Hz superimposed with 20% noise in an embodiment of the present invention; Fig.12 This is a spectrum diagram of the output results of a simulated human motion signal and a noise signal using 20 Hz superimposed with 20% noise in an embodiment of the present invention; Fig.13 The embodiment of the present invention is applied to a single chip microcomputer to test the spectrum diagram of outdoor without human motion; Fig.14 This is a spectrum diagram of an embodiment of the present invention applied to a single chip microcomputer to test outdoor human body motion. DETAILED DESCRIPTION
[0019] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.
[0020] For example, the specific flow diagram of the present invention is as follows Figure 4As shown, the input signal x(t) is obtained by continuous wave radar detection and then ADC sampling. The process of continuous wave radar detection and then ADC sampling to obtain the input signal x(t) belongs to the prior art and will not be described in detail here. The device with continuous wave radar is placed in the tested environment. The value tested by the radar sets a preset threshold D. The preset threshold D needs to be determined according to the user's detection distance. The longer the detection distance, the smaller the preset threshold should be set. The preset threshold D is a preset range according to the working environment and is not limited to a specific range. For example, the preset threshold D of human motion when the detection distance is 10m must be smaller than the preset threshold D of human motion when the detection distance is 5m. If there is human motion in the environment, the collected signal is recorded as a human motion signal. If there is no human motion in the environment, the collected signal is recorded as a noise signal. The experimenter sets the input signal x(t) as a human motion signal or a noise signal. For the system, it is unknown whether the input signal is a human motion signal or a noise signal; when the frequencies cancel each other out, the output result S is low and less than the preset threshold D, which means that the input signal x(t) is a noise signal; when the frequencies are concentrated at a few frequency points, after the input signal x(t) and h(t) are correlated, since there is no cancellation of other frequencies, the correlated output result S is large and greater than the preset threshold D, which means that the input signal x(t) is a human motion signal. After that, the experimenter compares the system's judgment result with the previously input signal to know whether the system's judgment is accurate. In actual applications, if the system judges that it is a human motion signal, it can trigger corresponding behaviors, such as turning on lights, opening doors, taking photos with cameras, low-power wake-up, etc. If it is judged to be a noise signal, it will not be triggered. The related signal h(t) is a continuously frequency modulated signal, and the formula for the related signal h(t) is: ; Where f represents frequency, the value of frequency f needs to be determined according to the signal bandwidth to be detected by the user, pi represents pulse, and t represents signal period; the correlation calculation is performed on the input signal x(t) and the correlation signal h(t), and the process block diagram is as follows Figure 8 As shown, noise(t) is a noise signal, signal(t) is a human motion signal, and the input signal x(t) is judged to be a noise signal or a human motion signal according to the correlation calculation result; the calculation method for the correlation between the input signal x(t) and the correlation signal h(t) is: ; The frequency components of the signal h(t) include the frequency band range of the Doppler frequency shift corresponding to a moving person.
[0021] Exemplarily, in order to verify the effect of the present invention on distinguishing noise signals from human motion signals, simulink was used for simulation. The simulation results show that the present invention can make the output value of a single frequency signal representing human motion correlated with h(t) increase rapidly; while the output value of a noise signal with a dispersed spectrum correlated with h(t) will not fluctuate much. A system with a sampling frequency of 250 Hz is used; the human motion signal is simulated by superimposing a sine wave with noise; h(t) is a sine wave with an instantaneous frequency that decreases gradually from 45 Hz to 5 Hz; the spectrum waveform of the human motion signal is as follows Figure 5 As shown, the spectrum waveform of the noise signal is as follows Figure 6 As shown, the spectrum waveform of the related signal h(t) is as follows Figure 7 shown.
[0022] By changing the frequency of the human motion signal, 5 Hz, 10 Hz, 15 Hz and 20 Hz are superimposed with 20% white noise to simulate the intermediate frequency signal output by the radar device when human motion is detected. By selecting different noises and comparing the output results S, it can be found that the output value S of the human motion signal is always much larger than the output value S of the noise signal. Figure 9-12 As shown, the thick line represents the output of the simulated human motion signal with 20% noise superimposed, and the thin line represents the output of the random noise signal.
