KR20210116318A - Systen for preventing indoor human casualties - Google Patents

Abstract

The present invention relates to a system for preventing indoor life accidents, which detects only dynamic movement of a human body through a low-resolution image-based human body sensor installed indoors to generate an accident risk signal when the dynamic movement does not occur for a set time or more, or is different from usual, so that a guardian is notified of the situation in real-time, thereby enabling the guardian to automatically check the health status of a single person from a distance. According to the present invention, the system comprises: a sensor module installed indoors to photograph a subject; an image monitoring module monitoring activity of the subject by analyzing the image captured by the sensor module; and a health care module inferring the subject's health state through inference based on an activity monitoring result transmitted from the image monitoring module to generate care information and notify a guardian of the care information on the subject.

Description

Indoor accident detection system {SYSTEN FOR PREVENTING INDOOR HUMAN CASUALTIES}

The present invention relates to an indoor human accident detection system, and more particularly, by detecting only the dynamic movement of the human body through a low-resolution image-based human body detection sensor installed indoors, an accident when the dynamic movement does not occur for more than a set time or is different from usual It relates to an indoor fatal accident detection system that generates a danger signal and notifies the guardian in real time, so that the guardian can automatically check the health status of a person living alone from a distance.

With the recent acceleration of nuclear family and aging, the number of senior citizens living alone without a supporter who can help is increasing.

In the case of the elderly living alone, even if their health suddenly becomes critical or an emergency occurs one day, the so-called death of the elderly living alone is a social issue, in which precious lives are lost due to failure to timely request for emergency rescue, or, worse, left alone for a long time after death. have.

In order to solve the problem of loneliness for the elderly living alone, government agencies, local governments and private security companies installed sensor-based devices to measure changes in the subject's physical condition to monitor the subject's activity level, gas risk, crime prevention, and fire. We have established a system for overall safety management, such as emergency services, and provide services in connection with rescue organizations such as 119 in case of emergency.

However, the service provided through such sensor-based device installation may cause the target object to resist monitoring/control due to the attachment of sensors and cameras in the home. In light of the fact that maintenance costs and labor costs for service monitoring continue to occur, which greatly increases the burden of service costs, there is a problem that effective/practical services cannot be provided to the low-income seniors living alone in need of real help.

The present invention has been devised to solve the above problems, and its purpose is to detect only the dynamic movement of the human body through a low-resolution image-based human body detection sensor installed indoors, so that the dynamic movement does not occur for more than a set time or is different from usual. In this case, it is to provide an indoor human accident detection system that can automatically check the health status of a person living alone from a distance by generating an accident risk signal to notify the guardian in real time.

According to the present invention, a sensor module installed indoors to photograph a subject; an image monitoring module for monitoring the activity of the subject by analyzing the image captured by the sensor module; and a health care module for generating and notifying care information for a subject by inferring the subject's health state through inference based on the activity monitoring result transmitted from the image monitoring module. It provides an indoor human accident detection system comprising a.

Preferably, the sensor module comprises: an image sensor that captures a designated region of interest to obtain image data for the region of interest; and a communication unit transmitting the image data transmitted from the image sensor to the image monitoring module. It includes, wherein the image data captured by the image sensor is photographed in units of one image frame, such as a still image, or is photographed in units of a plurality of image frames, such as a moving image.

Preferably, the image monitoring module includes: an image monitoring analysis unit for receiving image data transmitted in real time from the sensor module, performing image analysis, and outputting monitoring determination information on the subject's activity in the image; a determination standard setting unit for storing and managing determination reference information for monitoring determination of the image monitoring and analysis unit; and a communication unit that receives the image data transmitted from the sensor module, transmits the image data to the image monitoring analysis unit, and transmits the monitoring determination information transmitted from the image monitoring analysis unit to the health care module. It is characterized in that it includes.

Preferably, the image monitoring analysis unit identifies the posture the subject is currently taking in the image including the subject and counts the maintenance time of the identified posture to generate the posture identification value and the holding time information as monitoring determination information. characterized.

Preferably, the image monitoring and analysis unit includes: a person detection unit for detecting a person who is a subject from an image including the subject transmitted in real time from the sensor module; a posture identification unit for identifying a body posture currently taken by the person detected by the person detecting unit; Posture monitoring unit for calculating the posture holding time by counting the time that the posture identified by the posture identification unit is maintained; and an inference information generation unit for outputting the posture identification value of the posture identification unit and the posture maintenance time information of the posture monitoring unit as monitoring determination information capable of inferring a health state; It is characterized in that it includes.

Preferably, the determination criterion setting unit manages first posture identification information, which is an identification criterion using the direction of the major axis and the minor axis that can be derived from the person area of the person detected by the person detection unit, and the first posture identification reference information is reference information for identifying the person's posture as a standing posture when the long axis of the person area corresponds to the vertical axis of the image, and reference information for identifying the person's posture as a lying posture when the long axis of the person area corresponds to the horizontal axis of the image It is characterized in that it includes.

Preferably, the posture identification unit identifies a major axis and a minor axis of the person area with reference to the first posture identification reference information managed by the determination standard setting unit, and the long axis of the person area is an angle of ±44° with the vertical axis of the image If the position of the person is matched within the range, the current person's posture is identified as a standing posture, and the posture monitoring unit calculates the posture maintenance time by counting the time the identified standing posture is maintained, or the long axis of this person area is ±44 from the horizontal axis of the image. If it corresponds within the angle of °, the current posture of the person is identified as a supine posture, and the posture monitoring unit counts the time the identified supine posture is maintained to calculate the posture maintenance time.

Preferably, the determination criterion setting unit manages second posture identification information, which is an identification criterion using a ratio of a length of a major axis to a minor axis that can be derived from the person region of the person detected by the person detection unit, and the second posture identification standard The information includes reference information for identifying the posture of the person lying down as a crouched posture when the length of the major axis does not exceed three times the length of the minor axis for the posture identified as the lying posture according to the first posture identification reference information characterized in that

Preferably, after determining the supine posture, the posture identification unit identifies the major axis length and the minor axis length of the person area with reference to the second posture identification reference information managed by the determination criterion setting unit, and determines the posture identified as the lying posture. In contrast, if the length of the major axis does not exceed three times the length of the minor axis, the posture of the person lying down is identified as a crouched posture, and the posture monitoring unit counts the time the identified crouching posture is maintained to calculate the posture maintenance time. characterized.

Preferably, the image monitoring analysis unit identifies the subject's wearing clothes in the image including the subject and generates the clothes identification value and the wearing period information as monitoring discrimination information by counting the wearing period of the identified clothes. do.

Preferably, the image monitoring and analysis unit includes: a person detection unit for detecting a person who is a subject from an image including the subject transmitted in real time from the sensor module; a clothing identification unit for identifying the currently worn clothing of the person detected by the person detection unit; a clothes monitoring unit calculating a wearing period by counting the wearing period of the clothes identified by the wearing clothes identification unit; and an inference information generating unit for outputting the clothing identification value of the wearing clothes identification unit and the wearing period information of the clothes monitoring unit as monitoring identification information capable of inferring a health state; It is characterized in that it includes.

