JP7653679B2 - Drowning Judgment System - Google Patents

Description

The present invention relates to a drowning detection system that can detect drowning or drowning people over a wide area with high accuracy while significantly reducing the user's effort in video monitoring.

Conventionally, there has been known an in-pool imaging system that is attached to the lane ropes of a pool and captures the area adjacent to the lane ropes (see, for example, Patent Document 1).

JP 2018-061139 A

However, the above technology cannot fully function in places such as the ocean, where there are no course ropes and where wide areas must be monitored.

Therefore, the present invention aims to provide a drowning detection system that can detect drowning or drowning people over a wide area with high accuracy while significantly reducing the user's effort in video monitoring.

The present invention provides a drowning determination system that includes a memory unit that stores multiple patterns of changes in the exposed part of a human above the water when the human is drowning in water; a detection unit that detects the exposed part of the human subject above the water in a series of time-series images of the water surface captured by a camera; a determination unit that refers to the memory unit and determines whether the human subject is drowning or about to drown based on the changes in the exposed part in the series of time-series images; and a notification unit that outputs a notification signal when the determination unit determines that the human subject is drowning or about to drown.

With this configuration, it is possible to detect drowning or people who are likely to drown with a high degree of certainty based on drowning-specific features. In addition, because it is automatically determined whether a target person is drowning or likely to drown, the effort required of the user to monitor the video is significantly reduced. Furthermore, by using a detection unit with high detection accuracy, it is possible to detect drowning that occurs far away and cannot be detected by the human eye. Furthermore, by using multiple image capture devices to capture multiple different water surfaces, or by using a drone as the image capture device, it is possible to detect drowning that occurs in places that cannot be detected by normal monitoring.

In accordance with another aspect of the present invention, a drowning determination program is provided that is installed in a computer that stores multiple patterns of changes in the vertical direction of the exposed part of a human being above water when the human is drowning in water, the drowning determination program comprising the steps of: detecting the exposed part of the human being above water that is captured in a series of time-series images of the water surface taken by a camera; referring to the computer, determining whether the human being is drowning or about to drown based on the changes in the vertical direction of the exposed part detected in the series of time-series images; and outputting an alarm signal when it is determined that the human being is drowning or about to drown.

According to another aspect of the present invention, a drowning determination system is provided that includes a memory unit that stores the speed or size of waves that pose a high risk of a human being being swept away and drowning; a detection unit that detects waves heading toward the coastline captured in a series of time-series images of the water surface taken by an imaging device, and a target human being located between the waves and the coastline; a calculation unit that calculates a predicted speed or predicted size of the waves detected in the series of time-series images when they reach the target human; a determination unit that refers to the memory unit and determines whether the target human is at high risk of being swept away by waves and drowning based on the calculated predicted speed or predicted size; and an alarm unit that outputs an alarm signal when the determination unit determines that there is a high risk.

With this configuration, it is possible to detect with high certainty people who are at high risk of being swept away by waves and drowning, because it is possible to determine the "predicted speed and predicted size of the detected wave when it reaches the target human", which is difficult for the human eye to distinguish. In addition, because it is automatically determined whether the target human is at high risk of being swept away by waves and drowning, the labor of video surveillance by the user is significantly reduced. In addition, by using a detection unit with high detection accuracy, it is possible to detect drowning that occurs far away and cannot be detected by the human eye. Furthermore, by using multiple imaging devices to capture multiple different water surfaces, or by using a drone as the imaging device, it is possible to detect drowning that occurs in places that cannot be detected by normal surveillance.

According to another aspect of the present invention, there is provided a drowning determination program that is installed in a computer that stores the speed or size of waves that pose a high risk of a human being being swept away and drowning, and that includes the steps of: detecting waves heading toward the coastline in a series of time-series images of the water surface taken by a camera, and a target human being located between the waves and the coastline; calculating a predicted speed or predicted size of the waves detected in the series of time-series images when they reach the target human; referring to the computer, determining whether or not there is a high risk of the target human being being swept away by waves and drowning based on the calculated predicted speed or predicted size; and outputting an alarm signal when the determination unit determines that there is a high risk.

According to another aspect of the present invention, a drowning determination system is provided that includes a detection unit that detects a pair of currents flowing toward the coastline and a target human exposed above the water surface in a series of time-series images of the water surface captured by a camera device, a determination unit that determines that the target human is at risk of being swept away by a rip current occurring between the pair of currents if the detected target human is located between the pair of currents and the distance between the pair of currents is within a predetermined distance, and a notification unit that outputs a notification signal when the determination unit determines that there is a risk.