[0023] For example, the present invention is applied to a single-chip microcomputer. The method of writing into the single-chip microcomputer is a prior art and will not be described in detail here. Then, a test is conducted outdoors. The spectrum diagram of the outdoor environment without human motion is as follows: Fig.13 As shown, the spectrum diagram of outdoor human movement is as follows Fig.14 As shown. The black line represents a judgment cycle, the circled line represents the intermediate frequency signal, and the star line represents the output value S of the intermediate frequency signal after passing through the system. In the case of human motion, the star line will still have a large value at the end of a judgment cycle, but the noise signal will have a small value after a judgment cycle. This is because the frequency of the noise signal is dispersed, and after an integration cycle, its frequencies cancel each other out, while the human motion signal with concentrated frequencies will not have the situation where the frequencies cancel each other out; it can be seen that this method significantly improves the distinction between noise and human motion. Applying this method to a single-tone Doppler continuous wave detection radar can remove the influence of white noise, and can remove the influence of the swing of flowers, plants and trees outdoors, greatly reducing the probability of false alarms.
[0024] Exemplarily, to distinguish between frequency-dispersed signals and frequency-concentrated signals, the spectrum analysis method can also be used, that is, first collect 2 to the power of N points, generally 512 points, and then perform FFT after collection, and directly analyze whether the frequency of the signal is concentrated or dispersed through the spectrum diagram. Compared with the present application, this method has the following two disadvantages: First, the computing power requirement is high. Since the spectrum analysis method requires the chip to be able to do FFT, this requires the chip to have a DSP module, which increases the chip cost, while the related detection algorithm only requires the chip to be able to do simple addition, subtraction, multiplication and division, the calculation is simple, and the cost performance is higher. Second, there is a delay. Since the spectrum analysis method needs to be sampled first, it is necessary to wait until a certain number of points are collected before FFT can be performed for discrimination, which causes a delay. For example, if the sampling frequency is 250hz (empirical value), it takes 4ms to sample one point, and it takes 2048ms to collect 512 points, which means that it takes 2s from a person's exercise to display the movement of the human body. The delay of 2s is enough to make people feel that the system is not timely enough. However, the present application processes the sampled data in real time, and processes the data immediately after it is collected, and the result can be given after one cycle. In summary, the spectrum analysis method has high hardware requirements, high cost, and delay problems. Compared with the spectrum analysis method, the present application has the advantages of low cost and high reliability.
[0025] The above is only a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art shall not deviate from the spirit and scope of the present invention, and shall be within the scope of protection of the claims attached to the present invention.
Claims
1. A method for detecting human body movement using Doppler characteristics of human body movement, characterized in that: The following steps are involved: Step 1: Place the device with continuous wave radar in the environment to be tested, detect the environment, set a preset threshold D based on the value tested by the radar, and sample the signal collected by the device with continuous wave radar through ADC to obtain the input signal x(t); Step 2: Set a correlation signal h(t). The correlation signal h(t) is a continuously frequency modulated signal. The formula of the correlation signal h(t) is: ; Where f represents frequency, pi represents pulse, and t represents signal period; Step 3: Calculate the correlation between the input signal x(t) and the correlation signal h(t) to obtain a correlation result S. Compare the threshold D in step S1 with the correlation result S. If S > D, it is determined that the input signal is a human motion signal; if S < D, it is determined that the input signal is a noise signal. The calculation method for the correlation between the input signal x(t) and the correlation signal h(t) is as follows: ; Step 4: If the system determines that the signal is a human motion signal, the corresponding behavior is triggered. If the system determines that the signal is a noise signal, the corresponding behavior is not triggered.
2. The method for detecting human body movement using Doppler characteristics of human body movement according to claim 1, characterized in that: In step 1, if the detection distance of the continuous wave radar device is longer, the preset threshold D is smaller.
3. The method for detecting human body movement using Doppler characteristics of human body movement according to claim 1, characterized in that: In step 2, when setting the specific parameters of the signal h(t), the frequency component of the signal h(t) includes the frequency band range of the Doppler frequency shift corresponding to the moving person.
4. The method for detecting human body movement using Doppler characteristics of human body movement according to claim 1, characterized in that: The value of the frequency f in step 2 needs to be determined according to the signal bandwidth that the user wants to detect.
5. The method for detecting human body movement using Doppler characteristics of human body movement according to claim 1, characterized in that: In step 3, when the output signal x(t) is a noise signal and is correlated with h(t), the frequencies cancel each other out, and the output result S after correlation is small; when the output signal x(t) is a human motion signal, a signal with frequencies concentrated at several frequency points, when the human motion signal is correlated with h(t), there is no cancellation of other frequencies, and the output result S after correlation is large.
6. The method for detecting human body movement using Doppler characteristics of human body movement according to claim 1, characterized in that: The corresponding actions include lighting up a light, opening a door, taking a photo with a camera, or waking up from low power consumption.
