JP2016038686A - Monitoring apparatus and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a monitoring apparatus for detecting a person from a far-infrared image with high accuracy.SOLUTION: A monitoring apparatus 10 receives time-series data of far-infrared images captured by a far-infrared camera 90, compares background images generated in advance on the basis of the far-infrared images of the received time-series data, extracts a heat-source area from the far-infrared images, and determines a likelihood of person, on the basis of attribute information of the extracted heat-source area. It is determined whether the heat-source area extracted for each of the far-infrared images of the time-series data represents a person or not, on the basis of the determined likelihood of person.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a monitoring device and a program, and more particularly to a monitoring device and a program using a far infrared image.

  Conventionally, a method for detecting a person from an infrared image taken by an infrared camera or the like has been proposed and applied to various fields such as crime prevention and disaster prevention. For example, a method for detecting a person from an infrared image taken not only in a low-temperature environment such as nighttime but also in a high-temperature environment such as summer daytime has been proposed.

JP 2009-156776 A

  However, the technique described in Patent Document 1 requires a laser range finder, which is a distance measuring unit for measuring the distance to an object, in the infrared camera, which increases the cost of the infrared camera.

  The present invention has been made in view of the problems described above, and a monitoring device capable of accurately detecting a person from a far-infrared image even when only a far-infrared camera is used. The purpose is to provide a program.

  In order to achieve the above object, a monitoring apparatus according to the first invention includes: a receiving unit that receives time-series data of a far infrared image captured by a far infrared camera that captures a predetermined range; Each of the received far-infrared images of the time-series data is compared with a background image generated in advance based on the far-infrared image, thereby extracting the heat source region from the far-infrared image, and by the extracting means Based on the extracted attribute information of the heat source area, the first determination means for determining the character of the person and the heat source area extracted for each of the far-infrared images of the time-series data are determined by the first determination means. And a second determination unit that determines whether the heat source region represents a person based on the determination result.

  According to a second aspect of the present invention, there is provided a program for receiving a time series data of a far infrared image captured by a far infrared camera that captures a predetermined range, and a time series data received by the receiving means. Extraction means for extracting the heat source area from the far infrared image by comparing with each of the infrared images with a background image generated in advance based on the far infrared image, and attribute information of the heat source area extracted by the extraction means Based on the determination result determined by the first determination unit for the heat source region extracted for each of the far-infrared images of the time series data, Is a program for functioning as second determination means for determining whether or not a person is represented.

  According to the first invention and the second invention, the time-series data of the far-infrared image captured by the far-infrared camera that captures a predetermined range is received, and the far-infrared image of the received time-series data is received. For each, a heat source region is extracted from the far-infrared image by comparing with a background image generated in advance based on the far-infrared image.

  Then, based on the extracted attribute information of the heat source region, the person likeness is determined, and based on the person likeness determination result for each heat source region extracted for each far-infrared image of the time series data, the heat source region is It is determined whether or not a person is represented.

  Thus, even in the case of a far-infrared image, it is possible to accurately detect a person from the far-infrared image by determining the person-likeness for each of the far-infrared images of the time-series data.

  Further, the first determination means according to the first invention is a vertically long shape whose shape is long in the vertical direction based on the extracted attribute information of the heat source region, and the size of the heat source region is within a predetermined range. When the temperature of the heat source region is within a predetermined range, it is determined that the heat source region is likely to be a person. Accordingly, it is possible to narrow down in advance whether or not a person is included in the far-infrared image.

  Further, the second determination means according to the first invention is based on the determination result determined by the first determination means for the heat source region extracted for each of the far-infrared images of the time series data. Within the time range, the number of times that the heat source area was determined to be human was larger than the threshold, and the heat source area extracted from the latest far-infrared image in the time series data was determined to be human. In this case, it is determined that the heat source region represents a person. Thereby, the detection accuracy for detecting a person from a far-infrared image can be further improved.

  Further, the second determination means according to the first invention is based on the determination result determined by the first determination means for the heat source region extracted for each of the far-infrared images of the time series data. Within the time range, if the number of times that the heat source region is likely to be a person is determined more than once as a threshold, and the heat source region extracted from the latest far-infrared image in the time-series data is a person When it is determined that it is likely, the heat source region is determined to represent a person. Thereby, the detection accuracy for detecting a person from a far-infrared image can be further improved.

  Further, the second determining means according to the first invention is based on the determination result determined by the first determining means for the heat source region extracted for each of the far-infrared images of the time series data. When it is continuously determined that each of the heat source regions extracted from each of the latest predetermined number of far-infrared images in the series data is likely to be a person, it is determined that the heat source region represents a person. Thereby, the detection accuracy for detecting a person from a far-infrared image can be further improved.

  Further, the second determination means according to the first aspect of the invention is based on the attribute information of the heat source region extracted for each of the far-infrared images of the time series data, and the shape is a shape having an aspect ratio of 1 or left and right. The heat source area is located at the bottom of the far-infrared image, the fluctuation amount of the position of the heat source area is larger than a threshold value, and the temperature of the heat source area is within a predetermined range. If it is present and the size of the heat source area is smaller than the threshold, it is determined that the heat source area represents a pet. Thereby, a pet can be detected from a far-infrared image.

  Further, the second determination means according to the first invention is configured such that the heat source region is positioned above the far infrared image based on the attribute information of the heat source region extracted for each far infrared image of the time series data. When the amount of variation in the position of the heat source region is less than the threshold value and the temperature of the heat source region is lower than the threshold value in summer, or when the temperature of the heat source region is higher than the threshold value in winter Determine that the area represents an air conditioner. Thereby, the air conditioner can be detected from the far-infrared image.

  Further, the second determination means according to the first invention is configured such that the heat source region is positioned below the far infrared image based on the attribute information of the heat source region extracted for each of the far infrared images of the time series data. And the heat source region is heated when the variation in the position of the heat source region is less than the threshold, the temperature of the heat source region is higher than the threshold, and the size of the heat source region is less than the threshold. Determined to represent an appliance. Thereby, a heating appliance can be detected from a far-infrared image.

  Further, the second determination means according to the first invention is based on the attribute information of the heat source region extracted for each of the far-infrared images of the time series data, and the variation amount of the position of the heat source region is less than the threshold value. If the temperature of the heat source area is higher than the temperature of the heat source area at night and the size of the heat source area is larger than the threshold, it is determined that the heat source area represents a window. Thereby, the window can be detected from the far-infrared image.