With this configuration, it is possible to grasp the occurrence of a rip current F2, which is difficult for the human eye to distinguish, and therefore it is possible to detect with high certainty people who are at risk of being swept away by a rip current. In addition, because it is automatically determined whether or not the target person is at risk of being swept away by a rip current, the labor of video surveillance by the user is significantly reduced. Furthermore, by using a detection unit with high detection accuracy, it is possible to detect rip currents that occur far away that cannot be detected by the human eye. Furthermore, by photographing multiple different water surfaces with multiple imaging devices or using a drone as the imaging device, it is possible to detect rip currents that occur in places that cannot be detected by normal surveillance.

According to another aspect of the present invention, a drowning determination program is provided that includes the steps of: detecting a pair of currents flowing toward the coastline and a human subject exposed above the water surface in a series of time-series images of the water surface captured by a camera; determining that the human subject is at risk of being swept away by a rip current occurring between the pair of currents if the detected human subject is located between the pair of currents and the distance between the pair of currents is within a predetermined distance; and outputting an alarm signal if the determination unit determines that there is a risk.

According to another aspect of the present invention, a drowning determination system is provided that includes a memory unit that stores multiple patterns of ripples or bubbles on the water surface around a human drowning in relation to the position of the human, a detection unit that detects the exposed portion of the human subject above the water and the ripples or bubbles on the water surface around the exposed portion in a series of time-series images of the water surface captured by an image capture device, a determination unit that refers to the memory unit and determines whether the human subject is drowning or about to drown based on the position of the exposed portion and the ripples or bubbles detected in the series of time-series images, and an alarm unit that outputs an alarm signal when the determination unit determines that the human subject is drowning or about to drown.

With this configuration, it is possible to detect with high certainty a drowning person or a person who is about to drown, based on ripples or bubbles on the water surface around the person, which occur with a high probability when the person is drowning. In addition, because it is automatically determined whether the target person is drowning or about to drown, the effort required of the user to monitor the video is significantly reduced. Furthermore, by using a detection unit with high detection accuracy, it is possible to detect drowning that has occurred far away and cannot be detected by the human eye. Furthermore, by using multiple image capture devices to capture multiple different water surfaces, or by using a drone as the image capture device, it is possible to detect drowning that has occurred in a place that cannot be detected by normal monitoring.

According to another aspect of the present invention, there is provided a drowning determination program that is installed in a computer that stores multiple patterns of ripples or bubbles on the water surface around a human drowning in relation to the position of the human, the program comprising the steps of: detecting an exposed portion of the human subject above the water and ripples or bubbles on the water surface around the exposed portion, which are captured in a series of time-series images of the water surface taken by a camera; determining, with reference to the computer, whether the human subject is drowning or about to drown based on the position of the exposed portion and the ripples or bubbles detected in the series of time-series images; and outputting an alarm signal when the determination unit determines that the human subject is drowning or about to drown.

The drowning detection system of the present invention makes it possible to detect drowning or drowning people over a wide area with high accuracy while significantly reducing the effort required of the user to monitor video.

FIG. 1 is an explanatory diagram of a usage state of the drowning determination system according to the first embodiment of the present invention; Block diagram of a drowning determination system according to a first embodiment of the present invention. 1 is a flowchart of a drowning determination system according to a first embodiment of the present invention; FIG. 13 is an explanatory diagram of a usage state of the drowning determination system according to the second embodiment of the present invention. Block diagram of a drowning determination system according to a second embodiment of the present invention. 1 is a flowchart of a drowning determination system according to a second embodiment of the present invention. FIG. 13 is an explanatory diagram of a usage state of the drowning determination system according to the third embodiment of the present invention. Block diagram of a drowning determination system according to a third embodiment of the present invention. 10 is a flowchart of a drowning determination system according to a third embodiment of the present invention. FIG. 13 is an explanatory diagram of a usage state of the drowning determination system according to the fourth embodiment of the present invention. Block diagram of a drowning determination system according to a fourth embodiment of the present invention. 11 is a flowchart of a drowning determination system according to a fourth embodiment of the present invention.

The drowning determination system 1 according to an embodiment of the present invention will be described below with reference to Figures 1 to 3.

