Person detection system and method based on ultrasonic waves

Abstract

A method of personnel detection based on ultrasound. The method includes the steps of: (a) transmitting an ultrasound signal from a transmitter, the ultrasound signal including components at a first frequency; (b) receiving a reflection of the ultrasound signal, the received signal including components at frequencies higher and lower than the first frequency; (c) determining a difference between an upper portion of the received signal containing frequencies higher than the first frequency and a lower portion of the received signal containing frequencies lower than the first frequency; and (d) determining whether a person is present based on the difference between the upper portion and the lower portion.

Description

Technical Field

[0001] The present invention relates to an ultrasonic personnel detection system and method. Background Art

[0002] Video conferencing devices such as collaboration endpoints often have a standby mode, meaning they remain active when not in use. Beneficially, the standby mode reduces power consumption, for example by shutting down unneeded components (e.g., monitors). Another benefit of the standby mode is that it increases the life expectancy of components such as camera sensors and projector lamps.

[0003] Devices with a standby state must have a method for putting the device into standby mode and waking it up again. For example, previous video conferencing equipment operated on a timer and human input method. The device was programmed to enter standby mode after a predetermined period of inactivity (e.g., 10 minutes) and be woken up by human input into a control panel or similar device. However, this is not particularly intuitive for some users, especially if the control panel is not easily accessible or hidden from view.

[0004] It is also known to use a person detector, in which a device is placed in standby mode when it determines that no one is nearby. One such example is disclosed in US Pat. No. 9,319,633 B1, in which ultrasonic waves are transmitted into a spatial region. A controller then receives the reflected ultrasonic waves and calculates an error signal representing an estimate of the ultrasonic signal received without an echo. The controller then calculates a power estimate of the error signal over time and detects the presence of a person based on changes in this power estimate.

[0005] However, a problem with the ultrasonic method discussed above is that it is prone to false alarms. The systems discussed above detect within a relatively narrow frequency range. However, transient noises not generated by the transmitter (such as a door slamming or hands clapping) are relatively broad in frequency and therefore will have components that fall within the detection range. This can lead to false alarms, where transient noises detected near a device (but not indicating the presence of a person in the room) can cause the device to wake up from standby mode.

[0006] Therefore, a more robust method for person detection that is less prone to false positives is needed. Summary of the Invention

[0007] Therefore, in a first aspect, an embodiment of the present invention provides a method for detecting people based on ultrasound, comprising the following steps:

[0008] (a) transmitting an ultrasonic signal from a transmitter, the ultrasonic signal including a component at a first frequency f0;

[0009] (b) receiving a reflection of the ultrasonic signal, the received signal including components having frequencies higher and lower than the first frequency;

[0010] (c) determining a difference between an upper portion of the received signal comprising frequencies higher than the first frequency and a lower portion of the received signal comprising frequencies lower than the first frequency; and

[0011] (d) determining whether a person is present based on the difference between the upper portion and the lower portion.

[0012] This approach is more robust in detecting the actual presence of a person by mitigating false positives caused by broadband noise that does not indicate movement. It is also more robust to other devices that may be emitting ultrasonic signals near f0. Furthermore, this approach does not rely on the transmitter and receiver being synchronized in time.

[0013] The method may have any one or any combination of the following optional features, as long as they are compatible.

[0014] The determination may be based on a distinction between the upper frequency portion and the lower frequency portion.

[0015] The upper portion of the received signal may include higher frequencies immediately adjacent to the first frequency, and the lower portion of the received signal may include lower frequencies immediately adjacent to the first frequency.

[0016] The method may include dividing the received signal into a plurality of intervals, each interval representing a frequency range of the received signal, and wherein the upper portion is an upper frequency interval containing a portion of the received signal having a frequency higher than the first frequency, and the lower portion is a lower frequency interval containing a portion of the received signal having a frequency lower than the first frequency.

[0017] The determination may be performed based on a difference between a normalized power estimate for the upper frequency interval and a normalized power estimate for the lower frequency interval. A normalization factor may be the sum of the power estimates for the upper frequency interval and the lower frequency interval. Normalization may be performed by subtracting an estimate of an average of the upper frequency interval and the lower frequency interval.