Preferably, the wearing clothes identification unit is characterized in that the wearable clothes of the subject are identified through the clothes color and clothes shape information managed by the discrimination criterion setting unit.

Preferably, the image monitoring analysis unit identifies the movement movement of the subject in the image containing the subject and calculates the movement direction and movement speed of the identified movement movement to generate movement path and movement speed information as monitoring determination information. characterized.

Preferably, the image monitoring and analysis unit includes: a person and a target detection unit for detecting a person who is a subject and a target that is a moving destination of the subject from an image including the subject transmitted in real time from the sensor module; a movement identification unit for recognizing the movement of the person to the object detected by the person and the object detection unit; a movement monitoring unit for calculating the movement path and movement speed of the subject for the movement movement identified by the movement identification unit; and an inference information generator for outputting the movement path and movement speed information of the movement monitoring unit as monitoring determination information capable of inferring a health state; It is characterized in that it includes.

Preferably, the movement identification unit is characterized in that it distinguishes and recognizes the movement movement of the everyday object and the non-routine object managed by the determination criterion setting unit.

Preferably, the health care module receives the monitoring determination information transmitted in real time from the image monitoring module and compares the monitoring determination information for each subject with the care standard information set in advance for the subject through inferencing the subject's health status a care information processing unit generating care information for a subject by analogy; a care standard setting unit for storing and managing care reference information for inferring a health state of the care information processing unit; and a communication unit that receives the monitoring determination information transmitted from the image monitoring module, transmits it to a care information processing unit, and notifies the care information transmitted from the care information processing unit to a communication terminal of a preset guardian; It is characterized in that it includes.

Preferably, the care standard information of the care standard setting unit includes the reference duration of the lying posture and the necessary care information accordingly, the reference duration of the squatting posture and the necessary care information accordingly, the standard duration of the standing posture and the necessary care accordingly. It is characterized in that at least one or more of information, reference duration for each identification garment and necessary care information, movement route and corresponding necessary care information, movement speed, and necessary care information accordingly are included.

According to the present invention, only the dynamic movement of the human body is detected through a low-resolution image-based human body detection sensor installed indoors, and when the dynamic movement does not occur for more than a set time or is different from usual, an accident risk signal is generated and the situation is notified to the guardian in real time This has the effect of allowing the guardian to automatically check the health status of a single person from a distance.

In particular, since it is a method that detects only the dynamic movement of the human body as a low-quality image, rather than a method of accurately capturing an image of a person, it also has the advantage of reducing the objection to monitoring/controlling the subject.

In addition, it can be operated only by installing a simple low-quality image sensor without the cost burden of installing, maintaining, and repairing separate sensor-based equipment in the home, so it is possible to create an optimal effect that meets the intended purpose of preventing loneliness at a minimum cost. possible.

In addition, by inferring the subject's health status through inference based on the results of monitoring activities such as the subject's current posture, the subject's clothing, and the subject's movement in the video, it is more accurate and reasonable for the health of the person living alone. It also has the effect of enabling status check.

1 is a view for explaining an indoor human accident detection system according to a first embodiment of the present invention.
2 is a view for explaining the internal configuration of the human body detection sensor according to the first embodiment of the present invention.
3 is a diagram for explaining the circuit configuration of the image sensor according to the first embodiment of the present invention.
4 is a diagram for explaining another circuit configuration of the image sensor according to the first embodiment of the present invention.
5 is a diagram for explaining the internal configuration of the control terminal according to the first embodiment of the present invention.
6 is a view for explaining an indoor human accident detection system according to a second embodiment of the present invention.
7 is a view for explaining various configuration examples of the indoor human accident detection system according to the second embodiment of the present invention.
8 is a view for explaining an image monitoring analysis unit according to the first reasoning method of the indoor human accident detection system according to the second embodiment of the present invention.
9 is a view for explaining the driving principle of the first reasoning method of the indoor human accident detection system according to the second embodiment of the present invention.
10 is a view for explaining an image monitoring analysis unit according to the second reasoning method of the indoor human accident detection system according to the second embodiment of the present invention.
11 is a view for explaining the driving principle of the second reasoning method of the indoor human accident detection system according to the second embodiment of the present invention.
12 is a view for explaining an image monitoring analysis unit according to the third reasoning method of the indoor human accident detection system according to the second embodiment of the present invention.
13 is a view for explaining the driving principle of the third reasoning method of the indoor human accident detection system according to the second embodiment of the present invention.
14 is a view for explaining a driving method of an indoor human accident detection system according to a second embodiment of the present invention.

Since the present invention can have various changes and can have various embodiments, specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, and it should be understood to include all modifications, equivalents and substitutes included in the spirit and scope of the present invention. In describing each figure, like reference numerals have been used for like elements.

Terms such as first, second, A, and B may be used to describe various elements, but the elements should not be limited by the terms. The above terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, a first component may be referred to as a second component, and similarly, a second component may also be referred to as a first component. and/or includes a combination of a plurality of related listed items or any of a plurality of related listed items.

When a component is referred to as being “connected” or “connected” to another component, it is understood that the other component may be directly connected or connected to the other component, but other components may exist in between. it should be On the other hand, when it is said that a certain element is "directly connected" or "directly connected" to another element, it should be understood that no other element is present in the middle.

The terms used in the present application are only used to describe specific embodiments, and are not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly dictates otherwise. In the present application, terms such as “comprise” or “have” are intended to designate that a feature, number, step, operation, component, part, or combination thereof described in the specification exists, but one or more other features It should be understood that this does not preclude the existence or addition of numbers, steps, operations, components, parts, or combinations thereof.

Unless defined otherwise, all terms used herein, including technical and scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related art, and should not be interpreted in an ideal or excessively formal meaning unless explicitly defined in the present application. does not

Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the accompanying drawings.

1 is a view for explaining an indoor human accident detection system according to a first embodiment of the present invention.

The indoor human accident detection system according to the first embodiment of the present invention is a human body detection sensor 100, the human body detection sensor ( 100) connected by wire or wirelessly to the gateway 200 that relays the detection signal, the control terminal 300 that receives the detection signal through the gateway 200 to determine whether there is an accident risk, and the control terminal 300 When the result of determining whether there is an accident risk is received, the communication company server 400 may be included to notify the situation in real time to the set guardian's client terminal 500 .

A plurality of the human body detection sensor 100 may be installed at various locations in the indoor space where the subject is active, each of which captures a set range to acquire image data for the operation area and analyzes the image to detect only the subject's dynamic movement will do

2 is a view for explaining the internal configuration of the human body detection sensor according to the first embodiment of the present invention.

The human body detection sensor 100 includes an image sensor 110 that acquires image data for an operating area by photographing a set range, and an image sensor 110 that receives image data captured from the image sensor 110 and analyzes the received image data. a human body recognition unit 130 that extracts an image of the subject's human body, recognizes whether there is a dynamic movement of the human body within the operation zone based on this, and outputs an activity detection signal or an inactivity detection signal; and the human body recognition unit 130 ) may include a module control unit 140 that periodically transmits a detection signal transmitted from the gateway 200 to the gateway 200 .