The drowning determination system 1 is for detecting a drowning person or a person who is likely to drown from a video of the water surface. In this embodiment, as shown in FIG. 1, an example is described in which a drowning person A who is likely to drown is detected in approximately real time from a video (a series of time-series images X) taken over a wide area of the sea surface by a drone D.

As shown in FIG. 2, the drowning determination system 1 includes a memory unit 2, a detection unit 3, a determination unit 4, and a notification unit 5.

The memory unit 2 stores multiple patterns of changes in the vertical direction of the exposed part of a person above water when the person is drowning in water.

Possible patterns of change in the exposed parts in the up and down direction include "repeatly exposing only the head at high speed, then exposing up to the neck", "repeatly exposing only the head at medium speed, then exposing up to the chest", "exposing up to the neck with almost no up and down movement", etc.

In addition, in this embodiment, the multiple patterns of vertical changes in the exposed parts of the human body above the water include multiple patterns of vertical changes in the human body's arms exposed above the water surface.

Possible patterns of change in the up and down direction of the arms include "repeated exposure and submersion of the arms at high speed (while flapping)", "repeated exposure and submersion of the arms at medium speed (while flapping)", "no arms exposed", etc.

The memory unit 2 may further store multiple patterns of horizontal changes in the exposed parts of the human body above water.

The detection unit 3 detects the exposed portion of the target human A above the water that is captured in a series of time-series images X of the water surface captured by a photographing device (drone D).

This detection may be performed using well-known posture determination techniques, tracking techniques, or the like. It is preferable that the detection unit 3 obtains a series of time-series images X in approximately real time, and performs the above detection in approximately real time as well.

The boundary between the exposed part of human A above the water and the ocean surface can be determined based on the color boundary (the ocean surface is blue, humans are skin or hair color, etc.).

The determination unit 4 refers to the storage unit 2 and determines whether the target human A is drowning or about to drown based on the change in the vertical direction of the exposed portion detected in the series of time-series images X. In detail, if there is one among the multiple patterns stored in the storage unit 2 that has a predetermined degree of agreement or higher with the change in the vertical direction of the detected exposed portion, it determines that "the target human A is drowning or about to drown."

When it is determined that "target human A is drowning or about to drown," the notification unit 5 outputs a notification signal indicating danger. The notification signal preferably identifies target human A on the display screen of the user's information terminal showing the image captured by the imaging device (drone D), but it may simply output a display or sound indicating danger. Possible users include beach managers and lifeguards.

Next, the operation of the drowning determination system 1 will be explained using the flowchart in Figure 3.

First, when the detection unit 3 detects an exposed portion of the target human A above water in the series of time-series images X (S1: YES), the determination unit 4 refers to the memory unit 2 and determines whether the target human A is drowning or about to drown based on the vertical change in the exposed portion detected in the series of time-series images X (S2).

If it is determined that "target person A is drowning or about to drown" (S2: YES), a warning signal is output by the warning unit 5 (S3). Based on this warning signal, lifeguards waiting on the shore can be dispatched to quickly and efficiently rescue target person A.

As described above, in the drowning determination system 1 according to this embodiment, multiple patterns of changes in the vertical direction of the exposed part of a person above the water when the person is drowning in water are stored in the memory unit 2, and the drowning of the target person A is determined based on the vertical changes in the exposed part detected in the series of time-series images X by referring to the memory unit 2.

With this configuration, it is possible to detect with high certainty a drowning person or a person who is about to drown, based on the characteristics specific to drowning. In addition, since it is automatically determined whether the target person A is drowning or about to drown, the user's effort in video surveillance is significantly reduced. Furthermore, by using a detection unit 3 with high detection accuracy, it is possible to detect drowning that occurs far away and cannot be detected by the human eye. Furthermore, by using multiple imaging devices to capture multiple different water surfaces, or by using a drone as the imaging device, it is possible to detect drowning that occurs in a place that cannot be detected by normal surveillance.

In addition, in the drowning determination system 1 according to this embodiment, the multiple patterns of changes in the vertical direction of the part of the person exposed above the water include multiple patterns of changes in the vertical direction of the person's arms exposed above the water.

This configuration takes into account arm movements, which are likely to be the most violent up-and-down movements of a person drowning in water, making it possible to detect drowning or drowning people with a higher degree of certainty.

In addition, in the drowning determination system 1 according to this embodiment, the alarm signal is intended to identify the target human A on the display screen of the user's information terminal showing the image captured by the imaging device (drone D).