  Further, the second determination means according to the first invention is based on the attribute information of the heat source region extracted for each of the far-infrared images of the time series data, and the variation amount of the position of the heat source region is less than the threshold value. When the first determination unit determines that there is a human-like object in the far-infrared image at the same time including the heat source area and the size of the heat source area is greater than the threshold value, If the temperature is higher than the threshold, it is determined that the heat source area represents a television. Thereby, the television can be detected from the far-infrared image.

  Further, the second determination means according to the first invention is based on the attribute information of the heat source region extracted for each of the far-infrared images of the time series data, and the variation amount of the position of the heat source region is less than the threshold value. When the first determination unit determines that there is a human-like object in the far-infrared image at the same time including the heat source area and the size of the heat source area is smaller than the threshold value, If the temperature is higher than the threshold, it is determined that the heat source region represents an ashtray. Thereby, the ashtray can be detected from the far-infrared image.

  Furthermore, the monitoring apparatus according to the first invention further includes an output means for outputting an alarm when the second determination means determines that the heat source region represents a person. This makes it possible to report the presence of a suspicious person to the outside even when the user is away.

  The background image according to the first invention is updated every time a far-infrared image is received by the receiving means. Thereby, the detection accuracy for detecting a person from a far-infrared image can be further improved.

  As described above, according to the monitoring apparatus and program of the present invention, even when only a far-infrared camera is used, a person can be accurately detected from a far-infrared image, and further, far-red Even when the resolution of the outside image is not high, it is possible to obtain an effect that a person can be accurately detected from the far-infrared image.

FIG. 1 is a block diagram showing the configuration of the monitoring system. FIG. 2 is a schematic diagram for explaining an installation example of a far-infrared camera. FIG. 3 is a schematic diagram for explaining an example of a far-infrared image photographed by a far-infrared camera. FIG. 4 is a schematic diagram for explaining an example of a far-infrared image photographed by a far-infrared camera. FIG. 5 is a schematic diagram for explaining an example of processing for extracting an object region from an input image. FIG. 6 is a diagram illustrating an example of attribute information. FIG. 7 is a schematic diagram for explaining the attribute information. FIG. 8 is a diagram illustrating an example of determination items and determination conditions for preliminary determination. FIG. 9 is a diagram illustrating an example of a predetermined determination formula for each type of object. FIG. 10 is a diagram illustrating an example of a determination formula for determining a person. FIG. 11 is a diagram illustrating an example of another determination formula for determining a person. FIG. 12 is a diagram illustrating an example of another determination formula for determining a person. FIG. 13 is a flowchart showing the contents of a background image calculation routine in the monitoring apparatus according to the present embodiment. FIG. 14 is a flowchart showing the contents of a person determination routine in the monitoring apparatus according to the present embodiment.

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

  As shown in FIG. 1, the monitoring system 5 according to the present embodiment includes a far-infrared camera 90 and a monitoring device 10.

  The far infrared camera 90 is, for example, a camera that captures a wavelength of 3 μm to 1000 μm corresponding to the far infrared region. According to the present invention, a high-infrared far-infrared camera having tens of thousands of pixels may be used, but an effect can be obtained even when a low-resolution far-infrared camera having several hundred to several thousand pixels is used. Here, as an example, the far-infrared camera 90 will be described as being a low-resolution far-infrared camera of about several hundred to several thousand pixels.

  As shown in FIG. 2, the far-infrared camera 90 is fixedly attached to a predetermined location such as a house room, for example, and takes a far-infrared image within a predetermined range. In the example of FIG. 2, an air conditioner 100, a heating device 102, a window 104, a television 106, a pet 108, and a person 110 exist within the imaging range of the far infrared camera 90.

  FIG. 3 is a diagram showing an example of a far-infrared image when the state of the room shown in FIG.

Since the far-infrared camera 90 is a low-resolution camera that captures wavelengths in the far-infrared region, unlike a visible light camera or a high-resolution far-infrared camera, as shown in FIG. The outline of the included object is blurred, and the type of the object included in the far-infrared image cannot be specified only by viewing the far-infrared image. On the other hand, since the amount of far infrared rays emitted from the object differs according to the temperature of the object, each pixel of the far infrared camera 90 detects the temperature of the object from the amount of far infrared rays, and the pixel value corresponding to the detected temperature Is output from the far-infrared image. As an example, the pixel value output by the pixel of the far-infrared camera 90 increases as the temperature of the object increases. In addition, each pixel of the far-infrared camera 90 is arranged in a rectangle so that the number of pixels in the horizontal direction of the captured far-infrared image is W _image and the number of pixels in the vertical direction is H _image. .

  Note that the far-infrared image captured by the far-infrared camera 90 may be either a still image or a moving image.

  On the other hand, the monitoring apparatus 10 in FIG. 1 is configured by a computer including a CPU, a RAM, and a ROM that stores a program for executing a background image calculation routine and a person determination routine described later. As shown in FIG.

  The monitoring device 10 includes a reception unit 20, a calculation unit 25, an output unit 70, and a storage unit 80. The calculation unit 25 includes an extraction unit 30, an acquisition unit 40, an accumulation unit 50, and a determination unit 60.

  The accepting unit 20 accepts a far infrared image captured by the far infrared camera 90. As already described, since each pixel value of the far infrared image represents temperature, as shown in FIG. 4, a far infrared image showing the temperature distribution within the imaging range of the far infrared camera 90 can be acquired. In the example of the far infrared image shown in FIG. 4, some object having a temperature in the range of 25 ° C. to 35 ° C. different from the ambient temperature is recognized at the center of the far infrared image. In addition, the example of the far-infrared image shown in FIG. 4 focuses on one of the objects included in the far-infrared image, and an actual far-infrared image includes a plurality of objects as shown in FIG. Contains objects.

  In addition, the reception unit 20 further receives an instruction regarding the operation of the monitoring device 10 such as activation or stop of the monitoring device 10.

  The extraction unit 30 generates a background image 14 from the far-infrared image received by the reception unit 20, the generated background image 14, and the input image 12 that is a far-infrared image received by the reception unit 20 after the background image is generated. Are extracted from the input image 12 a region different from the temperature distribution of the background image 14, that is, a region where regions having higher or lower temperatures than the background image 14 are gathered with a certain size. FIG. 5 is a diagram schematically illustrating this process, and shows how the heat source region 16 is extracted from the input image 12 using the background image 14.