The image data photographed by the image sensor 110 may be photographed in units of one image frame, such as a still image, or in some cases, may be photographed in units of a plurality of image frames, such as a moving image. In addition, image data captured by each pixel of the image sensor 110 is transmitted to and stored in a separate memory unit (not shown), and the stored image data is an image for human body detection by the human body recognition unit 130 . It is used as basic data for analysis.

The human body recognition unit 130 analyzes the image data stored in the memory unit and extracts the human body of the subject located in the operation area as a monitoring object. Here, in extracting the monitoring object, the object can be extracted according to the change of the RGB value or the Black/White value by comparing every image frame according to the time series sequence of the image data, and the size and shape of the object are limited , animals, raindrops, curtains, etc. can be detected and extracted by filtering only the human body to be detected by filtering the detected objects.

In addition, the human body recognition unit 130 determines whether or not the subject is dynamic in the operation area according to whether or not a monitoring object for the human body is extracted, and generates an activity detection signal or an inactivity detection signal according to the determined result. will output

The module control unit 140 periodically transmits a detection signal including an activity detection signal or an inactivity detection signal transmitted from the human body recognition unit 130 to the gateway 200 to enable accumulation of real-time status signals.

An image sensor 110 according to a first embodiment of the present invention will now be described in detail with reference to FIGS. 3 and 4 .

3 is a diagram for explaining a circuit configuration of the image sensor according to the first embodiment of the present invention, and FIG. 4 is a diagram for explaining another circuit configuration of the image sensor according to the first embodiment of the present invention.

Referring to FIG. 3 , the image sensor according to the first embodiment of the present invention includes an optical sensor unit 111 including first and second photodiodes 111a and 111b for converting an optical signal into an electrical signal; When amplifying each electrical signal converted from the optical sensor unit 111, the amplifying unit 112 and the comparing unit 113 comparing the respective electrical signals amplified from the amplifying unit 112 and amplifying the difference and a signal converter 114 for converting the electrical signal output from the comparator 113 into a digital signal.

The optical sensor unit 111 may include first and second photodiodes 111a and 111b. The first photodiode 111a is a general photodiode, and when an optical signal (light) is incident, an electrical signal is detected. On the other hand, the second photodiode 111b has an external optical filter that blocks visible light or infrared light, so that even when an optical signal is incident, no electrical signal is detected.

Here, the optical filter may be made of a metal film or polysilicon capable of blocking visible or infrared rays, or other materials having an equivalent light blocking function, and a metal line ( A metal line) may be used as it is, and aluminum as a metal line may be deposited on the surface of the photodiode to block visible light incident to the second photodiode 111b.

In this case, in the above-described example, a form in which aluminum is deposited on the second photodiode 111b and acts as a light filter has been described, but the present invention is not limited thereto. It is also possible to form it to act as an optical filter.

The amplifying unit 112 may include a first AMP 112a and a second AMP 112b. The first AMP 112a is configured to amplify or buffer the signal input from the first photodiode 111a, and the second AMP 112b receives the signal input from the second photodiode 111b. A configuration for amplifying or buffering.

The comparison unit 113 compares the output value of the first AMP 112a with the output value of the second AMP 112b. The comparator 113 has a high gain and may be configured as a variable differential amplifier. Here, the reason for increasing the gain of the differential amplifier is to operate well in an ultra-low environment, and also to enable the entire module and device including the same to be driven with low power.

Basically, the first and second photodiodes 111a and 111b have a value unique to each pixel and have fixed pattern noise that does not change with time due to the characteristics of the photodiodes. This fixed pattern noise is noise generated from non-uniformity related to the AMP in the pixel, and is usually corrected in advance, but appears as noise when exposed for a long time.

In the present invention, an optical filter is formed in the second photodiode 111b constituting the optical sensor unit 111 in order to remove the influence of the fixed pattern noise.

Since the first AMP 112a is connected to the first photodiode 111a, it has the leakage fixed pattern noise of the diode together with the electrical signal according to the optical signal, but the second AMP 112b ) and connected to the second photodiode 111b, since an optical filter is formed, the second AMP 112b does not have an electrical signal according to the optical signal, but has only fixed pattern noise.

Therefore, in the process of comparing the output value of the first AMP 112a and the output value of the second AMP 112b in the comparator 113, only an electric signal according to the pure optical signal of the first photodiode 111a is detected. In this process, the fixed pattern noise of the first photodiode 111a is also removed.

In addition, the signal conversion unit 114 converts the two outputs of the comparison unit 113 into digital signals and transmits them to the human body recognition unit 130 so that the human body recognition unit 130 analyzes the image data for each pixel. The human body of the subject located within the operation area is to be extracted as a monitoring object.

Meanwhile, in the present invention, a separate auxiliary light source unit (not shown) is turned on in a low-illuminance environment so that the image sensor can recognize a subject even in a low-illuminance environment. In this case, in order to prepare for a situation in which the auxiliary light source is reluctant to turn on the lighting and to reduce the lighting cost, the auxiliary light source unit is made of a low-illuminance IR LED, and an optical filter that blocks only visible light is formed in the second photodiode 111b Since the second photodiode 111b has an external optical filter that blocks only visible light, it operates in response to the IR LED as it is, and as a result, outputs the same electrical signal as the first photodiode 111a. It becomes impossible to remove the pattern noise.

Of course, in order to solve this problem, an optical filter for blocking IR may be formed on the second photodiode 111b.

However, in order to more fundamentally and surely solve this problem, in the present invention, a compensation unit 117 that is switched in a low-illuminance environment and operates instead of the second photodiode 111b is added.

Referring to FIG. 4 , another image sensor according to the first embodiment of the present invention includes, in addition to the configuration of the optical sensor unit 111 , the amplifier 112 , the comparator 113 , and the signal conversion unit 114 , the In the optical sensor unit 111, a switch unit 116 for switching so that the second photodiode 111b is not operated by the auxiliary light source of the auxiliary light source unit in a low light environment, and the switch unit 116 in a low light environment It may be configured to include a compensating unit 117 operating in place of the second photodiode 111b of the optical sensor unit 111 .

The compensator 117 includes a capacitive component having the same or similar capacitance as the second photodiode 111b. Preferably, the capacitive element may be composed of a compensation capacitor 117a that does not respond to IR or visible light.

The switch unit 116 connects the second photodiode 111b or the compensation capacitor 117a to the second AMP 112b by switching the state according to a situation in which the auxiliary light source is turned on according to ambient illuminance.

That is, in a normal illuminance environment in which the auxiliary light source is turned off, the device controller 200 controls the switch 116 to connect the second photodiode 111b to the second AMP 112b. In this case, as described above, the second AMP 112b has only fixed pattern noise because the optical filter is formed on the second photodiode 111b. As a result, the comparison unit 113 removes the fixed pattern noise from image data. will get

Also, in a low-illuminance environment in which the auxiliary light source is turned on, the control unit that turns on the auxiliary light source controls the switch unit 116 so that the compensation capacitor 117a of the compensation unit 117 is connected to the second AMP 112b. In this case, the compensation capacitor 117a of the compensator 117 does not naturally respond to the lighting of the auxiliary light source, and the comparator 113 obtains only the image data of the first photodiode 111a.