This configuration allows lifeguards and others waiting on the shore to quickly and efficiently dispatch rescue missions to target person A.

Next, we will explain the drowning determination system 10 according to the second embodiment of the present invention.

In the second embodiment, as shown in FIG. 4, the presence of waves W is taken into consideration to detect a person B who is likely to be swept away by the waves.

As shown in FIG. 5, the drowning determination system 10 includes a memory unit 11, a detection unit 12, a calculation unit 13, a determination unit 14, and a notification unit 15.

The memory unit 11 stores the speed or size of waves that pose a high risk of drowning.

Whether or not a wave is dangerous is preferably stored based on a combination of wave speed and wave size, but it is also possible to store only one of them. It may also be stored as "if the wave speed (size) is above a certain level, it is highly dangerous regardless of the combination with the wave size (speed)."

The detection unit 12 detects waves W moving toward a coastline E and a target human B positioned between the waves W and the coastline E in a series of time-series images Y of the water surface captured by a photographing device (drone D).

When a wave W is detected that extends over a predetermined distance in a direction roughly parallel to the coastline or the horizon and resembles a white line heading toward the coastline E, it may be determined that a "wave W has been detected."

In addition, when the boundary between the brown of the beach and the white of the waves that have reached the beach is detected, it may be determined that "coastline E has been detected."

The calculation unit 13 calculates the predicted speed and predicted size of the detected wave W when it reaches the target human B.

At this time, the calculation unit 13 first judges whether the wave W will reach the target human B. For example, it is possible to judge whether the wave W is likely to reach the target human B based on a change in the positional relationship between the wave W and the target human B in a time-series image Y of multiple frames, a change in the relative speed between the wave W and the target human B, and a change in the distance from the target human B to the coastline E. For example, if the position of the target human B deviates from the traveling direction of the wave W, it can be judged that the wave W will not reach. Also, if the speed of the target human B is faster than the speed of the wave W, it can be judged that the wave W will not reach. Furthermore, if the target human B is likely to reach the coastline E before the wave W reaches the target human B, it can be judged that the wave W will not reach.

Then, when it is determined that the wave W will reach the target human B, the predicted speed and predicted size of the wave W at the time of arrival are calculated. As for the predicted speed, for example, it is possible to calculate the timing at which the wave W will reach the target human B based on the speed of the wave W in the multiple frames of the time-series image Y and the speed of the target human B, and use the speed at this arrival timing. In addition, it is possible to calculate the predicted size based on, for example, the change in size (acceleration) of the wave W in the multiple frames of the time-series image Y and the timing at which the wave W will reach the target human B.

In addition, since the positional relationship between the wave W and the target human B, the speed of the wave W, the speed of the target human B, the distance between the target human B and the coastline E, etc. are expected to change from moment to moment, it is preferable that the calculation unit 13 constantly performs calculations.

The determination unit 14 refers to the memory unit 11 and determines whether or not there is a high risk that the target human B will be swept away by the waves and drown, based on the calculated predicted speed or predicted size. In detail, if any of the wave speeds or sizes stored in the memory unit 11 have a predetermined degree of agreement or higher with the calculated predicted speed and predicted size, it determines that "the target human B is at high risk of being swept away by the waves and drowning."

The notification unit 15 outputs a notification signal indicating danger when it is determined that "there is a high risk that the target human B will be swept away by the waves and drown." The notification signal preferably identifies the target human B on the display screen of the user's information terminal showing the image captured by the imaging device (drone D), but it may simply output a display or sound indicating the danger. Possible users include beach managers and lifeguards.

Next, the operation of the drowning determination system 10 according to this embodiment will be explained using the flowchart in FIG.

First, when the detection unit 12 detects a wave W and a target human B in a series of time-series images Y (S11: YES), the calculation unit 13 calculates the predicted speed and predicted size of the wave W detected in the series of time-series images Y when it reaches the target human B (S12).

Next, the determination unit 14 refers to the memory unit 11 and determines whether or not there is a high risk of the target human B being swept away by the waves and drowning, based on the calculated predicted speed and predicted size (S13).

If it is determined that "there is a high risk that the target person B will be swept away by the waves and drown" (S13: YES), the alarm unit 15 outputs an alarm signal (S14). Based on this alarm signal, lifeguards and others waiting on the shore can be dispatched to quickly and efficiently rescue the target person B.