  Here, the background image 14 is an image calculated from a predetermined number N of far-infrared images received by the receiving unit 20.

  Specifically, the arrangement of each pixel of the far-infrared image is as follows. As shown in FIG. When the vertical direction of the far-infrared image is represented by (x, y) coordinates with the Y axis as the background, the background image 14 is calculated by the equations (1) and (2).

Here, I _n (x, y) is the pixel in the coordinates (x, y) of the n-th (n = 1,..., N) far-infrared image among the N far-infrared images. The pixel value, I ^{^} (x, y) is the average value of the pixels at the coordinates (x, y) of the background image 14, and σ (x, y) is the standard deviation of the pixels at the coordinates (x, y) of the background image 14. Represents.

  Then, the extraction unit 30 calculates the difference between the input image 12 and the background image 14 according to the equation (3), and calculates a filter F (x, y) for extracting the heat source region 16 from the input image 12.

  Here, I (x, y) represents the pixel value of the pixel at the coordinates (x, y) of the input image 12, and M represents the pixel value margin for extracting the heat source region 16. In other words, since the pixel value indicates temperature information, the pixel value margin can also be said to represent the temperature margin.

  The extraction unit 30 extracts a region where F (x, y) = 1 from the input image 12 as the heat source region 16 based on the value of the filter F (x, y). Then, the extraction unit 30 stores the background image 14 and the heat source region 16 in the storage unit 80.

  The acquisition unit 40 acquires attribute information for each heat source region 16 extracted by the extraction unit 30. Here, the attribute information of the heat source region 16 refers to a quantitative value obtained from the temperature distribution of each heat source region 16. FIG. 6 is a diagram illustrating an example of attribute information of the heat source region 16, and the acquisition unit 40 acquires information about, for example, position, size, area, and temperature from the heat source region 16 as attribute information.

As attribute information relating to the position of the heat source region 16, for example, there is a central point P _C and the center of gravity P _G of the heat source region 16. The central point P _C of the heat source regions 16, as shown in FIG. 7, four sides diagonal of the rectangle contacting the heat source region 16 refers to a coordinate point on the input image 12 at a position intersecting. Further, the center of gravity P _G heat source region 16 refers to a coordinate point on the input image 12 in the center of mass of heat sources region 16.

  The attribute information related to the dimensions of the heat source region 16 includes, for example, the width W and the height H of the heat source region 16. As shown in FIG. 7, the width W of the heat source region 16 refers to the number of pixels in the horizontal direction of a rectangle whose four sides are in contact with the heat source region 16. The height H of the heat source region 16 refers to the number of pixels in the vertical direction of the rectangle.

  The attribute information related to the area of the heat source region 16 includes, for example, the number of pixels S of the heat source region 16. The number S of pixels in the heat source region 16 is obtained by measuring the number of pixels in which F (x, y) = 1 in the filter F (x, y) calculated when the heat source region 16 is extracted from the input image 12. Can be obtained at.

The attribute information regarding the temperature of the heat source region 16 includes, for example, the average temperature T _ave , the maximum temperature T _max , and the minimum temperature T _{min of} the heat source region 16. The average temperature T _ave of the heat source region 16 is the average value of the pixel values in each pixel included in the heat source region 16, and the maximum temperature T _max of the heat source region 16 is the maximum pixel value of the pixels included in the heat source region 16, the heat source The minimum temperature T _min of the region 16 can be acquired as the minimum pixel value among the pixels included in the heat source region 16.

  The accumulation unit 50 associates the attribute information of the heat source region 16 acquired by the acquisition unit 40 with the time information indicating the passage of time of the attribute information, and accumulates the attribute information and time information of the heat source region 16 in the storage unit 80. To do.

  When the input image 12 includes a plurality of heat source regions 16 and attribute information is acquired from each of the heat source regions 16, the attribute information is attribute information obtained from the input image 12 taken at the same time. Therefore, the same time information is associated with each attribute information. As described above, in this embodiment, the time series is described as an example of time, but an index representing another time series such as a serial number may be used.

  After the attribute information of the heat source region 16 is accumulated in the storage unit 80 over a predetermined period, the determination unit 60, for example, every time the input image 12 is received by the reception unit 20, the heat source included in the received input image 12 Based on the attribute information of the region 16 and the attribute information of the heat source region 16 included in the past input image 12 accumulated so far in the storage unit 80, the object indicated by the heat source region 16 included in the newly received input image 12 The type of is determined.

  Therefore, the determination unit 60 first performs preliminary determination on the heat source region 16 based on the attribute information acquired by the acquisition unit 40.

  In the preliminary determination, the possibility that the object indicated by the heat source region 16 is a person, that is, “humanity” is determined. Therefore, even if it is determined by preliminary determination that the object indicated by the heat source region 16 is human, the heat source region 16 is not immediately determined to be a person.

  FIG. 8 is a diagram illustrating an example of determination items and determination conditions for preliminary determination. As shown in FIG. 8, for example, the preliminary determination includes determination items related to the aspect ratio, area, and temperature with respect to the attribute information.

  Generally, since the shape of a person is often vertically long in the vertical direction, it is determined whether or not the width W and the height H included in the attribute information satisfy the determination condition H> αW. Here, α is a predetermined coefficient for adjusting the magnitude relationship between the width W and the height H.

Further, if the person has a certain area or more in the input image 12 and the range of body temperature that can be normally taken can be predicted in advance, the condition of S> S _th is set for the area S included in the attribute information. It is determined whether the condition of T _th1 <T _ave <T _th2 is satisfied for the average temperature T _ave included in the attribute information. Here, S _th is an area necessary for determining the heat source region 16 as a person in the input image 12, T _th1 is a lower limit temperature for determining the heat source region 16 as a person, and T _th2 is a heat source region 16 as a person. Represents the upper limit temperature for determination.

Here, instead of the condition of _{S> S th} to the area _S, it may be an upper limit value _{S max} with respect to S max>_{S> S} area S as _th. S _max is an upper limit area value that can be considered as the area of a person in the input image 12. By providing the upper limit value S _max with respect to the area S, it is possible to prevent the heat source region 16 having a size that is impossible for a person from being determined to be human.