Here, the gateway 200 collects a detection signal periodically transmitted from the human body detection sensor 100 and transmits it to the control terminal 300 .

5 is a diagram for explaining the internal configuration of the control terminal according to the first embodiment of the present invention.

The control terminal 300 receives and accumulates the detection signals of the human body detection sensor 100 through the gateway 200 to determine whether there is an accident risk.

The control terminal 300 includes a setting unit 350 for setting an allowable range for the dynamic movement of a subject, an occupancy detection unit 340 for detecting whether the subject is present and outputting an occupancy signal or an outing signal, and the gateway ( 200) through the receiving unit 320 for periodically receiving a detection signal including an activity detection signal or an inactivity detection signal of the human body detection sensor 100, the presence signal from the occupancy detection unit 340 is received in the state A target state signal including an accident risk signal or an accident safety signal to determine whether there is an accident risk by analyzing the accumulation of detection signals including the activity detection signal or inactivity detection signal with reference to the allowable range of the setting unit 350 An accident determination unit 360 that outputs can be configured.

The setting unit 350 sets and stores the allowable range for the subject's dynamic movement. Such an allowable range may be set as a range for an allowable time and an allowable place.

For example, the allowable range for the subject's dynamic movement may be set to 24 hours as the allowable time. That is, a case in which the inactivity detection signal of the subject collected from the human body detection sensor 100 continues for 24 hours or more may be set as an accident risk case.

In addition, the allowable range for the subject's dynamic movement may be set as a combination of a specific area and allowable time as an allowable place. That is, a case in which a target's inactivity detection signal collected from the human body detection sensor 100 continues for 24 hours or more in a specific area (eg, a kitchen) may be set as an accident risk case.

In this case, the human body detection sensor 100 may be installed dispersedly in various locations in the material space, and the control terminal 300 includes an activity detection signal or an inactivity detection signal from each of the human body detection sensors 100 . The sensing signal is periodically received and integrated. Therefore, when the allowable range of the setting unit 350 is set only for the allowable time, the dynamic movement of the subject will be analyzed with respect to a state in which all detection signals of the installed human body detection sensors 100 are combined.

The occupancy detection unit 340 detects whether the subject is occupant and outputs a signal indicating whether the position includes an occupancy signal or an outing signal. In fact, such an indoor fatal accident detection system functions only in the occupancy situation of the subject and should not operate when the subject is out.

Therefore, the presence detection unit 340 can collect the communication terminal device location information of the subject from the communication company server 400 in real time or in units of a predetermined period, and it is possible to check whether the subject is present through the collected location information.

The accident determination unit 360 includes a positioning signal including an occupancy signal or an outing signal of the occupancy detection unit 340 and an activity detection signal or inactivity of the human body detection sensor 100 transmitted from the receiving unit 320 . By comparing and analyzing the detection signal including the detection signal with the allowable range for the subject's dynamic movement of the setting unit 350, the subject state signal including the accident risk signal or the accident safety signal is finally output.

That is, whether or not an accident has occurred for the subject will be determined based on whether the inactivity detection signal is detected over an allowable range while the presence of the subject is confirmed.

In this way, the target state signal including the accident risk signal or the accident safety signal analyzed and output from the control terminal 300 is collected in the communication company server 400, and the communication company server 400 periodically includes the guardians set for the target. Real-time situation notification (subject accident risk, subject safety) is made to the client terminal 500 . Therefore, the guardians of the single person will be able to automatically check the health status of the single person from a distance.

On the other hand, in the above-described embodiment, it has been described as a preferred embodiment that the system determines whether there is an accident risk by detecting the dynamic movement of an indoor subject and recognizing whether there is a dynamic movement of the human body within the operation zone, but the present invention is limited thereto it doesn't happen There is a possibility of error in the method of determining the state of the subject only with the presence of actual dynamic movement, and the number of states that can be determined is also very limited.

Therefore, in the present invention, the present invention does not stop at deriving the subject's current state through the above-described dynamic movement, but also converts the movement information of the indoor subject transmitted from the human body sensor 100 into data and machine-learns it for each subject to derive more diverse and accurate states. can make it possible

For example, it will be possible to derive a separate subject's health state by comparing the movement progress speed among the movement information of the indoor subject with the existing same movement progress speed.

In addition, it will be possible to derive a separate subject's health status by comparing the occurrence time of a specific action (wake up, etc.) among the movement information of the indoor subject with the occurrence time of the same behavior.

In addition, it will be possible to derive a separate subject's health state by comparing the detailed motion of a specific action among the motion information of the indoor subject with the detailed motion of the same action.

For this purpose, the control terminal 300 or a separate server may be equipped with an algorithm for machine learning of images, and detailed health conditions analyzed and output here are collected and set in the communication company server 400, client terminals of guardians. (500) can be made to be notified of the real-time situation.

Next, an indoor human accident detection system and a driving method thereof according to a second embodiment of the present invention will be described in detail with reference to FIGS. 6 to 14 .

First, FIG. 6 is a view for explaining an indoor human accident detection system according to a second embodiment of the present invention.

Referring to FIG. 6 , the indoor human accident detection system according to the second embodiment of the present invention includes a sensor module 600 installed indoors to photograph a subject, and a subject by analyzing the image captured by the sensor module 600 . Health to generate and notify care information for a subject by inferring the subject's health status based on the image monitoring module 700 for monitoring the activity of the subject and the activity monitoring result transmitted from the image monitoring module 700 The care module 800 may be included.

The indoor human accident detection system according to the second embodiment of the present invention is composed of a low-resolution image sensor and a sensor module 600 having only a simplified communication function is easily installed in an indoor area of interest, and a low-resolution image of a subject to be monitored within the area of interest. is taken, and the image is analyzed by the image monitoring module 700 to derive monitoring values for the posture of the person, the person's clothes, or the movement speed (average speed, movement direction) of the person, such monitoring values It is characterized by inferring the health status of the subject to be monitored through inference based on the information and notifying the designated caregiver of appropriate care information.

7 is a view for explaining various configuration examples of the indoor human accident detection system according to the second embodiment of the present invention.

Subjects who need the provision of an actual indoor human accident detection system clearly show physical characteristics with a lower than a certain level of exercise ability and a narrow range of activity, and in addition, the problem of lack of understanding and utilization of electronic devices is a problem with such indoor monitoring equipment. It is a major obstacle to the installation and operation of Therefore, in the present invention, the sensor module 600 installed indoors to photograph a subject is composed of a low-resolution image sensor and a small module having only a simplified communication function, so that even the elderly who cannot handle electronic devices can be installed and operated quickly and easily. It has its characteristics to make it happen.

For example, only the sensor module 600 of a simplified configuration is installed in the house as shown in FIG. By installing and operating the service, it is possible to reduce the installation difficulty and maintenance burden of the installer of the service for the elderly.