As described above, in the drowning determination system 10 according to this embodiment, the speed or size of waves that pose a high risk of a human being being swept away and drowning is stored in the memory unit 11, and the memory unit 11 is referenced to determine whether or not there is a high risk of the target human B being swept away by a wave and drowning, based on the predicted speed and predicted size when the detected wave W reaches the target human B.

With this configuration, the "predicted speed and predicted size when the detected wave W reaches the target human B" is determined, which is difficult for the human eye to distinguish, so it is possible to detect with high certainty people who are at high risk of being swept away by the waves and drowning. In addition, because it is automatically determined whether the target human B is at high risk of being swept away by the waves and drowning, the user's labor in video surveillance is significantly reduced. In addition, by using a detection unit 12 with high detection accuracy, it is possible to detect drowning that occurs far away and cannot be detected by the human eye. Furthermore, by using multiple imaging devices to capture multiple different water surfaces, or by using a drone as the imaging device, it is possible to detect drowning that occurs in places that cannot be detected by normal surveillance.

Next, we will explain the drowning determination system 20 according to the third embodiment of the present invention.

In the third embodiment, we detect people who are likely to be swept away by rip currents.

As shown in Figure 7, when there is a pair of currents F1 heading toward the coastline E, a rip current refers to a current F2 that occurs between the pair of currents F1 and flows away from the coastline E. Rip currents tend to move slowly, but because they are so slow, it is possible for a situation to occur in which "person C is caught in a rip current and unknowingly finds himself or herself carried far away from the coastline E."

As shown in FIG. 8, the drowning determination system 20 includes a detection unit 21, a determination unit 22, and a notification unit 23.

The detection unit 21 detects a pair of flows F1 heading toward the coastline E and a human subject C exposed above the water surface, which are captured in a series of time-series images Z of the water surface captured by an imaging device (drone D). The flow F1 heading toward the coastline E can be detected, for example, by detecting waves heading toward the coastline E (white line-shaped waves that extend over a predetermined distance in a direction approximately parallel to the coastline or the horizon and head toward the coastline E).

When the detected target human C is located between a pair of currents F1 and the distance L between the pair of currents F1 is within a predetermined distance, the determination unit 22 determines that "the target human C is at risk of being swept away by a rip current F2 occurring between the pair of currents F1." As the distance L, for example, it is possible to measure the distance between the end of a wave included in one current F1 and the end of a wave included in the other current F1. Rip currents usually occur with a width of about 10 to 30 meters, so the predetermined distance can be determined with reference to the said width of about 10 to 30 meters.

When it is determined that "the target human C is at risk of being swept away by a rip current," the notification unit 23 outputs a notification signal indicating the danger. The notification signal is preferably one that identifies the target human C on the display screen of the user's information terminal showing the image captured by the imaging device (drone D), but it may simply output a display or sound indicating the danger. Possible users include beach managers and lifeguards.

Next, the operation of the drowning determination system 20 according to this embodiment will be explained using the flowchart in FIG.

First, when the detection unit 21 detects a pair of flows F1 and a target human C (S21: YES), the judgment unit 22 determines whether the target human C is located between the pair of flows F1 and whether the distance L between the pair of flows F1 is within a predetermined distance (S22).

If the target human C is located between a pair of currents F1 and the distance L between the pair of currents F1 is within a predetermined distance (S22: YES), it is determined that "the target human C is at risk of being swept away by the rip current F2" (S23), and a warning signal is output by the warning unit 23 (S24). Based on this warning signal, lifeguards waiting on the coast can be dispatched to rescue the target human C quickly and efficiently.

As described above, in the drowning determination system 20 according to this embodiment, when the target person C is located between a pair of flows F1 and the distance L between the pair of flows F1 is within a predetermined distance, it is determined that the target person C is at risk of being swept away by the rip current F2.

With this configuration, it is possible to grasp the occurrence of a rip current F2 that is difficult for the human eye to distinguish, and therefore it is possible to detect with high certainty a person who is at risk of being swept away by a rip current F2. In addition, since it is automatically determined whether or not the target person C is at risk of being swept away by a rip current F2, the user's labor in video surveillance is significantly reduced. In addition, by using a detection unit 21 with high detection accuracy, it is possible to detect a rip current F2 that occurs far away and cannot be detected by the human eye. Furthermore, by photographing multiple different water surfaces with multiple imaging devices or using a drone as the imaging device, it is possible to detect a rip current F2 that occurs in a place that cannot be detected by normal surveillance.