Note that the coefficient α, the area S _th , S _max , the lower limit temperature T _th1 , and the upper limit temperature T _th2 are the number of pixels of the far infrared camera 90, the photographing range, the temperature around the installation location of the far infrared camera 90, etc. Determine according to environmental conditions. As an example, when a camera with a far-infrared camera 90 with 50 × 50 pixels in the horizontal direction × vertical direction and a shooting range of 60 degrees is installed on the ceiling of the room, α = 1.0 and S _th = 0.008. × W _image × H _image , T _th1 = 20, T _th2 = 40.

  The determination unit 60 associates R = 1 information with attribute information that satisfies all the determination conditions regarding the aspect ratio, area, and temperature, and R = 0 with other attribute information.

  Then, when the time information associated with the attribute information of the heat source region 16 at a certain timing is t, the determination unit 60 determines the type of the object indicated by the heat source region 16 according to a determination formula that is predetermined for each type of object. judge.

  FIG. 9 is a diagram illustrating an example of a determination formula predetermined for each type of object.

  No. in FIG. 1 is a determination formula for determining whether or not the object indicated by the heat source region 16 is a person, and FIG. 10 is a diagram illustrating an example of a situation represented by the determination formula 1).

  The determination formula 1) is a sum of R (t−n) representing “humanity” of attribute information at each time included in a range retroactively from the latest time t to a predetermined time point n represented by param1. Is larger than param2 and R (t) = 1, that is, when it is determined that the heat source region 16 at the latest time t is human, the object indicated by the heat source region 16 extracted at the latest time t It is a determination formula for determining the type as a person.

  FIG. 10 shows an example in which the sum total of R (t−n) is 5, and the time t, (t−n1), (t−n2), (t -N3) and an example in which it is determined that the heat source region 16 in (t-n4) is human. Here, n1 to n4 represent arbitrary natural numbers having a magnitude relationship of 1 to param1 and n1 <n2 <n3 <n4, and “...” In FIG. Indicates the range.

  When the determination formula 1) is satisfied, it is determined that there is a person in the input image 12. However, the determination formula for determining whether the object indicated by the heat source region 16 is a person is not limited to the determination formula 1).

  For example, as shown in FIG. 11, when it is determined that the heat source region 16 at each time point of the param1A that goes back to the past from the latest time is likely to be human, the heat source region extracted from the latest time t The type of the object indicated by 16 may be determined as a person. The example of FIG. 11 shows a situation in which it is continuously determined that the heat source region 16 at each of the five points in time from the latest time is likely to be human.

  Also, for example, as shown in FIG. 12, the number of times that the heat source region 16 is determined to be human is continuously more than param2B in the range that dates back from the latest time t to the time point represented by param1B. In addition, when it is determined that the heat source region 16 at the latest time t is human, the type of the object indicated by the heat source region 16 extracted at the latest time t may be determined as a person. The example of FIG. 12 shows a situation in which the number of times that it is determined that the person seems to be human twice is consecutive three times in the range traced back to the past from the latest time to the time point represented by param1B. “...” In FIG. 12 indicates a range of time that is not determined to be human, as in FIG. 10.

  In the example of FIG. 12, it is determined whether or not the number of times determined to be human twice is equal to or greater than param2B, but the number of continuous times determined to be human is not limited to two. For example, it goes without saying that it may be determined whether or not the number of times determined to be human is P or more is param2B or more P (P is a natural number of 2 or more).

  10 to 12, each parameter of param1, param2, param1A, param1B, and param2B represents a natural number, the number of pixels of the far infrared camera 90, the photographing range, and the temperature around the installation location of the far infrared camera 90. It is determined in advance according to the environmental conditions.

  Since a person not only stands and moves in a house such as a room in a house but also sits on a chair or lies on the floor, the shape of the person photographed by the far-infrared camera 90 is long in the vertical direction. It is not always vertically long. Therefore, even if the object indicated by the heat source region 16 is a person, the determination items regarding the aspect ratio of the preliminary determination shown in FIG. 8 are not satisfied, so the heat source region 16 representing a person is determined not to be a person and input There is a possibility of losing a person from the image 12. For example, when the performance of the far-infrared camera 90 is poor or when a person is hidden behind furniture, the heat source region 16 representing the person is determined not to be a person, and the person is selected from the input image 12. Possible loss of sight.

  However, even if the person loses sight of the person from the input image 12, while the person is in the room, such as when the person enters or leaves the room, the person stands and moves through the room for a plurality of times. A situation occurs. Therefore, in the person determination method shown in FIGS. 10 to 12, the preliminary determination determination items shown in FIG. 8 are obtained during the period from when the person enters the room to when the person leaves. Even if a person is lost from the input image 12 without satisfying any of them, the type of the object indicated by the heat source region 16 extracted at the latest time t can be determined as a person.

  For the above reason, param1 may be set to a number corresponding to about 1 to 10 seconds, but the value of param1B is set to a number corresponding to an order of about 10 seconds to several minutes.

  In addition, it is determined whether the object indicated by the heat source region 16 is a person in FIG. In the determination formula 1) 1, the determination regarding the change in the position of the heat source region 16 is not particularly performed. For example, when the amount of change in the position of the heat source region 16 is larger than a predetermined threshold, the object indicated by the heat source region 16 A determination condition for determining the type of a person as a person may be added.

  Depending on the situation, the size of a person and a stationary object such as a window, which will be described later, may be approximately the same. Therefore, by adding a determination condition regarding the amount of change in the position of the heat source region 16 to the determination formula 1). In addition, an effect of reducing erroneous determination between a stationary object such as a window and a person can be expected. Needless to say, the determination condition regarding the amount of change in the position of the heat source region 16 may be added to the person determination method described with reference to FIGS. 11 and 12.

  Depending on the installation position of the far-infrared camera 90, a person comes directly under the lens of the far-infrared camera 90, and the aspect ratio of the heat source area 16 corresponding to the person is approximately 1: 1, that is, the heat source area 16 The shape may be close to a circle. Such a situation is likely to occur when, for example, a far-infrared camera 90 equipped with a wide-angle lens is installed on the ceiling near the center of the room to capture a wider range.