Further, as in Fig. 7 (b), in the house, only the sensor module 600 of a simplified configuration and the image monitoring module 700 for monitoring the activity of the subject by analyzing the image transmitted from the module are installed and , the health care module 800 connected to the network may be installed and operated outdoors or in an external concentration center. In this case, since only the monitoring value is transmitted through the network instead of directly transmitting the captured image to the outside in real time through the network, it will be possible to reduce the communication load and avoid the cost increase due to the application of the high-speed communication method.

Also, it is possible to install all of the sensor module 600 , the image monitoring module 700 , and the health care module 800 in the house as shown in FIG. 7C . In this case, faster and more accurate parental notification of care information will be possible.

Here, the sensor module 600 , the image monitoring module 700 , and the health care module 800 may be connected through a wired/wireless communication network.

In addition, a plurality of the sensor modules 600 may be installed at various locations in the indoor space in which the subject is active, and each of the sensor modules 600 will photograph a designated area of interest.

In particular, the sensor module 600 can operate with a battery without a separate external power source, and due to such a self-powered method, physical restrictions on the installation location of the sensor module 600 are reduced, and installation convenience can also be improved. .

Referring back to FIG. 6 , the sensor module 600 includes an image sensor 610 that acquires image data for a region of interest by photographing a designated region of interest, and an image monitoring module for image data transmitted from the image sensor 610 . The communication unit 620 for transmitting to 700 may be included.

The image data captured by the image sensor 610 may be photographed in units of one image frame, such as a still image, but is preferably photographed in units of a plurality of image frames, such as moving images. In addition, image data obtained by photographing each pixel of the image sensor 610 may be transmitted to and stored in a separate memory unit (not shown), and the photographed image data may be transmitted to an image monitoring module through the communication unit 620 . It is transmitted in real time to 700 and used as basic data of image analysis for monitoring target activity of the image monitoring module 700 .

Since the detailed circuit configuration of the image sensor 610 has been described in detail with reference to FIGS. 3 and 4 above, it will be omitted.

On the other hand, the image monitoring module 700 receives the image data transmitted in real time from the sensor module 600, performs image analysis, and an image monitoring analysis unit 710 that outputs monitoring determination information on the activity of the subject in the image. And, the image monitoring analysis unit 710 by receiving the image data transmitted from the determination criterion setting unit 720 and the sensor module 600 for storing and managing the determination criterion information for the monitoring determination of the image monitoring and analysis unit 710 . ) and may include a communication unit 730 that transmits monitoring determination information transmitted from the image monitoring and analysis unit 710 to the health care module 800 .

The image monitoring and analysis unit 710 analyzes real-time image data captured from the sensor module 600 to extract a context inference image related to the subject's health status from the subject's activity, and based on the criterion for determining the context inference image Monitoring identification information is generated.

The context inference image is an image capable of inferring the health status of the subject from the image including the subject, and in this context inference image, the image monitoring and analysis unit 710 generates monitoring discrimination information according to a predetermined criterion.

In the indoor human accident detection system according to the second embodiment of the present invention, the image monitoring module 700 performs circumstantial inference through one of the following three methods, and can check the health status of the subject here. It will produce monitoring identification information.

Context inference according to the first method identifies the posture the subject is currently taking in the image containing the subject and counts the holding time of the identified posture to generate the posture identification value and the holding time information as monitoring determination information. will be

With the recent acceleration of nuclear family and aging, the number of elderly living alone living alone without a person obligated to provide assistance is increasing. The so-called loneliness problem for the elderly living alone, in which precious lives are lost or, worse, left unattended for a long time after death, is emerging as a social issue.

In general, the elderly with physical or mental health problems tend to take more lying or squatting postures than standing or sitting positions. There will be no difficulty in determining, and timely notification of such a health risk situation to external guardians will be of great help in health care for the elderly who are not present.

Accordingly, in the context inference according to the first method, the image monitoring and analysis unit 710 identifies the body posture that the subject is currently taking in the image including the subject transmitted in real time from the sensor module 600 and maintains the identified posture. By counting the time, the posture identification value and its holding time information will be generated as monitoring determination information. At this time, the posture identification reference information for posture identification will be managed by the determination criterion setting unit 720 as determination reference information.

8, in the context inference according to the first method, the image monitoring analysis unit 710 includes a person detection unit 711a that detects a person who is a subject from an image including the subject transmitted in real time from the sensor module 600, and; The posture identification unit 711b that identifies the body posture that the person detected by the person detection unit 711a is currently taking, and the time that the posture identified by the posture identification unit 711b is maintained is counted to determine the posture maintenance time Inference information for outputting the calculated posture monitoring unit 711c, the posture identification value of the posture identification unit 711b, and the posture maintenance time information of the posture monitoring unit 711c as monitoring determination information capable of inferring a health state It may be configured to include a generator 711d.

Here, the posture identification unit 711b will be able to identify the posture that the subject is currently taking through the first posture identification reference information and the second posture identification reference information managed by the determination criterion setting unit 720 .

The first posture identification reference information is an identification criterion using the direction of the major axis and the minor axis that can be derived from the person area (area occupied by the person in the image) detected by the person detection unit 711a.

In more detail, the first posture identification reference information includes reference information for identifying the posture of the person as a stand when the long axis of the person area corresponds to the vertical axis of the image. In addition, the first posture identification reference information includes reference information for identifying the posture of the person as a lying posture when the long axis of the person area corresponds to the horizontal axis of the image.

Usually, the sensor module 600 for photographing a subject is installed so that the ground becomes the horizontal axis of the image, and thus, if the long axis of the person area corresponds to the vertical axis, it is determined as a standing posture, and the long axis of the person area is the horizontal axis and the horizontal axis. If it is matched, there will be no big crowd or identification error in judging it as a lying position.

Here, since the person does not always maintain an upright posture parallel to the vertical axis of the image, and the lying posture may also be skewed from the horizontal axis of the image, the first posture identification reference information can be supplemented more precisely. That is, the first posture identification reference information may include reference information for identifying the posture of the person as a stand when the long axis of the person area corresponds to the vertical axis of the image within an angle of ±44°. In addition, the first posture identification reference information may include reference information for identifying the posture of the person as a lying posture when the long axis of the person area corresponds to the horizontal axis of the image within an angle of ±44°.

Meanwhile, the second posture identification reference information is an identification criterion using a ratio of the length of the major axis to the minor axis that can be derived from the person area (area occupied by the person in the image) detected by the person detection unit 711a.

In more detail, in the second posture identification reference information, the ratio of the length of the major axis to the length of the minor axis is 3:1 or less, that is, the length of the major axis is the length of the minor axis with respect to the posture identified as the lying posture according to the first posture identification reference information. It includes reference information to identify it as a crouched posture if it does not exceed 3 times the comparison. In general, the length of the major axis of a person is more than three times the length of the minor axis when a person is lying upright. As such, the ratio of the length of the major axis to the length of the minor axis is 3:1 or less, that is, the length of the major axis is three times the length of the minor axis. If it cannot be crossed, there will be no big crowd or identification error in judging it as a crouched posture.

Referring to FIG. 9 together, (a), (b) and (c) of FIG. 9 are images including a subject transmitted in real time from the sensor module 600, and the person detecting unit 711a in the image including the subject ) as shown in (a), (b) and (c) of FIG. 9 detects a person who is a subject.