Next, we will explain the drowning determination system 30 according to the fourth embodiment of the present invention.

In the fourth embodiment, as shown in FIG. 10, it is determined whether the target human G is drowning or about to drown based on the ripples or bubbles around the target human G.

As shown in FIG. 11, the drowning determination system 30 includes a memory unit 31, a detection unit 32, a determination unit 33, and an alarm unit 34.

The memory unit 31 stores multiple patterns of ripples or bubbles on the water surface around a human being relative to the position of the human being when the human is drowning in water.

Possible ripple or bubble patterns include "ripples with a radius of XX m centered on the human", "ripples centered on a position within a specified distance from the human", "large bubbles (splashes) within a specified distance from the human", "small bubbles (splashes) within a specified distance from the human", etc. The specified distance can be set based on the position where a person's outstretched hands will land in the water.

In addition, in this embodiment, the memory unit 31 further stores the pattern of arm movement of a person when the person is drowning in water in association with the multiple patterns of ripples or bubbles.

Possible arm movement patterns include "flapping with the elbow extended," "flapping with the elbow bent," "flapping alternately from left to right," etc.

The detection unit 32 detects the exposed portion of the target human G above the water, and the ripples or bubbles on the water surface surrounding the exposed portion, captured in a series of time-series images V of the water surface captured by a photographing device (drone D).

This detection may be performed using well-known posture determination techniques, tracking techniques, or the like. It is preferable that the detection unit 32 acquires a series of time-series images V in approximately real time, and performs the above detection in approximately real time as well.

The boundary between the exposed part of the human subject G above the water and the ocean surface can be determined based on the color boundary (the ocean surface is blue, humans are skin or hair color, etc.).

The determination unit 33 refers to the memory unit 31 and determines whether the target human G is drowning or about to drown based on the position of the exposed part and the ripples or blisters detected in the series of time-series images V. In detail, if there is one among the multiple patterns stored in the memory unit 31 that has a predetermined degree of match or higher with the position of the detected exposed part and the ripples or blisters, it determines that "the target human G is drowning or about to drown."

In particular, in this embodiment, the determination unit 33 refers to the memory unit 31 and determines whether the target human G is drowning or about to drown, taking into account the arm movements detected in the series of time-series images V.

The alarm unit 34 outputs an alarm signal indicating danger when it is determined that the target human G is drowning or about to drown. The alarm signal preferably identifies the target human G on the display screen of the user's information terminal showing the image captured by the imaging device (drone D), but it may simply output a display or sound indicating danger. Possible users include beach managers and lifeguards.

Next, the operation of the drowning determination system 30 will be explained using the flowchart in Figure 12.

First, when the detection unit 32 detects an exposed portion of the target human G above the water and ripples or bubbles on the water surface surrounding the exposed portion in the series of time-series images V (S31: YES), the determination unit 33 refers to the memory unit 31 and determines whether the target human G is drowning or about to drown based on the position of the exposed portion and the ripples or bubbles detected in the series of time-series images V (S32).

If it is determined that the target human G is drowning or about to drown (S32: YES), a warning signal is output by the warning unit 34 (S33). Based on this warning signal, lifeguards waiting on the shore can be dispatched to the target human G to rescue him or her quickly and efficiently.

As described above, in the drowning determination system 30 according to this embodiment, multiple patterns of ripples or bubbles on the water surface around a human being relative to the position of the human being when the human is drowning in water are stored in the memory unit 31, and by referring to the memory unit 31, it is determined whether the target human being G is drowning or about to drown based on the position of the exposed part and the ripples or bubbles detected in the series of time-series images V.

With this configuration, it is possible to detect with high certainty a drowning person or a person who is about to drown, based on ripples or bubbles on the water surface around the person, which occur with a high probability when the person is drowning. In addition, since it is automatically determined whether the target person G is drowning or about to drown, the user's effort in video surveillance is significantly reduced. Furthermore, by using a detection unit 32 with high detection accuracy, it is possible to detect drowning that has occurred far away and cannot be detected by the human eye. Furthermore, by capturing images of multiple different water surfaces with multiple imaging devices, or by using a drone as the imaging device, it is possible to detect drowning that has occurred in a place that cannot be detected by normal surveillance.

In addition, the drowning determination system 30 according to this embodiment also takes into account the arm movements detected in the series of time-series images V to determine whether the target human G is drowning or about to drown.