[0018] Determining the presence of a person may include determining a logit function of the normalized power of the upper frequency interval. Determining the presence of a person may include determining a logit function of the normalized power of the lower frequency interval. Determining the presence of a person may include determining a first logit function of the normalized power of the upper frequency interval and determining a second logit function of the normalized power of the lower frequency interval.

[0019] The logit function can take the following form:

[0020]

[0021] Here, X(t,f0+1) is the coefficient representing the upper frequency interval at time t, and X(t,f0-1) is the coefficient representing the lower frequency interval at time t.

[0022] Steps (b) to (d) may be repeated at a predetermined rate. Each repetition of steps (b) to (d) may be performed over a time window of the received signal. The time window may be approximately 20 milliseconds.

[0023] After determining the presence of a person, the method may further include determining whether the person is moving toward or away from the receiver. Determining whether the person is moving toward or away from the receiver may be further based on a first likelihood ratio test for determining whether the person is moving toward the receiver and a second likelihood ratio test for determining whether the person is moving away from the receiver. A log-likelihood ratio may be derived for each likelihood ratio and recursively calculated from previous values ​​of the corresponding log-likelihood ratio. Advantageously, such log-likelihood ratios are easier to implement and may have lower computational complexity, resulting in a computationally less expensive method.

[0024] When the presence of a person has been determined, the method may include taking the video conferencing device out of standby mode.

[0025] In some embodiments, there may be a first receiver and a second receiver, wherein the first receiver receives reflections of ultrasonic signals having components with frequencies greater than the first frequency, and the second receiver receives reflections of ultrasonic signals having components with frequencies less than the first frequency. Thus, the first receiver may receive an upper portion of the received signal, while the second receiver may receive a lower portion of the received signal. Determining the difference between the upper portion of the received signal and the lower portion of the received signal may include determining whether the first receiver and / or the second receiver received the corresponding signal. If both receivers receive the signal, it may be determined that the received signal is a false alarm. If only one receiver receives the signal, it may be determined that a person is present.

[0026] In a second aspect, an embodiment of the present invention provides a personnel detection system, the system comprising:

[0027] a transmitter configured to transmit an ultrasonic signal including a component at a first frequency f0;

[0028] one or more receivers configured to receive reflections of the ultrasonic signal; and

[0029] One or more processors configured, in response to the receiver receiving a receive signal including components at frequencies above and below the first frequency, to:

[0030] (a) determining a difference between an upper portion of the received signal comprising frequencies higher than the first frequency and a lower portion of the received signal comprising frequencies lower than the first frequency; and

[0031] (b) determining whether a person is present based on the difference between the upper portion and the lower portion.

[0032] This system is more robust in detecting whether a person is actually present by mitigating false positives caused by broadband noise that does not represent movement. It is also more robust to other devices that can emit ultrasonic signals near f0. Furthermore, it does not rely on the transmitter and receiver being time-synchronized.

[0033] The system can have any one or any combination of the following optional features, as compatible.

[0034] The determining can be based on a difference between the upper portion of the received signal and the lower portion of the received signal.

[0035] The upper portion of the received signal can comprise higher frequencies immediately adjacent to the first frequency, and the lower portion of the received signal can comprise lower frequencies immediately adjacent to the first frequency.

[0036] The processor can be further configured to divide the received signal into a plurality of bins, each bin representing a range of frequencies of the received signal, and wherein the upper portion is an upper frequency bin comprising a portion of the received signal having frequencies higher than the first frequency, and the lower portion is a lower frequency bin comprising a portion of the received signal having frequencies lower than the first frequency.

[0037] The determining can be performed based on a difference between a normalized power estimate of the upper frequency bin and a normalized power estimate of the lower frequency bin. The normalization factor can be a sum of the power estimates of the upper frequency bin and the lower frequency bin.

[0038] The determining of the presence of a person can comprise determining a logit function of the normalized power of the upper frequency bin. The determining of the presence of a person can comprise determining a logit function of the normalized power of the lower frequency bin. The determining of the presence of a person can comprise determining a first logit function of the normalized power of the upper frequency bin, and determining a second logit function of the normalized power of the lower frequency bin.