Here, in Fig. 9 (a), the person is taking a standing posture, (b) is taking a lying posture, (c) is taking a crouching posture.

In the image in which the subject person is detected, the posture identification unit 711b first refers to the first posture identification reference information managed by the determination criterion setting unit 720, and as shown in FIG. 9(a), the long axis L of the person area. ) and the minor axis (S), and if the major axis (L) of this person area corresponds to the vertical axis (V) of the image (more precisely, if it corresponds within an angle of ±44° with the vertical axis of the image), the current person’s posture is determined. The standing posture is identified, and accordingly, the posture monitoring unit 711c counts the time the identified standing posture is maintained to calculate the posture holding time.

On the other hand, in the image in which the subject, the person, is detected, the posture identification unit 711b refers to the first posture identification reference information managed by the determination criterion setting unit 720, and as shown in FIG. 9(b), the long axis ( L) and minor axis (S) are identified, and if the major axis (L) of this person area corresponds to the horizontal axis (H) of the image (more precisely, if it corresponds within an angle of ±44° with the horizontal axis of the image), the current person’s posture is identified as a supine posture, and accordingly, the posture monitoring unit 711c counts the time the identified supine posture is maintained to calculate the posture maintenance time.

After the determination of the lying posture, the posture identification unit 711b refers to the second posture identification reference information managed by the determination criterion setting unit 720, as shown in FIG. When the length of the minor axis (S) is identified and the ratio of the length of the major axis to the length of the minor axis is 3:1 or less for the posture identified as lying down, that is, when the length of the major axis does not exceed 3 times the length of the minor axis, it is identified as a crouched posture Accordingly, the posture monitoring unit 711c calculates the posture holding time by counting the time the identified crouching posture is maintained.

Accordingly, the inference information generating unit 711d outputs the posture identification value of the posture identification unit 711b and the posture maintenance time information of the posture monitoring unit 711c as monitoring determination information that can infer a health state, and this The same monitoring determination information will be transmitted to the health care module 800 .

Meanwhile, the situational inference according to the second method identifies the subject's clothes in the image containing the subject and counts the wearing period of the identified clothes, thereby generating the clothing identification value and the wearing period information as monitoring discrimination information. will be.

In general, the elderly who have physical or mental health problems have a great burden on the act of changing clothes, and accordingly, the number of times of changing clothes rapidly decreases as the elderly with poor health and mobility are poor. If you wear one piece of clothing for a long time without changing clothes, it will not be too difficult to determine that it is a health risk situation. will be of great help to

Accordingly, in the context inference according to the second method, the image monitoring and analysis unit 710 identifies the clothes currently worn by the subject in the image including the subject transmitted in real time from the sensor module 600, and maintains the identified clothes. By counting (wearing time), the clothing identification value and the wearing period information will be generated as monitoring identification information. At this time, information on the shape of clothes and the color of clothes for identification of clothes will be managed by the discrimination criteria setting unit 720 as discrimination criteria information.

10, in the context inference according to the second method, the image monitoring analysis unit 710 includes a person detection unit 712a that detects a person who is a subject from an image including the subject transmitted in real time from the sensor module 600, and; Clothes for calculating the wearing period by counting the wearing period of the clothes identified by the wearing clothes identification unit 712b for identifying the currently worn clothes of the person detected by the person detecting unit 712a, and the wearing clothes identification unit 712b A monitoring unit 712c, an inference information generating unit that outputs the clothing identification value of the wearing clothes identification unit 712b and the wearing period information of the clothes monitoring unit 712c as monitoring determination information capable of inferring a health condition ( 712d) may be included.

Here, the worn clothes identification unit 712b will be able to identify the subject's wearing clothes through the clothing color and clothing shape information managed by the determination criterion setting unit 720, and the determination criterion setting unit 720 Information on the color and shape of clothes can be written according to the general clothes classification rules, but can be written as customized clothes classification rules by the guardian according to the actual clothes worn by the subject.

Referring to FIG. 11 together, (a) of FIG. 11 is an image including a subject transmitted in real time from the sensor module 600 .

In the image including the subject as described above, the person detection unit 712a detects the person who is the subject as shown in FIG. 11( b ).

In addition, from the image in which the person, the subject, is detected, the clothes identification unit 712b identifies the currently worn clothes of the person as shown in FIG. 11C . Here, the upper and lower garments may be identified separately, or the upper and lower garments may be identified together.

At this time, information on the shape of clothes and the color of clothes for identification of clothes will be managed and provided by the discrimination criterion setting unit 720 as discrimination criterion information.

Accordingly, the clothes monitoring unit 712c will be able to calculate the subject's clothes wearing period by counting the identification maintenance period of the identified clothes, that is, the wearing period of the identified clothes.

As a result, the inference information generating unit 712d outputs the clothing identification value of the worn clothes identification unit 712b and the wearing period information of the clothes monitoring unit 712c as monitoring determination information that can infer the health condition, The same monitoring determination information will be transmitted to the health care module 800 .

On the other hand, context inference according to the third method identifies the movement of the subject in the image containing the subject and calculates the movement direction and speed of the identified movement, thereby generating movement path and movement speed information as monitoring determination information. will do

In general, the elderly who have physical or mental health problems often move to a different non-routine route than usual, and there is a big difference in the speed of their movement compared to the normal case. If this occurs or an unusual movement speed appears, it will not be too difficult to determine that it is a health risk situation. will be.

Accordingly, in the context inference according to the third method, the image monitoring and analysis unit 710 calculates the movement path and movement speed of the subject from the image including the subject transmitted in real time from the sensor module 600, thereby determining the movement path as monitoring determination information. and movement speed information. In this case, the information on the intra-image path calculation formula and the intra-image speed calculation formula for calculating the moving path and the moving speed will be managed by the discrimination criterion setting unit 720 as discrimination criterion information.

12, in the context inference according to the third method, the image monitoring and analysis unit 710 detects a person who is a subject and an object that is a moving destination of the subject from the image including the subject transmitted in real time from the sensor module 600 A person and object detection unit 713a, a movement identification unit 713b for recognizing a movement movement of a person detected by the person and object detection unit 713a to a target, and a movement movement identified by the movement identification unit 713b Inference information for outputting the movement monitoring unit 713c for calculating the movement path and movement speed of the subject, and the movement path and movement speed information of the movement monitoring unit 713c as monitoring determination information capable of inferring the health condition It may be configured to include a generator 713d.

In this case, the target will be divided into an ordinary target and an non-routine target. The everyday object exists in the image and may be an object (eg, a sofa, etc.) that the subject accesses on a daily basis. In addition, the non-routine object may be an object (eg, washing machine, etc.) that exists in the image and is not accessed by the subject on a daily basis.

Here, the information on the everyday object and the non-routine object existing in the image will be managed by the determination standard setting unit 720 as the determination standard information.

Referring to FIG. 13 together, (a) of FIG. 13 is an image including a subject transmitted in real time from the sensor module 600 .

As shown in (b) of FIG. 13 , in the image including the subject, the person and object detection unit 713a detects a person who is a subject and an object (eg, a chair and a washing machine) that is a moving destination of the subject. Here, the object as the moving destination may include at least one or more of an everyday object (eg, a chair) and a non-routine object (eg, a washing machine).

