JPH05118613A - Life scene inferring device and air conditioner - Google Patents

Abstract

PURPOSE:To provide a device to infer life scene based on human information and environmental information drawn from sensor information with no manipulation of the user and provided an air conditioner which utilizes this device. CONSTITUTION:A pyroelectric type IR ray sensor 12 which is a human detection means and light sensor 13 which is an environment detection means are provided in a room 11, and thermal images in the room which are obtained by measuring the inside of the room with a certain time interval are processed by a feature drawing means 14 to obtain as feature quantities the number of persons who are in the room, their positions, postures and action quantities with respect to the floor surface. From those feature quantities a candidate for life scene is inferred by candidate life scene inferring means 17. A life scene determination means 20 judges whether or not life scene is determined by counting the temporal continuity of the candidate life scene that is the output of a candidate life scene inferring means 17. With this arrangement human information with little error can be obtained, and the life scene inference that is not reactive too sensitively to transitory life actions such as movements in the room in a short time but nearer the reality can be made.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for automatically switching the driving method for each life scene in the control of equipment for providing a thermal environment, lighting, sound, scent, etc. for home use. The present invention relates to a scene estimation device and an air conditioner using the same.

[0002]

2. Description of the Related Art In recent years, particularly in the field of air conditioners and the field of lighting equipment, equipment that provides an environment suited to the indoor living scene has been commercialized, but most of them are switching operation for each living scene. Was performed by the user's switch operation. Also, using an infrared sensor,
A device for detecting the movement of a hot object in a room and operating the device by considering it as entering a resident, and Japanese Patent Laid-Open No. 2-4910.
As described in Japanese Unexamined Patent Publication No. 2 (1998), the position of a person in the room is detected by using an infrared sensor that divides the room into a plurality of areas and detects the movement of a heat object in each area, and then moves toward that position. There was an air conditioner that regulated the wind direction.

[0003]

However, in the conventional method of switching the life scene only by the operation of the user, the device has to be operated every time the life scene changes in the daily life, and the user can use it. As a result, it has become a troublesome person, and as a result, it has not been used so much, and there is a problem in that even if a driving method for each life scene is prepared for the device, it is not fully utilized. In addition, if the infrared sensor detects only the movement of a hot object and starts and stops the device, it will start and stop the device in response to objects other than humans, or detect when the human is stationary. There was a problem that was not done. Furthermore, the one that detects the position of the person in the room and controls the device is
There is no distinction between life scenes such as whether the person is sleeping, being in a group, or just getting up.Therefore, switching the driving method based on these detection information is only a control method centered on the occupant's whereabouts. There was a problem of not having.

The present invention solves such a conventional problem, and automatically changes the life scene based on the number of people in the room, position, posture, movement, and environmental changes over time. An object is to provide an excellent life scene estimation device that can be estimated and an air conditioner using the same.

[0005]

In order to achieve the above object, the life scene estimating apparatus of the present invention comprises a person detecting means for detecting a state of a person in a room, and an environment detecting means for detecting an indoor environment. A feature extraction means for extracting a feature amount based on these detection information, a candidate life scene estimation means for estimating a life scene that is a candidate from the obtained feature amount, and a life scene determination means for determining a life scene. It is equipped with.

[0006]

With the above structure, the life scene estimating apparatus of the present invention first measures the occupants in the room and the environmental condition by the human detecting means and the environment detecting means a plurality of times within a unit of processing for a fixed time. .. Next, by the feature extraction means, the presence / absence of persons in the room, the number of persons, the position / posture, and the amount of activity with respect to the occupants as the feature amount, and the detected environmental information regarding the environment are the intentions regarding the room use of the occupants. Extract whether or not the appearance of. The candidate life scene estimation means estimates a life scene to be a candidate based on the feature amount obtained in the current processing unit and the time change of the feature amount of the plurality of processing units. In the life scene confirmation means, the continuity of the candidate life scene and the difference from the current life scene are judged, and the candidate life scene is adopted.
In addition to deciding not to be adopted, a life scene showing the state change of the occupant such as entering, leaving, waking up and going to bed is determined.

[0007]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows the configuration of a life scene estimation device according to an embodiment of the present invention. In FIG. 1, 11 is a room to be estimated, 12 is a pyroelectric infrared sensor device for measuring a thermal image of the room, 13 is an optical sensor device for detecting light in the room, and 14 is feature extraction means. The thermal image processing unit 15 and the environmental knowledge conversion unit 16 that determines whether the light in the room is due to illumination or sunlight. Reference numeral 17 denotes a candidate life scene estimation means, which is a displacement amount calculation unit 18 for calculating the time displacement of the feature amount obtained by the thermal image processing, and a candidate life scene estimation unit 19.
Have and. Reference numeral 20 is a life scene determining means for estimating the life scene based on the temporal continuity of the candidate life scene and the property of the candidate scene. Reference numeral 25 is a clock / calendar generating means for obtaining the month, the day of the week and the time, and the feature extracting means 1
4 gives time information including the season to the candidate life scene estimation means 17 and the life scene determination means 20.