[0039] The logit function can take the following form:

[0040]

[0041] where X(t, f0+1) is a coefficient representing the upper frequency bin at time t, and X(t, f0-1) is a coefficient representing the lower frequency bin at time t.

[0042] The processor may also be configured to repeat steps (a) to (b) at a predetermined rate. Each repetition of steps (a) to (d) may be performed over a time window of the received signal. The time window may be around 20 milliseconds.

[0043] After determining the presence of a person, the processor may be further configured to determine whether the person is moving toward or away from the receiver. Determining whether the person is moving toward or away from the receiver may be further based on a first likelihood ratio test for determining whether the person is moving toward the receiver and a second likelihood ratio test for determining whether the person is moving away from the receiver. A log-likelihood ratio may be derived for each likelihood ratio and may be recursively calculated from previous values ​​of the corresponding log-likelihood ratio.

[0044] The processor may also be configured to bring the video conferencing device out of standby mode when the presence of a person has been determined.

[0045] In some embodiments, there may be a first receiver and a second receiver, wherein the first receiver is configured to receive reflections of ultrasonic signals having components with frequencies greater than a first frequency, and the second receiver is configured to receive reflections of ultrasonic signals having components with frequencies less than the first frequency. Thus, the first receiver may receive an upper portion of the received signal, while the second receiver may receive a lower portion of the received signal. Determining the difference between the upper portion of the received signal and the lower portion of the received signal may include determining whether the first receiver and / or the second receiver received the corresponding signal. In the event that both receivers receive the signal, it may be determined that the received signal is a false alarm. In the event that only one receiver receives the signal, it may be determined that a person is present.

[0046] Other aspects of the present invention provide: a computer program comprising code that causes the computer to perform the method of the first aspect when run on a computer; a computer-readable medium storing the computer program, the computer program comprising code that causes the computer to perform the method of the first aspect when run on a computer; and a computer system programmed to perform the method of the first aspect.

[0047] The optional features of the aspects of the invention described above may be applied alone or in combination with any aspect of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Embodiments of the present invention will now be described by way of example with reference to the accompanying drawings, in which:

[0049] Figure 1 A system according to an embodiment of the present invention is shown;

[0050] Figure 2 Shown Figure 1Spectrum (time-frequency diagram) of spatial points near the microphone of the system;

[0051] Figure 3 is a flowchart of a method according to an embodiment of the present invention;

[0052] Figure 4 It shows that when a broadband signal is received, |X(t,f0-1)| 2 ,|X(t,f0+1)| 2 and Graph of

[0053] Figure 5 shows a graph of L(t) and the corresponding histogram when no motion occurs;

[0054] Figure 6 It shows that when motion occurs, |X(t,f0-1)| 2 ,|X(t,f0+1)| 2 and L(t) graphs;

[0055] Figure 7 shows a graph of L(t), and corresponding graphs of log-likelihood ratio and detection threshold; and

[0056] Figure 8 is a flow chart of a method according to a variant embodiment of the present invention. DETAILED DESCRIPTION

[0057] Aspects and embodiments of the present invention will now be discussed with reference to the accompanying drawings. Other aspects and embodiments will be apparent to those skilled in the art.

[0058] Figure 1 A room including a system of the present invention is shown. The system includes an ultrasonic transmitter 101 that transmits an ultrasonic signal 102 at a first frequency f0 (also referred to as a tone). In this example, f0 is 22,000 Hz, but it can take any ultrasonic frequency value (e.g., at least 20 kHz and not more than 24 kHz). The transmitter continuously emits the tone. In this example, the transmitter is a speaker that is also used in video conferencing equipment. The system also includes a receiver 103, which in this example is a microphone that is also part of the video conferencing equipment. The receiver is configured to detect not only the ultrasonic signal at f0 but also reflections of the signal that has been Doppler-shifted. The system also includes one or more processors (not shown) configured to use the signal received from receiver 103 to determine whether a person is present in the room.