The information on the everyday object and the non-routine object existing in the image will be managed by the determination standard setting unit 720 as the determination standard information.

In addition, the movement identification unit 713b recognizes the movement of the detected person to the target as shown in (c) or (d) of FIG. do.

Accordingly, the movement monitoring unit 713c will calculate the movement path and movement speed of the subject for the identified movement movement.

Accordingly, the inference information generation unit 713d outputs the movement path and movement speed information of the movement monitoring unit 713c as monitoring determination information capable of inferring a health state, and such monitoring determination information is transmitted to the health care module 800 ) will be transferred.

On the other hand, the health care module 800 receives the monitoring determination information transmitted from the image monitoring module 700, and generates care information for the subject by inferring the subject's health state through inference based on the monitoring determination information. Parents will be notified.

The health care module 800 receives the monitoring determination information transmitted in real time from the image monitoring module 700 and compares the monitoring determination information for each subject with the care reference information set in advance for the subject to infer the subject's health status. A care information processing unit 810 that generates care information for a subject by inferring through Receives monitoring determination information transmitted from the monitoring module 700 and delivers it to the care information processing unit 810 , and notifies the care information transmitted from the care information processing unit 810 to a preset communication terminal (eg, a smartphone) of the guardian The communication unit 830 may be included.

The care information processing unit 810 generates customized care information by inferring the health status of the subject according to the individual criteria of the subject with the monitoring determination information transmitted from the image monitoring module 700 .

In addition, the care standard setting unit 820 provides care reference information serving as an inference criterion so that the care information processing unit 810 can infer the health status of the subject.

The care reference information may be set according to the three context inference methods according to the second embodiment of the present invention.

In the context inference according to the first method, the image monitoring module 700 identifies the posture the subject is currently taking in the image containing the subject and counts the maintenance time of the identified posture as monitoring determination information, the posture identification value and the holding time information will be generated and delivered to the health care module 800 .

Accordingly, the care standard information of the care standard setting unit 820 may be individually set by the guardian in consideration of various factors such as the subject's age, gender, medical history, living space, and usual posture habits, and the standard of the lying posture The duration and corresponding necessary care information, the reference duration of the crouched posture and necessary care information accordingly, the standard duration of the standing posture and necessary care information accordingly may be included.

Based on this, the care information processing unit 820 receives monitoring determination information including the posture identification value and the holding time information from the image monitoring module 700, and the care standard setting unit 820 through the holding time of the posture. After selecting the standard duration for each posture included in the care standard information of

And the care information processing unit 820 notifies the care information including the extracted necessary care information to the communication terminal of the guardian set in advance through the communication unit 830, and customized care information for the current health status of the subject identified through the posture will be able to be communicated to the guardian in real time.

In addition, in the context inference according to the second method, the image monitoring module 700 identifies the subject's wearing clothes in the image including the subject and counts the wearing period of the identified clothes. generated and delivered to the health care module 800 .

Accordingly, the care standard information of the care standard setting unit 820 may be individually set by the guardian in consideration of various factors such as the subject's age, gender, medical history, living space, and usual posture and habits, and the criteria for identification clothes Duration and consequent care needs information may be included.

Based on this, the care information processing unit 820 receives the monitoring determination information including the clothing identification value and the wearing period information from the image monitoring module 700, and the care standard setting unit 820 through the maintenance time of the corresponding clothing. After selecting the standard duration for each identification garment included in the care standard information, it will be possible to extract necessary care information.

And the care information processing unit 820 notifies the care information including the extracted necessary care information to the communication terminal of the guardian set in advance through the communication unit 830, and customized care information for the current health condition to the subject identified through the clothes. will be able to be communicated to the guardian in real time.

In addition, in the context inference according to the third method, the image monitoring module 700 identifies the movement of the subject in the image containing the subject, and calculates the movement direction and speed of the identified movement, so that the movement path and movement speed as monitoring determination information. Information will be generated and transmitted to the health care module 800 .

Accordingly, the care standard information of the care standard setting unit 820 may be individually set by the guardian in consideration of various factors such as the subject's age, gender, medical history, living space, and usual posture habits, and the movement route (daily routine). target object and non-routine object) and necessary care information, movement speed and necessary care information accordingly.

Based on this, the care information processing unit 820 receives the monitoring determination information including the movement path (normal object and non-routine object) and movement speed information from the image monitoring module 700, and sets the care standard through the movement path It will be possible to extract necessary care information according to the movement path included in the care reference information of the unit 820 and to extract necessary care information according to the movement speed.

And the care information processing unit 820 notifies the care information including the extracted necessary care information to the communication terminal of the guardian set in advance through the communication unit 830, and customized care for the current health condition to the subject identified through the movement movement Information will be communicated to the caregiver in real time.

By introducing machine learning or deep learning to the health care module 800 , it is possible to increase the accuracy of situational inference through the above-described monitoring identification information and extraction of necessary care information accordingly. Through continuous learning of various and extensive subject monitoring and care information, the health care module 800 will be able to accurately check even minute changes in the subject's behavior, and will be able to significantly reduce the frequency of data interpretation errors.

14 is a view for explaining a driving method of an indoor human accident detection system according to a second embodiment of the present invention.

Referring to FIG. 14, first, the sensor module 600 composed of a low-resolution image sensor is easily installed in an indoor area of interest by a guardian or a person to be cared for, and takes a low-resolution image of a subject to be monitored within the area of interest ( S10).

The image monitoring module 700 that receives the image data transmitted in real time from the sensor module 600 performs image analysis (S20), and generates monitoring determination information on the activity of the subject in the image (S30).

Here, the monitoring determination information is generated by identifying the posture the subject is currently taking through the image including the subject in the image monitoring module 700 for context inference according to the first method, and counting the maintenance time of the identified posture It may be information about the posture identification value and its holding time.

In addition, the monitoring determination information is generated by identifying the subject's wearing clothes through the image including the subject in the image monitoring module 700 for context inference according to the second method and counting the wearing period of the identified clothes. It may be information about a value and its wearing period.

In addition, the monitoring determination information identifies the movement movement of the subject through the image including the subject in the image monitoring module 700 for context inference according to the third method, and calculates the movement direction and speed of the identified movement movement. It may be generated movement path and movement speed information.

Such monitoring determination information is transmitted to the health care module 800, and the health care module 800 generates care information for the subject by inferring the subject's health state through inference based on the transmitted monitoring determination information ( S40), the generated care information is notified to the designated guardian (S50).

As described above, the best embodiment has been disclosed in the drawings and the specification. Although specific terms are used herein, they are used only for the purpose of describing the present invention and are not used to limit the meaning or the scope of the present invention described in the claims. Therefore, those skilled in the art will understand that various modifications and equivalent other embodiments are possible therefrom. Therefore, the true technical protection scope of the present invention should be determined by the technical spirit of the appended claims.