This configuration takes into account arm movements, which are the most likely to occur when a person is drowning in water, making it possible to detect drowning or drowning people with a higher degree of certainty.

The drowning determination system of the present invention is not limited to the above-described embodiment, and various modifications and improvements are possible within the scope of the claims.

For example, in the first embodiment, ripples or bubbles on the water surface around the target human A may be further taken into consideration.

In this case, the memory unit 2 further stores multiple patterns of ripples or bubbles on the water surface around the human being when he or she is drowning in water, in association with multiple patterns of changes in the vertical direction of the exposed portion described above, and the detection unit 3 further detects ripples or bubbles on the water surface around the subject human A captured in the series of time-series images X.

For example, it is thought that a change in the vertical direction of the torso (up and down movement) will generate ripples according to the magnitude and speed of the change. Similarly, a change in the vertical direction of an arm will generate ripples and bubbles according to the magnitude and speed of the change at the position on the water surface where the arm repeatedly becomes exposed and submerged.

Then, the determination unit 4 refers to the memory unit 2 and determines whether the target human A is drowning or about to drown, taking into consideration the ripples or bubbles on the water surface detected in the series of time-series images X. In detail, if there is one among the multiple patterns stored in the memory unit 2 that has a predetermined degree of match or higher with the detected combination of "changes in the vertical direction of the exposed part" and "ripples or bubbles", it is determined that "the target human A is drowning or about to drown."

In this way, by taking into account ripples or bubbles on the water surface around target human A, it becomes possible to detect drowning or drowning people with a higher degree of certainty.

In addition, in the first embodiment, the facial direction of the target human A may also be taken into consideration.

In this case, the memory unit 2 further stores the pattern of the face direction when a person is drowning in water in association with multiple patterns of changes in the up and down direction of the exposed part described above, and the detection unit 3 further detects the face direction of the target human A in the water reflected in the series of time-series images X.

For example, if the person is "looking up" or "facing forward," it is likely that the person is struggling, and therefore there is a high possibility that there will be a large change in the vertical direction of the exposed parts (both the torso and arms). On the other hand, if the person is "looking down," it is likely that the person is unconscious, and therefore there is a high possibility that there will be almost no change in the vertical direction of the exposed parts (both the torso and arms).

Then, the determination unit 4 refers to the storage unit 2 and determines whether the target human A is drowning or about to drown, taking into consideration the facial orientation detected in the series of time-series images X. In detail, if there is one among the multiple patterns stored in the storage unit 2 that matches the detected combination of "change in the vertical direction of the exposed part" and "facial orientation" to a degree equal to or higher than a predetermined degree, it determines that "the target human A is drowning or about to drown."

In this way, by taking into account the direction of the face of the target person A, it is possible to detect people who are drowning or about to drown with a higher degree of certainty.

In addition, in the first embodiment, other people besides the target person A may also be taken into consideration.

In this case, the detection unit 3 further detects other humans who appear in the series of time-series images X, and the judgment unit 4 determines whether the target human A is drowning or about to drown, taking into account the distance between the target human A and the other humans or any variation in the distance.

For example, if another human is detected near target human A (e.g., within a first predetermined distance), this may be a situation in which children are playing, and in this case, it may be possible to reduce the probability of determining that the human is "drowning or about to drown."

On the other hand, if no other human is detected near target human A (e.g., within a second specified distance), it is possible to increase the probability of determining that the target human is "drowning or about to drown."

In addition, if another human is detected approaching target human A at a speed greater than or equal to a predetermined value, it is possible that the human is attempting to rescue the target human. In this case, it is possible to increase the probability of determining that the target human is drowning or near drowning.

In this way, by taking into account people other than the target person A, it is possible to detect people who are drowning or about to drown with a higher degree of certainty.

In addition, in the first embodiment, the distance from the target person A to the coastline E may be further taken into consideration.

In this case, the detection unit 3 further detects the coastline E captured in the series of time-series images X, and the determination unit 4 determines whether the target human A is drowning or about to drown, taking into account the distance from the target human A to the coastline E.

For example, if the distance from the target human A to the coastline E is close (e.g., within a third predetermined distance), it is possible that children are playing, and in this case, it is possible to reduce the probability of determining that the target human A is "drowning or about to drown."

On the other hand, if the distance from the target human A to the coastline E is far (e.g., greater than or equal to a fourth predetermined distance), it may be possible to increase the probability of determining that the target human A is "drowning or about to drown."