[0059] As an ultrasonic signal propagates through a room, it reflects from various objects and / or interfaces. For example, after reflecting from a wall, an undeflected reflection 104, i.e., a reflection still at f0, is returned to receiver 103. This undeflected reflection is ignored because it provides little information about the presence of a person in the room (indicated by motion). However, after reflecting from a person 105 moving toward receiver 103, an upward-moving reflection 106 is returned to the receiver. Upward-moving reflection 106 has a frequency higher than f0. This upward-moving reflection provides information about the presence of a person in the room, specifically that the person is moving toward receiver 103. Similarly, after reflecting from a person 107 moving away from receiver 103, a downward-moving reflection 108 is returned to the receiver. The frequency of downward-moving reflection 108 is lower than f0. This downward-moving reflection also provides information about the presence of a person in the room, specifically that the person is moving away from receiver 103.

[0060] However, as previously mentioned, transient noises 110 (e.g., the noise produced by a door 109 slamming or clapping (perhaps from outside the room)) have a relatively wide frequency range and may contain components having the same or similar frequencies as the upward or downward moving components. These transient noises that do not originate from the transmitter 101 may be interpreted by the receiver (or a processor connected thereto) as indicating the presence of a person.

[0061] Figure 2 The following is a spectrum diagram (time-frequency diagram) of the spatial region near the receiver, illustrating this principle. The transmitter emits a tone at a frequency of 22,000 Hz, thus providing a narrowband signal with a long duration. At t0, a person walks toward the receiver at a first speed, so the receiver receives an upward movement signal 201. In this example, the person then increases their speed toward the receiver, causing the receiver to receive a further upward movement signal 202. The person then stops, and the receiver no longer receives the upward movement signal. At t1, the person walks away from the receiver, so the receiver receives a downward movement signal 203.

[0062] Next, at time t2, the receiver receives transient signal 204. The signal is transient because it has a limited presence on the "x" axis. However, the transient signal includes components with the same frequency as the upward signal 201, the further upward signal 202, and the downward signal 203. There is a risk that a processor connected to the receiver might interpret transient signal 204 as indicating the presence of a person.

[0063] Figure 3is a flow chart of a method according to an embodiment of the present invention. In a first step 301, an ultrasonic tone is transmitted at a frequency f0. Next, in step 302, the signal received by one or more receivers is transformed from a microphone frame (i.e., a short time frame of the microphone signal) to the time-frequency domain. This is performed for a short time window or time frame (e.g., 20 ms) of the received signal, and a filter bank is used to transform the received signal to the frequency domain. This results in a plurality of coefficients describing a plurality of time-frequency intervals denoted as X(t,f), where t is the time frame, and t=0,1,2,…, f is the frequency interval index, and f=0,1,2,…,K-1, where K is the discrete Fourier transform (DFT) size. The filter bank is designed so that the filter is clear (almost no leakage from adjacent frequency intervals) and has a sufficiently narrow bandwidth to detect slow walking speeds, e.g., 65 Hz.

[0064] Next, in step 303, the logit of the normalized Doppler shift power is calculated. Let f0 denote the frequency bin index containing the frequency of the transmitted tone (e.g., 22000 Hz). The logit of the normalized Doppler shift power is then defined as:

[0065]

[0066] where |…| represents the absolute value, and is the logit function of p. The parameters of the logit function, namely is the normalized power estimate of the frequency bins above f0, and the normalization factor is the sum of the power estimates of the frequency bins above and below f0.

[0067] This means that p is a number between 0 and 1 that can be compared to a probability. The logit function then transforms this probability so that it can take values ​​between ±∞.

[0068] After calculating this for a given time window, the method moves to steps 304 and 307, which are performed simultaneously. At step 304, a first log-likelihood ratio, log-likelihood ratio 0, is updated based on the calculated logit to indicate how likely it is that there is motion toward the receiver. Simultaneously, at step 307, a second log-likelihood ratio, log-likelihood ratio 1, is updated based on the calculated logit to indicate how likely it is that there is motion away from the receiver.

[0069] In general, there is no closed-form expression for the likelihood ratio, so calculating it can be computationally expensive. However, since the values ​​of L(t) are found to be approximately independent and normally distributed, a simple expression for the log-likelihood ratio can be derived.

[0070] The log-likelihood ratio of a type known per se in the art has the general expression:

[0071]

[0072] in is the possibility of motion towards or away from the receiver, and is the possibility of no motion. See, for example, "CuSum Algorithm - A Small Review" by Pierre Granjon, the contents of which are incorporated herein by reference.