  Accordingly, in the input image 12 of the far-infrared camera 90, information indicating a range corresponding to a position directly below the lens of the far-infrared camera 90 is stored in the storage unit 80 in advance, and the heat source region 16 exists in the range. Alternatively, the value of α for adjusting the aspect ratio determination condition shown in FIG. 8 may be changed to a value closer to “1”.

  In this way, whether or not a person is included in the input image 12 by adjusting the value of α from the information related to the installation position of the far-infrared camera 90 and the information related to the setting of the lens or the like attached to the far-infrared camera 90. The determination accuracy can be improved.

  No. in FIG. 2 determination formulas 2) to 4) are determination formulas for determining whether or not the object indicated by the heat source region 16 is an air conditioner. The air conditioner is fixedly installed at a relatively upper portion of the input image 12, and has a tendency to show an ambient temperature in summer, that is, a temperature lower than room temperature, and to show a temperature higher than room temperature in winter.

Therefore, the judgment formula 2) is a determination formula in which central point P _C attribute information showed that the top half of the input image 12. Here _P cy (t) represents the y coordinate of the center point _{P C} of the attribute information at time t. Hereinafter, with respect to the symbol Z representing any attribute information, Z (t) represents the attribute information at time t.

Determination equation 3) is a discriminants central point P _C of attribute information indicates that no change. _Here, dP cy (t) / dt is, for example, a value representing the time variation of the y-coordinate of the center point _{P C} at time t from y-coordinate of the center point _{P C} at time (t-n). Note that the time (t−n) (n is an integer of 0 or more) represents the time at the time of going back n times from the time t.

The determination formula 4) is a determination formula for determining the relationship between the average temperature T _ave of the attribute information at the time t and the room temperature T _room . The judgment formula 4) is different from the judgment formula used for judgment according to the season. For example, the formula (4) is used during the summer period from June to September, and the formula (5) is used during the winter period from December to March.

Formula (4) is a determination formula indicating that the average temperature T _ave of the attribute information at time t is lower than the _room temperature T _room . On the other hand, equation (5) is a determination equation indicating that the average temperature T _ave is higher than the _room temperature T _room .

Here, param4 is a margin for adjusting the _room temperature Troom. Incidentally, paramβ (β is a natural number) and room temperature T _{room room} in Figure 9, in advance according to the far-infrared pixel number and the shooting range of the camera 90, and environmental conditions such as temperature around the installation place of the far-infrared camera 90, etc. Make a decision.

  And the determination part 60 determines with the object which the said heat-source area | region 16 shows is an air conditioner, when the attribute information in the time t satisfies all the determination formulas 2) -4). Note that the summer and winter periods shown above are examples, and the summer is not limited to the period from June to September, and the winter is not limited to the period from December to March.

  No. in FIG. 3 determination formulas 5) to 8) are determination formulas for determining whether or not the object indicated by the heat source region 16 is a heating appliance.

  Heating appliances are fixedly installed in a relatively lower part of the room and show a temperature higher than the ambient temperature, that is, a temperature higher than room temperature, and the range showing the temperature higher than the room temperature tends to be limited to a relatively narrow range. . The determination formulas 5) to 8) formulate the tendency of attribute information characteristic of such a heating appliance, and the determination unit 60 determines the attribute information of the heat source region 16 at time t as the determination formulas 5) to 8). Is satisfied, it is determined that the object indicated by the heat source region 16 is a heating appliance.

  In addition, in FIG. 4 determination formulas 9) to 11) are determination formulas for determining whether or not the object indicated by the heat source region 16 is a window.

  The windows are fixedly installed in the building, and the daytime temperature is higher than the nighttime temperature, and the size is relatively large. The determination formulas 9) to 11) formulate the tendency of attribute information characteristic of such windows, and the determination unit 60 determines that the attribute information of the heat source region 16 at time t is determined by the determination formulas 9) to 11). When all are satisfied, it is determined that the object indicated by the heat source region 16 is a window.

In the determination formula 10), the average temperature T _ave (night) represents the average temperature of the attribute information at the time corresponding to the time corresponding to the night, and the average temperature T _ave (day) corresponds to the daytime. It represents the average temperature of the attribute information at the time corresponding to the time. For example, the period from 6 o'clock to less than 18 o'clock is determined as daytime, and the period from after 18 o'clock to less than 6 o'clock is determined as night, and determination formula 9) is determined. However, the period setting for daytime and night is limited to this. It is not a thing.

  In addition, in FIG. 5 determination formulas 12) to 14) are determination formulas for determining whether or not the object indicated by the heat source region 16 is a television.

  Since TVs are installed indoors and are often turned on and watched when there are people, the temperature tends to be higher than room temperature and relatively large while people are present. It can be seen. The determination formulas 12) to 14) formulate the tendency of attribute information characteristic of such a television, and the determination unit 60 determines that the attribute information of the heat source region 16 at time t is determined by the determination formulas 12) to 14). When all are satisfied, it is determined that the object indicated by the heat source region 16 is a television.

In the determination formula 13), “T _ave (t)> param 11 if R (t) = 1” is attribute information having the same time and a determination target of the type of the object currently focused on. This indicates whether the average temperature T _ave (t) of the specific attribute information is higher than param11 when attribute information whose result of the preliminary determination is R = 1 exists in attribute information different from the specific attribute information. It is a judgment formula.

  In addition, in FIG. 6 determination formulas 15) to 17) are determination formulas for determining whether or not the object indicated by the heat source region 16 is an ashtray.

  The ashtray is placed at a fixed position such as a table during use, and a lit cigarette is often placed during smoking. Tend to be small. The determination formulas 15) to 17) formulate the tendency of attribute information characteristic of such an ashtray, and the determination unit 60 determines the attribute information of the heat source region 16 at time t as the determination formulas 15) to 17). When all are satisfied, it is determined that the object indicated by the heat source region 16 is an ashtray.

  In addition, in FIG. 7 is a determination expression for determining whether or not the object indicated by the heat source region 16 is a pet.

  The shape of the pet is the same in length and width, that is, the aspect ratio is 1, or the landscape is long in the left-right direction, and the pet is positioned relatively lower in the input image 12 because it moves on the floor in the room, There is a tendency that the amount of movement is large, the size is smaller than that of a person, and the range of body temperature that can be taken is determined. The determination formulas 18) to 22) formulate the tendency of the attribute information characteristic of such pets, and the determination unit 60 determines that the attribute information of the heat source region 16 at the time t is the determination formulas 18) to 22). When all are satisfied, it is determined that the object indicated by the heat source region 16 is a pet.