FIG. 2 shows the structure of the life scene determining means 20. In FIG. 2, 21 is a candidate life scene storage unit that stores the history of the candidate life scenes output from the candidate life scene estimation unit 17 every M seconds, and 22 is the duration only when the candidate life scenes are the same continuously. A counter for counting, 23 is a candidate life scene, a current life scene, and adoption of a candidate life scene from the value of the counter 22,
Life scene determination means for deciding not to be adopted, and 24 is a life scene change determination means for determining whether or not the change of the candidate life scene corresponds to a meaningful state change life scene.

Next, the operation of the life scene estimating apparatus configured as described above will be described. In the following description of the operation, one processing unit time from the measurement of the room to the determination of the life scene is M seconds, and the processing unit M is
Let K be the number of thermal images captured during the period. In FIG. 1, first, in the processing unit, first, K thermal images of the room 11 are continuously captured using the infrared sensor device 12.
Moreover, the lighting state of the light in the room when the thermal image is captured is measured using the optical sensor device 13. Next, still image processing is performed on each of the K thermal images captured by the thermal image processing unit 15, and the number of people in the room, the position with respect to the floor, and the posture are extracted. Here, the number of people is 0, 1, 2, 3 or more. Also, the position with respect to the floor is
The room is divided into 9 equal parts, and the position where the person in the room is located is extracted. There are three categories of posture: lying (lying), standing (sitting on a chair or floor), and standing (standing). Next, the activity amount of the person in the room is obtained from the change in the position of the person in the room with respect to the floor surface obtained from two consecutive thermal images of the captured thermal images. In addition, the size of the room for determining the reference floor surface is extracted from the thermal image as environmental information. Next, the environmental knowledge conversion unit 16 determines the lighting use state of the room at night from the time and the lighting state of the light.

Since the human information feature-extracted by the thermal image processor 15 is only for the processing unit time, it is too short to estimate the life scene from this feature amount alone. Therefore, the displacement amount calculation unit 18 calculates a representative value in consideration of the time change of the feature amount of S processing units. Here, the time for calculating the displacement amount is F = S × M seconds. The representative value here is the degree of fluctuation X of the number of people, the position stay degree Y indicating at which position in the room the person in the room often stays, and a period indicating the degree of activity of each person in the room. The amount of activity Z. The calculation method of each representative value and the meaning in life scene estimation are shown below. The number-of-people fluctuation X is the sum of the absolute values of the displacement of the number of people in the room between each successive processing unit divided by the sum of the absolute values of the maximum number of people fluctuation that can occur. The high degree of fluctuation X of the number of people means that the living scene in the room is a transitional state accompanying movement of the room or an active living scene. Moreover, it can be said that the number of people coming in and going out is small because the degree of fluctuation X of the number of people is low. Next, the position staying degree Y divides the room into 9 equal parts and arranges this into an array YA.
Corresponding to [9], the position of the person in the room obtained by the still image processing is counted in each processing unit in this array, and the position exceeding the predetermined threshold value is extracted. Like the array YA, the extracted location is an array Y that has a corresponding relationship with the floor position.
Set to [9]. The position where the occupant stays for a long time is considered as the room, and the array Y is used for the judgment. The period activity amount Z has the maximum activity amount of S processing units as a representative value. It can be said that a high period activity amount Z is an active life scene indoors, and a low period activity amount Z is a static life scene.

Next, in the candidate life scene estimation unit 19, from the displacement amount obtained by the displacement amount calculation unit 18, the feature amount obtained by the thermal image processing unit 15, and the variable obtained by the environmental knowledge making unit 16. , Estimate the life scenes that are candidates. The name of the target life scene and its meaning are shown in FIG.