[0073] Furthermore, the log-likelihood ratios can be recursively calculated using the previous and new values ​​of L(t). Initialization of the log-likelihood ratios can include initializing them to zero, meaning that the initial likelihood ratio is 1. This means that, at initialization, the likelihood of motion is the same as the likelihood of no motion. Let LLR0(t) represent the log-likelihood ratio of motion toward the receiver and LLR1(t) represent the log-likelihood ratio of motion away from the receiver, and the update equation for the log-likelihood ratio can be specified as:

[0074]

[0075]

[0076] In these expressions, δ is the expected change in amplitude, that is, the expected deviation of the mean of L(t) from zero mean during motion. This is a constant set during the initialization phase. The variance of L(t) is denoted as var. This is either set to a fixed value during the initialization phase or estimated when computing the value of L(t).

[0077] Once the log-likelihood ratios have been calculated using some or all of the information from the calculated logits, each log-likelihood ratio is compared to a threshold in steps 305 and 308. If one of the likelihood ratios exceeds its threshold, then "yes" in steps 305 and / or 308, motion towards or away from the receiver can then be determined in steps 306 and 309, respectively.

[0078] Once motion is detected, or no motion is detected ("No" in steps 305 and 308), the method returns to step 302 to obtain a new time window. In this way, the motion detection method can run continuously. In the example discussed below, the value of δ is selected to be 5, and var is estimated based on the value of L(t). In one example, when it is known that there is no motion, the maximum likelihood estimator of L(t) is used in the time window to obtain an estimate of var. When it is known that there is no motion, the maximum likelihood estimate can be calculated as L(t) for t in the time window. 2 average value.

[0079] The logit function discussed above is particularly suitable for motion detection for three reasons: (1) immunity to transient noise; (2) normally distributed values; and (3) indication of motion direction.

[0080] First, regarding point (1), Figure 4 It shows that when a transient broadband signal is received, the lower frequency interval |X(t,f0-1)| 2 , upper frequency interval |X(t,f0+1)| 2 and 's curve graph.

[0081] Figure 4 The above figure is |X(t,f0-1)| 2 The graph of power versus time is therefore a graph of the power in the frequency bin immediately below f 0. As can be seen from the two peaks (at approximately 3 and 5 seconds), this frequency bin contains transient, broadband noise components. Figure 4 The middle graph in |X(t,f0+1)| 2 A plot of power versus time is therefore a plot of power in the frequency bin immediately above f0. Again, two peaks can be seen at approximately 3 and 5 seconds. It can then be determined that the transient, broadband signal is adding roughly equal amounts to the upper and lower portions of the signal.

[0082] Therefore, if Figure 4 As shown in the following figure in , which is a plot of L(t), the roughly equal contributions from the upper and lower parts of the signal are offset by the ratio in L(t), so the logit function is noise-immune to transients, broadband, noise, or signals.

[0083] Then about point (2), the normal distribution value, Figure 5 The graph of L(t) and the corresponding histogram when no motion occurs are shown. If there is no motion, and therefore no Doppler shift, then the values ​​of L(t) for t = 0, 1, 2, ... follow a distribution similar to a normal distribution. This has been verified experimentally, and the results are shown in Figure 5 In the lower middle histogram, we can also expect that the values ​​of L(t) are almost independent of each other. As mentioned above, since these values ​​are independent and normally distributed, a simple expression for the log-likelihood ratio can be derived.

[0084] Next, regarding the motion detection of point (3), Figure 6 It shows that when motion occurs |X(t,f0-1)| 2 ,|X(t,f0+1)| 2Graph of L(t) and L(t). As previously mentioned, the values ​​of L(t) for t = 1, 2, ... contain information about the direction of motion. In the absence of motion, L(t) is close to zero. When there is motion toward the receiver, L(t) is typically positive, for example, a few decibels above zero. Conversely, when there is motion away from the receiver, L(t) is typically negative, for example, a few decibels below zero. Figure 6 The upper figure in FIG is a graph of the frequency range below the frequency containing f0, showing the person walking away from the receiver between 6 and 10 seconds by increasing amplitude. Figure 6 The middle figure in FIG is a graph of the frequency interval above the frequency interval containing f0, showing the person walking towards the receiver between 4 and 6 seconds by increasing amplitude. Figure 6 The figure below is a graph of the logit function L(t), showing that the function takes positive values ​​between 4 and 6 seconds and negative values ​​between 6 and 10 seconds, indicating that L(t) can be used to determine the direction of motion relative to the receiver.