  Note that the values of param1 to param19 shown in the determination result column of FIG. 9 are examples, and depend on the type of the far infrared camera 90 to be used and the environmental conditions such as the temperature around the location where the far infrared camera 90 is installed. Changed.

  In the example of FIG. 1-No. Although the determination formulas for determining the types of the seven objects up to 7 are shown, the types of the objects determined by the determination unit 60 are not limited to these seven objects. If a determination formula is added according to the location or environment where the far infrared camera 90 is installed, the determination unit 60 can determine various object types.

  For example, when the determination unit 60 determines that a person is included in the input image 12, the output unit 70 notifies the administrator of the monitoring system 5 to that effect. Note that the input image 12 is not limited to the case where a person is included, but may be notified to the administrator of the monitoring system 5 when other objects such as pets are included.

  The storage unit 80 is, for example, a far-infrared image received from the far-infrared camera 90, a background image 14, attribute information of the heat source region 16 associated with time information, and a determination formula determined in advance for each type of object, etc. Data necessary for object detection in the monitoring device 10 is stored.

  Next, the operation of the monitoring apparatus 10 according to the present embodiment will be described. Here, it is assumed that the background image 14 is not yet stored in the storage unit 80. First, after the monitoring apparatus 10 is activated, a background image calculation routine shown in FIG. 13 is executed.

  In step S <b> 10, a far infrared image captured by the far infrared camera 90 is acquired. Note that when a moving image having a predetermined frame rate is output from the far-infrared camera 90, an image for one frame is accepted as a far-infrared image.

  Next, in step S20, a smoothing process is performed on the far-infrared image acquired in the process of step S10. The smoothing process is a process of smoothing a change in pixel values included in the far infrared image and removing a noise component included in each pixel of the far infrared image. For the smoothing process, for example, a known technique using a moving average filter or a Gaussian filter is applied.

  In step S30, the number of far-infrared images acquired so far is measured, and it is determined whether or not the number of far-infrared images acquired is equal to or greater than the specified number N. If the determination is affirmative, the process proceeds to step S50. If the determination is negative, the process proceeds to step S40.

  Here, the prescribed number N of far-infrared images is a value that defines the number of far-infrared images necessary to generate the background image 14, for example, the type of far-infrared camera 90 to be used, and It is set according to environmental conditions such as the temperature around the place where the far infrared camera 90 is installed.

  In step S40, the far-infrared image smoothed by the process of step S20 is stored in the storage unit 80, and the process proceeds to step S10. And the far-infrared image is memorize | stored in the memory | storage part 80 until it reaches the regulation number N by repeating the process of step S10-step S40.

  On the other hand, when affirmative determination is made in the processing of step S30, that is, when a far-infrared image having a prescribed number N or more is received, N far-infrared images are obtained for each pixel based on the above-described equation (1) in step S50. An average value obtained by averaging the pixel values is calculated.

  In step S60, the standard deviation calculated from the N far-infrared images for each pixel and the average value for each pixel calculated in step S50 is calculated based on the above-described equation (2).

  In step S70, the background image 14 having the average value calculated in step S50 and the standard deviation calculated in step S60 is generated and stored in the storage unit 80.

  With the above processing, the background image calculation routine shown in FIG. 13 ends. Then, after the background image calculation routine ends, a person determination routine shown in FIG. 14 is executed.

  In step S100, for example, it is determined whether or not an instruction to stop the person determination routine has been received by pressing a far-infrared image acquisition end button (not shown) provided in the monitoring apparatus 10, and in the case of an affirmative determination, the person determination is performed. End the routine. On the other hand, if a negative determination is made, the process proceeds to step S110. The stop instruction of the person determination routine is not limited to pressing the far-infrared image acquisition end button, and may be received via a communication line such as the Internet (not shown) connected to the monitoring apparatus 10, for example. .

  In step S110, a far infrared image captured by the far infrared camera 90 is acquired. Hereinafter, the far-infrared image acquired in this step is particularly referred to as the input image 12 in order to distinguish it from the far-infrared image acquired in step S10 of the background image calculation routine shown in FIG.

  In step S120, a smoothing process is performed on the input image 12 acquired in the process of step S110, as in the process of step S20 of the background image calculation routine shown in FIG.

  In step S130, the background image 14 stored in the storage unit 80 in the process of step S70 is read, and the background image 14 and the input image 12 smoothed in the process of step S120 for each pixel based on the above-described equation (3). The difference is calculated. Then, a filter F (x, y) is generated in which the filter value of the coordinates where the difference is equal to or larger than “σ (x, y) + M” is set to “1”, and the filter value of the other coordinates is set to “0”. Note that the filter F (x, y) is adjusted by adjusting the pixel value margin M in accordance with the type of the far infrared camera 90 to be used and the environmental conditions such as the temperature around the location where the far infrared camera 90 is installed. The size of the region where the filter value is 1 can be adjusted.

  In step S140, the pixel value of the pixel included in the input image 12 smoothed in the process of step S120 is multiplied for each pixel by the filter value of the filter F (x, y). A pixel having a pixel value that is more than a predetermined difference from the background image 14, that is, “σ (x, y) + M”, is extracted.

  The region in which the pixels extracted in this way are formed adjacent to each other is a region different from the temperature distribution of the background image 14, and the heat source region 16 representing some object located within the imaging range of the far infrared camera 90. Can be considered. When the input image 12 includes a plurality of objects different from the temperature distribution of the background image 14, a plurality of heat source regions 16 are extracted from the input image 12.

  In step S150, for example, the attribute information shown in FIG. 6 is acquired for each of the heat source regions 16 extracted from the input image 12 in the process of step S140. Then, the preliminary determination shown in FIG. 8 is performed on each of the heat source regions 16 based on the acquired attribute information to determine whether or not the object indicated by the heat source region 16 is human. When it is determined that the object indicated by the heat source region 16 is human, information on the preliminary determination result set to R = 1 is included in the attribute information of the heat source region. On the other hand, if it is not determined that the object indicated by the heat source region 16 is human, information on the preliminary determination result set to R = 0 is included in the attribute information of the heat source region.