In this way, by taking into account the distance from the target person A to the coastline E, it is possible to detect people who are drowning or about to drown with a higher degree of certainty.

In addition, in the first embodiment, the change in the vertical direction of the human arm exposed above the water was considered, but the change in the horizontal direction of the human arm exposed above the water may also be considered. In particular, in the example described above that considers ripples or bubbles on the water surface around the subject human A, ripples and bubbles will appear at the position where the arm on the water surface repeatedly appears and disappears, depending on the magnitude and speed of the change, so it is effective to also consider the change in the horizontal direction of the arm.

In addition, in the above embodiment, a person who is drowning or who is about to drown is detected from the video (series of time-series images X) captured by the drone D, but a person who is drowning or who is about to drown may also be detected from video captured by a camera installed on a beach or the like.

Furthermore, when filming with a drone, human movements are detected as moving at a speed relative to the flight speed of the drone, and the detected movements may differ significantly from the various patterns stored in the memory unit. Therefore, when filming with a moving object such as a drone, a sensor capable of detecting the moving object's own speed and direction of movement, such as a gyro sensor, may be provided to the moving object so that human movements are judged at absolute speed. Conversely, human movements can also be accurately judged by storing patterns corresponding to multiple relative speeds in the memory unit and judging human movements based on the patterns corresponding to the relative speeds.

The present invention can also be applied to a program or method equivalent to the processing performed by the drowning determination system, and to a recording medium storing the program. In the case of a recording medium, the program is installed in a computer or the like. Here, the recording medium storing the program may be a non-transient recording medium. A non-transient recording medium may be, but is not limited to, a CD-ROM or the like.

1, 10, 20, 30 Determination system 2, 11, 31 Memory unit 3, 12, 21, 32 Detection unit 4, 14, 22, 33 Determination unit 5, 15, 23, 34 Notification unit 13 Calculation unit A, B, C, G Target human D Photographing device (drone)
E Coastline F1 Flow F2 Rip current L Distance W Waves X, Y, Z, V Time series images

Claims (3)

A memory section that memorizes the size of waves that pose a high risk of drowning, and
A detection unit that detects waves moving toward a coastline and a target human being located between the waves and the coastline in a series of time-series images of the water surface captured by an imaging device;
a calculation unit that calculates a speed of a target human being detected in the series of time-series images, a speed of a wave detected in the series of time-series images, and a predicted size of the wave detected in the series of time-series images when it reaches the target human being;
a judgment unit that judges that there is a high risk of the target human being being swept away by the wave and drowning when it is judged that the wave will reach the target human before the target human reaches the coastline based on the calculated target human speed and the calculated wave speed, and when the predicted size corresponds to the stored wave size;
a notification unit that outputs a notification signal when the determination unit determines that the risk is high;
A drowning determination system comprising: A program installed in a computer that stores the size of waves that pose a high risk of drowning to humans,
A step of detecting a wave moving toward a coastline and a target human being located between the wave and the coastline, the target human being being detected in a series of time-series images of a water surface captured by an image capturing device;
Calculating a speed of a human subject detected in the series of time-series images, a speed of a wave detected in the series of time-series images, and a predicted size of the wave detected in the series of time-series images when it reaches the human subject;
a step of determining that there is a high risk of the target human being being swept away by the wave and drowning when it is determined that the wave will reach the target human before the target human reaches the coastline based on the calculated target human speed and the calculated wave speed, and when the predicted size corresponds to the stored wave size;
outputting a notification signal when the determination unit determines that the risk is high;
A drowning determination program comprising: 1. A computer implemented method in which wave sizes that pose a high risk of human drowning are stored, the method comprising:
A step of detecting a wave moving toward a coastline and a target human being located between the wave and the coastline, the target human being being detected in a series of time-series images of a water surface captured by an image capturing device;
Calculating a speed of a human subject detected in the series of time-series images, a speed of a wave detected in the series of time-series images, and a predicted size of the wave detected in the series of time-series images when it reaches the human subject;
a step of determining that there is a high risk of the target human being being swept away by the wave and drowning when it is determined that the wave will reach the target human before the target human reaches the coastline based on the calculated target human speed and the calculated wave speed, and when the predicted size corresponds to the stored wave size;
outputting a notification signal when the determination unit determines that the risk is high;
A drowning determination method comprising:

Images