[0085] Figure 7 A plot of L(t) is shown, along with corresponding plots of the log-likelihood ratio and the detection threshold. Figure 7 The upper figure in is L(t), corresponding to Figure 6 The following figure. Figure 7 The bottom figure in Figure 7 is a graph of the log-likelihood ratio and the detection threshold used to detect the presence of motion toward or away from the receiver. Line 701 shows the value of LLR0(t) discussed above, and line 702 shows the value of LLR1(t) discussed above. The dashed line 703 is the threshold, which is 100 in this example.

[0086] As can be seen, line 701 rises above threshold 703 between 4 and 5 seconds, indicating motion towards the receiver. At approximately 7 seconds, line 702 rises above threshold 703, while line 701 falls below the threshold, indicating motion away from the receiver.

[0087] Figure 8 Flowchart of a variant method according to an embodiment of the present invention. Figure 8 and Figure 3 Where features are shared by the flowcharts shown in FIG. 1 , the same features are indicated by the same reference numerals. Figure 3 Compared with the method shown, Figure 8 The method utilizes two logit functions: the first logit function L1(t) is adjusted to better detect motion towards the video system, and the second logit function L2(t) is adjusted to better detect motion away from the video system.

[0088] The following observations can be used to improve Figure 3Consider the logit function discussed above. During movement toward the videoconferencing equipment in a room, reflections from moving objects will result in higher received frequencies. However, before being received by the receiver in the videoconferencing equipment, the reflections hit the back wall, then the moving object, and then the back wall again, resulting in a lower received frequency. It is worth noting that the received frequencies constitute the range of Doppler shifts. With movement toward the videoconferencing equipment, most of these Doppler shifts will have higher frequencies, but some will have lower frequencies.

[0089] Therefore, L1(t) can be expressed as:

[0090]

[0091] That is, X(t, f0-1) in L(t) is replaced by X(t, f0-3). This results in a more robust signal for detecting motion toward the video conferencing device because, at normal walking speeds, the received Doppler shift down to f0-3 is minimal. In addition, noise immunity remains good because broadband noise (such as a door slamming or clapping) has very similar energy in the two frequency bins f0-3 and f0+1. However, when motion is directed away from the video conferencing device, the logit function L1(t) performs poorly. Therefore, a second logit function L2(t) is used, which is formulated as:

[0092]

[0093] This is shown in steps 303a to 309a and 303b to 309b of the two logit functions executed in parallel.

[0094] Although the present invention has been described in conjunction with the above exemplary embodiments, many equivalent modifications and variations will be apparent to those skilled in the art upon presentation of this disclosure. Therefore, the exemplary embodiments of the present invention set forth above are intended to be illustrative rather than restrictive. Various changes may be made to the described embodiments without departing from the spirit and scope of the present invention.

Claims

1. A method for detecting people based on ultrasound, comprising the following steps: (a) transmitting an ultrasonic signal from a transmitter, the ultrasonic signal including a component at a first frequency f0; (b) receiving a received signal including reflections of the ultrasonic signal, the received signal including components having frequencies higher and lower than the first frequency, including an upper portion of the received signal including frequencies higher than the first frequency and a lower portion of the received signal including frequencies lower than the first frequency; (c) determining a difference between the upper portion of the received signal including frequencies higher than the first frequency and the lower portion of the received signal including frequencies lower than the first frequency; and (d) determining whether a person is present based on the difference between the upper portion and the lower portion.

2. The method according to claim 1, wherein An upper portion of the received signal includes higher frequencies immediately adjacent to the first frequency, and a lower portion of the received signal includes lower frequencies immediately adjacent to the first frequency.