  In step S155, the attribute information of each heat source region 16 and the current time information are associated and stored in the storage unit 80. Note that the time information associated with the attribute information may be a time that clearly indicates the time of the input image 12, and may be, for example, the reception time of the input image 12, the extraction time of the heat source region 16, or the like. Further, the time information can be obtained by using, for example, a calendar function built in the CPU.

  Next, in step S160, the background image 14 is updated using the input image 12 smoothed by the process of step S120. Specifically, using the input image 12 smoothed by the process of step S120 and the far-infrared image stored in the storage unit 80 by the process of step S40 shown in FIG. Based on this, the average value for each pixel of the background image 14 is updated, and the standard deviation for each pixel of the background image 14 is updated based on the equation (2).

  At this time, since one input image 12 is added to update the background image 14, N in the expressions (1) and (2) is read as (N + 1). Then, the storage unit 80 stores the input image 12 smoothed in the process of step S120 and the background image 14 having the average value and the standard deviation updated in this step.

  In step S170, the number of input images 12 acquired so far is measured, and it is determined whether or not the acquired number of input images 12 is equal to or greater than a specified number J. If a negative determination is made, the process proceeds to step S100, and the processes of steps S100 to S170 are repeated until the number of acquired input images 12 is equal to or greater than the specified number J. At this time, in the process of step S160, the number K of the acquired input images 12 is measured every time the input image 12 is acquired, and N in the equations (1) and (2) is read as (N + K) and newly accepted. The background image 14 is updated based on the input image 12 and the far-infrared image and the input image 12 stored in the storage unit 80.

  Here, the prescribed number J of input images 12 is a value that defines the number of input images 12 necessary for accurately extracting an object from the input image 12 in the processing of step S180 described later. It is set according to the type of the far infrared camera 90 to be performed and the environmental conditions such as the temperature around the place where the far infrared camera 90 is installed. Note that instead of the specified number J of input images 12, a specified time of the input image 12 such as 1 hour or 1 day is specified, and when the input image 12 is acquired over the specified time, the process proceeds to step S180. It may be.

  On the other hand, if the determination in step S170 is affirmative, the process proceeds to step S180.

  In step S180, when the time information associated with the attribute information corresponding to the input image 12 acquired most recently is t, the type of the object indicated by the heat source region 16 corresponding to the attribute information in the time information t is shown in FIG. The determination is made according to the determination formula shown in.

  For example, first, in order to determine whether or not the heat source region 16 represents a person based on the determination result determined in step S150 with respect to the specified number J of input images 12, No. 2 in FIG. The judgment formula 1) of 1 is evaluated. Thereafter, for the heat source region 16 extracted in step S140 with respect to the specified number J of input images 12, and the position of the heat source region 16 whose position does not change over time, the type of object represented by the heat source region 16 Determine. Specifically, the No. for determining the pet in FIG. Other than the judgment formula of No. 7, 2-No. 6 is evaluated, and the type of the object indicated by the heat source region 16 is specified.

  In this step, whether or not the heat source region 16 is a heat source region representing a person is determined according to the determination formula 1) in FIG. 9, that is, the determination method illustrated in FIG. 10, but is illustrated in FIG. 11 or FIG. The determination may be made according to the determination method.

  As a result of the preliminary determination in step 150, among the heat source regions 16 extracted in step S140 with respect to the specified number J of input images 12, the position of which has changed, R = 0. No. 2 in FIG. 7 is evaluated to determine whether or not the object indicated by the heat source region 16 is a pet.

  Note that the tracking process may be performed on the heat source region 16 extracted in step S140 on the specified number J of input images 12, and the determination in step 180 may be performed on the tracked heat source region 16. . For the tracking processing, for example, a known tracking processing method such as a method for estimating the position and size of an object using a motion model of the object or a method for paying attention to the similarity of the shape of the heat source region 16 is applied. .

The determination equation 3 shown in FIG. 9), 6), 9), 12), 15), and the 20), with respect to the center point _{P C} of the attribute information, determination for determining the variation over time Contains an expression. Therefore, it is possible better to determine the variation situation over time using a center point P _C of the attribute information in more time, to determine whether they meet the more accurate determination expression. In other words, the object determination accuracy can be improved as the prescribed number J referred to in step S170 is set larger.

In the example of judging expression of each type of object as shown in FIG. 9, but the judgment formula is not shown only determines the time variation with respect to the center point P _C, other items of the attribute information by the object type, For example, a determination formula for determining a time change with respect to temperature may be defined.

  For example, when it is determined that a person is included in the input image 12, notification is made to the outside of the monitoring device 10. Specifically, when a person is detected from the input image 12 in a time zone in which there is no person in the room, an alarm sound is output from a speaker (not shown) built in the monitoring apparatus 10 and also to the monitoring apparatus 10. You may make it report to the administrator of the monitoring system 5, etc. via the connected communication lines, such as the internet which are not shown in figure. Thus, for example, the monitoring system 5 can be used for crime prevention purposes.

  Not only when a person is detected from the input image 12, but also when another object such as a pet is detected, a notification to that effect may be sent to the administrator of the monitoring system 5.

  Thus, after determining the type of the object indicated by the heat source region 16 corresponding to the attribute information at time t, the process proceeds to step S100. And when the determination of the type of the object which the heat-source area | region 16 contained in the input image 12 shows sequentially is performed by repeating the process of step S100-step S180, and the determination of step S100 becomes affirmation determination, FIG. The person determination routine shown in FIG.

  Note that the monitoring apparatus 10 according to the present embodiment can determine the type of an object even if the object does not have a predetermined shape such as a flame. As the characteristics of the flame, since the position and area vary with time and the temperature is high, for example, the judgment formula for the flame can be expressed by the equations (6) to (8).

Here, Param20 the amount of movement of the center point _{P C} of the heat source regions _{16, T fire} the minimum temperature required to identify the object and flames, Param21 is a parameter which defines the amount of variation in the number of pixels of the heat source regions 16 For example, what is necessary is just to set according to environmental conditions, such as the kind of the far infrared camera 90 to be used, and the temperature around the installation location of the far infrared camera 90.