100: human body detection sensor 200: gateway
300: control terminal 400: telecommunication company server
500: client terminal 600: sensor module
700: video monitoring module 800: health care module

Claims (17)

a sensor module installed indoors to photograph a subject;
an image monitoring module for monitoring the activity of the subject by analyzing the image captured by the sensor module; and
a health care module for generating and notifying care information for a subject by inferring the subject's health state through inference based on the activity monitoring result transmitted from the image monitoring module; Indoor fatal accident detection system comprising a.
The method of claim 1,
The sensor module is
an image sensor that captures a designated region of interest to acquire image data for the region of interest; and
a communication unit transmitting the image data transmitted from the image sensor to the image monitoring module; includes,
The image data captured by the image sensor is photographed in units of one image frame, such as a still image, or in units of a plurality of image frames, such as a moving image.
The method of claim 1,
The video monitoring module,
an image monitoring analysis unit for receiving image data transmitted in real time from the sensor module, performing image analysis, and outputting monitoring identification information on the subject's activity in the image;
a determination standard setting unit for storing and managing determination reference information for monitoring determination of the image monitoring and analysis unit; and
a communication unit that receives the image data transmitted from the sensor module and transmits the image data to the image monitoring analysis unit and transmits the monitoring determination information transmitted from the image monitoring analysis unit to the health care module; Indoor fatal accident detection system comprising a.
4. The method of claim 3,
The video monitoring analysis unit,
Indoor fatal accident detection system, characterized in that by identifying the posture the subject is currently taking from the image containing the subject and counting the maintenance time of the identified posture, the posture identification value and the holding time information are generated as monitoring discrimination information.
5. The method of claim 4,
The video monitoring analysis unit,
a person detection unit for detecting a person who is a target in an image including the target transmitted in real time from the sensor module;
a posture identification unit for identifying a body posture currently taken by the person detected by the person detecting unit;
Posture monitoring unit for calculating the posture holding time by counting the time that the posture identified by the posture identification unit is maintained; and
an inference information generation unit for outputting the posture identification value of the posture identification unit and the posture maintenance time information of the posture monitoring unit as monitoring determination information capable of inferring a health state; Indoor fatal accident detection system comprising a.
6. The method of claim 5,
The determination criterion setting unit manages first posture identification information, which is an identification criterion using the direction of the major axis and the minor axis that can be derived from the person area of the person detected by the person detection unit,
The first posture identification reference information includes reference information for identifying the posture of the person as a standing posture when the long axis of the person area corresponds to the vertical axis of the image, and reference information for identifying the posture of the person as a standing posture when the long axis of the person area corresponds to the horizontal axis of the image. Indoor fatal accident detection system, characterized in that it includes reference information to identify the posture.
7. The method of claim 6,
The posture identification unit identifies the long axis and the short axis of the person area with reference to the first posture identification reference information managed by the determination criterion setting unit,
If the long axis of this person area corresponds to the vertical axis of the image within an angle of ±44°, the current person's posture is identified as a standing posture, and the posture monitoring unit counts the time the identified standing posture is maintained to calculate the posture maintenance time and
Alternatively, if the long axis of this person area corresponds within an angle of ±44° with the horizontal axis of the image, the current person's posture is identified as a supine posture, and the posture monitoring unit counts the time the identified supine posture is maintained to determine the posture maintenance time. Indoor casualty detection system, characterized in that the calculation.
8. The method of claim 7,
The discrimination criterion setting unit manages second posture identification information, which is an identification criterion using the ratio of the length of the long axis and the short axis that can be derived from the person area of the person detected by the person detection unit,
The second posture identification reference information identifies the posture of the person lying down as a crouched posture when the length of the major axis does not exceed three times the length of the minor axis with respect to the posture identified as the supine position according to the first posture identification reference information Indoor human accident detection system, characterized in that it includes reference information to be made.
9. The method of claim 8,
After determining the lying posture, the posture identification unit identifies the major axis length and the minor axis length of the person area with reference to the second posture identification reference information managed by the determination criterion setting unit,
If the length of the major axis does not exceed three times the length of the minor axis for the posture identified as lying down, the posture of the person lying down is identified as a crouched posture, and the posture monitoring unit counts the time the identified crouching posture is maintained. Indoor fatal accident detection system, characterized in that calculating the posture maintenance time.
4. The method of claim 3,
The video monitoring analysis unit,
An indoor human accident detection system, characterized in that the clothes identification value and the wearing period information are generated as monitoring discrimination information by identifying the subject's wearing clothes from the image containing the subject and counting the wearing period of the identified clothes.
11. The method of claim 10,
The video monitoring analysis unit,
a person detection unit for detecting a person who is a target in an image including the target transmitted in real time from the sensor module;
a clothing identification unit for identifying the currently worn clothing of the person detected by the person detection unit;
a clothes monitoring unit calculating a wearing period by counting the wearing period of the clothes identified by the wearing clothes identification unit; and
an inference information generating unit for outputting the clothing identification value of the wearing clothes identification unit and the wearing period information of the clothes monitoring unit as monitoring identification information capable of inferring a health state; Indoor fatal accident detection system comprising a.
12. The method of claim 11,
The indoor casualty detection system, characterized in that the wearing clothes identification unit identifies the subject's wearing clothes through the clothing color and clothing shape information managed by the determination criterion setting unit.
4. The method of claim 3,
The video monitoring analysis unit,
An indoor human accident detection system, characterized in that it generates movement path and movement speed information as monitoring identification information by identifying the movement movement of the object in the image containing the object and calculating the movement direction and movement speed of the identified movement movement.
14. The method of claim 13,
The video monitoring analysis unit,
a person and object detection unit that detects a person who is a target person and a target that is a moving destination of the target from the image including the target transmitted in real time from the sensor module;
a movement identification unit for recognizing the movement of the person to the object detected by the person and the object detection unit;
a movement monitoring unit for calculating the movement path and movement speed of the subject for the movement movement identified by the movement identification unit; and
an inference information generation unit for outputting the movement path and movement speed information of the movement monitoring unit as monitoring determination information capable of inferring a health state; Indoor fatal accident detection system comprising a.
15. The method of claim 14,
The movement identification unit is an indoor human accident detection system, characterized in that it recognizes the movement movement for the everyday object and the non-routine object managed by the determination criterion setting unit separately.
The method of claim 1,
The health care module,
Receives monitoring determination information delivered in real time from the image monitoring module and generates care information for a subject by inferring the subject's health status by inferring the subject's health status by comparing the monitoring determination information for each subject with the care standard information set in advance for the subject care information processing unit;
a care standard setting unit for storing and managing care reference information for inferring a health state of the care information processing unit; and
a communication unit for receiving the monitoring determination information transmitted from the image monitoring module and transmitting it to a care information processing unit, and notifying the care information transmitted from the care information processing unit to a communication terminal of a preset guardian; Indoor fatal accident detection system comprising a.
17. The method of claim 16,
The care standard information of the care standard setting unit includes the reference duration of the lying posture and necessary care information accordingly, the reference duration of the squatting posture and necessary care information accordingly, the standard duration of the standing posture and necessary care information accordingly, and identification clothes Indoor fatal accident detection system, characterized in that at least one of a star standard duration and corresponding necessary care information, moving route and corresponding necessary care information, moving speed and corresponding necessary care information are included.

Images