Next, an algorithm for estimating the life scene shown in FIG. 5A will be described with reference to FIG. First, in step 31, it is determined whether the number of people during the time F is 0, or 1 or more. If 0, the candidate life scene is absent in step 32. If there is a person in the room, it is determined in step 33 whether or not the degree of fluctuation X in the number of people is larger than a predetermined threshold value. If the number of people variation X is high, the duration is counted in step 34. Whether the counter value is larger than a predetermined value or not is determined in step 3
In step 36, the candidate life scene is determined as active housework. If the value of the counter is small, it is considered as room movement in step 37, and the candidate life scene is transitional. On the other hand, when the degree of fluctuation X in the number of people in step 33 is smaller than the threshold value, it is determined in step 38 whether the illumination is on or off. If the light is not on, it is determined in step 39 whether the period activity amount Z is smaller than a predetermined threshold value. If larger,
In step 40, the candidate life scene is transitional. If it is smaller, it is determined in step 41 whether or not the posture is in the lying position.
If the subject is not in the decubitus position, the candidate life scene is relaxed in step 42. When lying down, consider sleeping. Step 3
Returning to step 8, if the light is on, it is determined in step 44 whether the period activity amount Z is smaller than a predetermined threshold value. If it is larger, in step 45, it is determined whether or not there is a person with a high degree of position stay Y among the people in the room. If there is, the candidate life scene is gathered in step 46, and if it is not, active housework is performed in step 47. .. Returning to step 44, if the period activity amount Z is smaller than the threshold value, it is determined in step 48 whether or not the number of persons is one. When the number of persons is plural, in step 49, it is determined whether or not all of the occupants in the room have a high degree of positional stay Y, and if so, in step 50, the candidate life scene is made to be relaxing. If not, step 51 is to unite. Returning to step 48 again, if the number of people is one, it is determined in step 52 whether the position staying degree Y is high. If the position staying degree Y is low, the candidate life scene is transitioned in step 53. If the position staying degree Y is high, it is determined in step 54 whether or not the posture of the occupant is in the lying position. If the lying position is in the lying position, the candidate life scene is relaxed in step 55, and otherwise the learning is performed in step 56.

With the above algorithm, the candidate of the present life scene is obtained every M seconds based on the information mainly from the thermal image in the past F seconds. Here, the reason why I made a candidate is that transitional living activities such as changing the place of living in the room or going to the toilet for a short time are performed several times in the life scene. If the measurement timing coincides with the occupants in these transitional states, life scenes of opposite nature will be output in an unstable manner, and this will make use of the life scene estimation results for device control. In that case, the operation becomes unstable. Therefore, the life scene mainly estimated from the thermal image is considered as a candidate, and the life scene determining means 20 is used.
Then, the life scene is determined on the condition of the temporal continuity of the candidates.

FIG. 4 shows an algorithm in the life scene determining means 20. Here, the current time is t, the time in the processing unit before the temporary point is t−1, the candidate life scene is B,
The life scene output by the life scene determination means 20 is referred to as an output life scene A. Figure 5 shows the life scene that A can take.
As shown in (a), a state-like life scene that means a state in which the occupants continue to carry out similar life activities, and FIG.
As shown in (b), it is roughly divided into state change life scenes in which changes in state life scenes are meaningful. First, step 6
1, the candidate life scene output at time t is time t
It is determined whether the life scene in -1 is different. If they are the same, the output life scene A at time t is made the same as the output life scene A at time t-1 in step 62. If different, it is determined in step 63 whether the candidate life scene B at time t is the same as the stored candidate life scene B at time t-1. If different, the counter is cleared in step 64, and the output life scene A at time t is made the same as the output life scene A at time t-1 in step 62. If the same in step 63,
In step 65, a threshold value which is a duration condition for adopting the candidate life scene B as the output life scene A is set. At that time, based on the change relationship from the output life scene A to the candidate life scene B, the threshold values are made into a table in advance, and the threshold values are determined from this table. Then, in step 66, the duration of the candidate life scene B is counted. Next, at step 67, it is judged if the counter is larger than the threshold value determined at step 65. If it is smaller than the threshold value, candidate life scene B
It is determined that sufficient time has not passed until the job is adopted, and in step 62, the output life scene A at time t is time t-
It is the same as the output life scene A in 1. If the threshold value is exceeded in step 67, time t is set in step 68.
The candidate life scene B at is adopted as the output life scene A at time t, and the counter is cleared. And
In step 69, it is judged from the change of the output life scene A at time t-1 and the change of the output life scene A at time t whether or not it corresponds to the state change life scene. The corresponding conditions of the state change life scene are shown in FIG. Finally, in step 70, the candidate life scene B at time t is stored.

As described above, according to this embodiment, by providing the pyroelectric infrared sensor device 12 and the feature extracting means 14 which are the person detecting means, the number of people in the room, the position with respect to the floor, the posture, and the activity. Life scene candidate estimating means 17 for detecting the amount, detecting indoor light by an optical sensor device 13 as an environment detecting means, and estimating a life scene as a candidate from the obtained feature amount and time information, and a candidate. By providing the life scene determination means 20 that determines the life scene based on the temporal continuity of the life scene and the property of the candidate scene, it is possible to estimate a life scene that is closer to the actual situation. In addition, when the candidate life scene is determined, by setting the continuity of the time as a condition, it is unstable with respect to a short-lived transitional living activity of an occupant such as room movement that appears in a stable life scene. You can prevent it from becoming.