3. The method according to claim 1, wherein The method includes dividing the received signal into a plurality of intervals, each interval representing a frequency range of the received signal, and wherein the upper portion is an upper frequency interval, comprising a portion of the received signal having a frequency higher than the first frequency, and the lower portion is a lower frequency interval, comprising a portion of the received signal having a frequency lower than the first frequency.

4. The method according to claim 3, wherein The determining is performed based on a difference between a normalized power estimate of the upper frequency bin and a normalized power estimate of the lower frequency bin.

5. The method according to claim 4, wherein The normalization factor is the sum of the power estimates of the upper frequency bin and the lower frequency bin.

6. The method according to claim 3, wherein Determining the presence of a person includes determining a logit function of a normalized power estimate of the upper frequency bin.

7. The method according to claim 6, wherein The logit function takes the following form: Here, X(t,f0+1) is the coefficient representing the upper frequency interval at time t, and X(t,f0-1) is the coefficient representing the lower frequency interval at time t.

8. The method according to claim 1, wherein Steps (b) to (d) are repeated at a predetermined rate.

9. The method of claim 1, further comprising the step of: after determining that a person exists, determining whether the person is moving toward or away from the receiver.

10. The method according to claim 9, wherein Determining whether the person is moving toward or away from the receiver is further based on a first likelihood ratio test for determining whether the person is moving toward the receiver and a second likelihood ratio test for determining whether the person is moving away from the receiver.

11. The method according to claim 10, wherein A log-likelihood ratio is derived for each likelihood ratio and is recursively calculated from the previous value of the corresponding log-likelihood ratio.

12. The method according to any one of claims 1 to 11, wherein When the presence of a person has been determined, the method includes taking the video conferencing device out of standby mode.

13. A personnel detection system, comprising: a transmitter configured to transmit an ultrasonic signal including a component at a first frequency f0; one or more receivers configured to receive a reception signal generated by reflections of the ultrasonic signal; and One or more processors configured to, in response to the one or more receivers receiving the received signal, the received signal including components having frequencies above and below the first frequency, including an upper portion of the received signal including frequencies above the first frequency and a lower portion of the received signal including frequencies below the first frequency, to: (a) determining a difference between the upper portion of the received signal comprising frequencies higher than the first frequency and the lower portion of the received signal comprising frequencies lower than the first frequency; and (b) determining whether a person exists based on the difference between the upper portion and the lower portion.

14. The system according to claim 13, wherein: An upper portion of the received signal includes higher frequencies immediately adjacent to the first frequency, and a lower portion of the received signal includes lower frequencies immediately adjacent to the first frequency.

15. The system according to claim 13, wherein: The processor is also configured to divide the received signal into a plurality of intervals, each interval representing a frequency range of the received signal, and wherein the upper portion is an upper frequency interval, comprising a portion of the received signal having a frequency higher than the first frequency, and the lower portion is a lower frequency interval, comprising a portion of the received signal having a frequency lower than the first frequency.

16. The system according to claim 15, wherein: The determining is performed based on a difference between a normalized power estimate of the upper frequency bin and a normalized power estimate of the lower frequency bin.

17. The system according to claim 16, wherein: The normalization factor is the sum of the power estimates of the upper frequency bin and the lower frequency bin.

18. The system of claim 15, wherein: Determining the presence of a person includes determining a logit function of a normalized power estimate of the upper frequency bin.

19. The system of claim 18, wherein: The logit function takes the following form: Here, X(t,f0+1) is the coefficient representing the upper frequency interval at time t, and X(t,f0-1) is the coefficient representing the lower frequency interval at time t.

20. The system according to any one of claims 13 to 19, wherein: The processor is further configured to repeat steps (a) to (b) at a predetermined rate.

21. The system according to any one of claims 13 to 19, wherein: The processor is further configured to determine whether the person is moving toward or away from the receiver after determining that the person is present.

22. The system of claim 21, wherein: Determining whether the person is moving toward or away from the receiver is further based on a first likelihood ratio test for determining whether the person is moving toward the receiver and a second likelihood ratio test for determining whether the person is moving away from the receiver.

23. The system according to any one of claims 13 to 19, wherein: When the presence of a person has been determined, the processor is configured to bring the video conferencing device out of standby mode.

Images