Further, by incorporating such a life scene estimation device in the control unit of the air conditioner, more detailed air conditioning control can be performed. In particular, this device is incorporated into the "air conditioner control device" described in the specification of Japanese Patent Application No. 3-216847 filed by the applicant of the present application, and the outputs of this device are respectively referred to as "operation content estimation means" and "learning". By inputting "means", more comfortable automatic air conditioning operation can be realized.

[0017]

As described above, according to the present invention, the person detecting means, the environment detecting means, the feature extracting means, the candidate life scene estimating means,
An excellent life scene estimation that can extract the detailed information of the people in the room and output a stable life scene considering the temporal continuity by providing the life scene decision means The device and the air conditioner using the device can be realized.

[Brief description of drawings]

FIG. 1 is a schematic block diagram showing the configuration of a life scene estimation device according to an embodiment of the present invention.

FIG. 2 is a schematic block diagram showing the configuration of a life scene estimating means in the embodiment.

FIG. 3 is a flowchart showing a candidate life scene estimation algorithm in the same embodiment.

FIG. 4 is a flowchart showing a life scene determination algorithm in the same embodiment.

5A is a list diagram showing names of candidate life scenes and their meanings in the same embodiment. FIG. 5B is a list diagram showing names of condition change life scenes and their conditions in the embodiment.

[Explanation of symbols]

 11 Indoors 12 Infrared sensor device (human detection means) 13 Optical sensor device (environment detection means) 14 Feature extraction means 15 Thermal image processing section 16 Environmental knowledge conversion section 17 Candidate life scene estimation means 18 Displacement amount calculation section 19 Candidate life scene estimation Part 20 Life scene confirmation means 21 Candidate life scene storage means 22 Counter 23 Life scene determination means 24 Life scene change determination means 25 Clock / calendar generation means

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Ikuo Akamine 1006 Kadoma, Kadoma, Osaka Prefecture Matsushita Electric Industrial Co., Ltd. (72) Shin Makoto Shimizu, 1006 Kadoma, Kadoma, Osaka Matsushita Electric Industrial Co., Ltd. Company (72) Inventor Katsuhiko Fujiwara 1006 Kadoma, Kadoma City, Osaka Prefecture Matsushita Electric Industrial Co., Ltd.

Claims (8)

[Claims] 1. A person detecting means for detecting human information in a room, an environment detecting means for detecting environment information in a room, and a feature extracting means for extracting and calculating a characteristic amount from the detected human information and environment information. And a candidate life scene estimation means for estimating a candidate life scene from the feature amount and time information obtained by the feature extraction means, and a life scene based on the temporal continuity of the candidate life scene and the nature of the candidate scene. A life scene estimation device comprising a life scene confirmation means for confirming. 2. A clock for obtaining the month, day and time.
The life scene estimating apparatus according to claim 1, further comprising a calendar generating unit. 3. The thermal detection device is a pyroelectric infrared sensor device that captures a thermal image of a room, and the feature extraction unit determines at least the presence / absence of persons in the room, the number of persons, the position, posture, and movement with respect to the floor. The life scene estimation device according to claim 1 or 2, wherein the life scene estimation device is further extracted. 4. The life scenes to be detected are absent, sleeping, gathering, learning, relaxing, active housework as status life scenes, and waking up, going to bed, entering, and leaving as status change life scenes. The life scene estimation device according to claim 1. 5. The life scene estimating apparatus according to claim 1, wherein the environment detecting means detects a lighting state of the indoor lighting. 6. The candidate life scene storage means for storing the candidate life scenes obtained by the life scene decision means by the candidate life scene estimation means, the counter for counting the temporal continuity of the candidate life scenes, and the state of the candidate life scenes. And a life scene change judging means for judging the occurrence of a state change life scene from the change relationship between the candidate life scene and the life scene up to now. Item 7. The life scene estimation device according to any one of items 1 to 5. 7. When a series of processes for estimating a life scene after measuring a room is used as one processing unit, the feature extracting means determines the candidate life scene as an amount that extends over a plurality of processing units. The degree of fluctuation of the number of persons, which represents the magnitude of the fluctuation of the number of persons in the room obtained, and the degree of position stay, which indicates at which position in the room the person in the room is staying for a long time,
The life scene estimation device according to any one of claims 1 to 6, wherein a period activity amount that represents a maximum value of the activity amount of a plurality of processing units is used. 8. An air conditioner incorporating the life scene estimation device according to any one of claims 1 to 7.