  For example, the judgment formulas shown in the formulas (6) to (8) for the heat source region 16 that does not satisfy the judgment formula for any object shown in FIG. 9 and the result of the preliminary judgment is R = 0. Is evaluated, and the object indicated by the heat source region 16 is determined to be a flame when all the determination formulas (6) to (8) are satisfied. In this case, since there is a possibility that a fire has occurred, an alarm sound is output from a speaker (not shown) built in the monitoring device 10 and a communication line such as the Internet (not shown) connected to the monitoring device 10 is used. Then, the administrator of the monitoring system 5 may be notified.

  As described above, according to the monitoring device 10 according to the present embodiment, the object included in the input image 12 every time when the predetermined range is monitored and the input image 12 is acquired over a predetermined period. Get attribute information obtained from the temperature of. Then, it is determined whether or not a person is included in the input image 12 in accordance with a determination formula set based on the attribute information and the temporal change of the attribute information accumulated over a predetermined period.

  Therefore, the low-resolution far-infrared camera 90 in which the contour of the object is unclear to the extent that the type of the object included in the input image 12 cannot be specified from the captured input image 12 is used. The person included in the input image 12 can be detected with high accuracy. In addition, the low-resolution far-infrared camera 90 as described above is less expensive than an infrared camera that can determine the type of an object from the contour of the object, and thus reduces the cost required for constructing the monitoring system 5. Can do.

  Furthermore, the monitoring system 5 according to the present embodiment does not need to be equipped with a filter or the like that cuts a part of the wavelength band in the far-infrared camera 90, and can be constructed with one far-infrared camera 90. it can. Therefore, the cost required for construction of the monitoring system 5 can be reduced.

  Further, since the type of the object is determined based on the time change of the attribute information, the detection accuracy of the object is improved as the attribute information is accumulated over time. Furthermore, every time the input image 12 is acquired, the background image 14 used in extracting the heat source region 16 is updated. Therefore, the average value and standard deviation of the background image 14 calculated in steps S50 and S60 in FIG. Compared with the case where an object is extracted from the input image 12 without updating, the detection accuracy of the object is improved.

  Note that the present invention is not limited to the above-described embodiment, and various modifications and applications are possible without departing from the gist of the present invention. For example, the order of the processes of the background image calculation routine shown in FIG. 13 and the person determination routine shown in FIG. 14 may be changed without departing from the gist of the invention.

  In the above description, the background image calculation routine and the person determination routine are stored in advance in the computer ROM. However, the present invention is not limited to this. The program according to the present embodiment can be provided in a form recorded on a computer-readable recording medium. For example, the program according to the present embodiment can be provided in a form recorded on a portable recording medium such as a CD-ROM, a DVD-ROM, and a USB memory. The program according to the present embodiment can be provided in a form recorded in a semiconductor memory such as a flash memory.

  Further, in the monitoring apparatus 10 according to the present embodiment, the case where the background image calculation routine and the person determination routine are realized by a software configuration has been described, but the present invention is not limited to this, and for example, background image calculation The routine and the person determination routine may be realized by a hardware configuration using an ASIC (Application Specific Integrated Circuit) or the like. In this case, higher processing speed is expected compared to the above embodiment.

DESCRIPTION OF SYMBOLS 5 Monitoring system 10 Monitoring apparatus 12 Input image 14 Background image 16 Heat source area | region 20 Reception part 25 Calculation part 30 Extraction part 40 Acquisition part 50 Accumulation part 60 Determination part 70 Output part 80 Storage part 90 Far-infrared camera

Claims (14)

Receiving means for receiving time-series data of a far-infrared image captured by a far-infrared camera that captures a predetermined range;
A heat source region is extracted from the far-infrared image by comparing each of the far-infrared images of the time-series data received by the receiving unit with a background image generated in advance based on the far-infrared image. Extraction means;
Based on attribute information of the heat source region extracted by the extraction means, a first determination means for determining person-likeness;
Based on the determination result determined by the first determination unit for the heat source region extracted for each of the far-infrared images of the time-series data, it is determined whether the heat source region represents a person. Two determination means;
Monitoring device. The first determination unit is a vertically long shape whose shape is long in the vertical direction based on the attribute information of the heat source region extracted by the extraction unit, and the size of the heat source region is within a predetermined range. The monitoring apparatus according to claim 1, wherein when the temperature of the heat source region is within a predetermined range, the heat source region is determined to be a person. The second determination means is within a predetermined time range based on the determination result determined by the first determination means for the heat source region extracted for each of the far-infrared images of the time-series data. When the number of times that the heat source area is determined to be a person is larger than a threshold, and the heat source area extracted from the latest far-infrared image in the time-series data is determined to be a person. The monitoring device according to claim 1, wherein the heat source region is determined to represent a person. The second determination means is within a predetermined time range based on the determination result determined by the first determination means for the heat source region extracted for each of the far-infrared images of the time-series data. In the case where the number of times that the heat source region is likely to be a person is determined to be multiple times consecutively is greater than a threshold, and the heat source region extracted from the latest far-infrared image in the time series data is likely to be a person The monitoring apparatus according to claim 1, wherein the heat source region is determined to represent a person when it is determined. The second determination unit is configured to determine the time-series data based on the determination result determined by the first determination unit for the heat source region extracted for each far-infrared image of the time-series data. The heat source region is determined to represent a person when it is continuously determined that each of the heat source regions extracted from each of the latest predetermined number of far-infrared images is likely to be a person. Item 3. The monitoring device according to Item 2. The second determination means has a shape with an aspect ratio of 1 or a horizontally long shape in the left-right direction based on the attribute information of the heat source region extracted for each of the far-infrared images of the time-series data. The heat source region is located at the bottom of the far-infrared image, the amount of variation in the position of the heat source region is greater than a threshold value, and the temperature of the heat source region is within a predetermined range. And the monitoring apparatus of any one of Claims 1-5 which determines with the said heat-source area | region showing a pet when the magnitude | size of the said heat-source area | region is smaller than a threshold value. The second determination means, based on the attribute information of the heat source region extracted for each of the far infrared images of the time series data, the heat source region is located at the upper part of the far infrared image, and When the amount of variation in the position of the heat source region is less than the threshold, and the temperature of the heat source region is lower than the threshold in the summer, or the temperature of the heat source region is higher than the threshold in the winter The monitoring apparatus according to claim 1, wherein the heat source region represents an air conditioner. The second determination means, based on the attribute information of the heat source region extracted for each of the far infrared images of the time series data, the heat source region is located below the far infrared image, and When the amount of variation in the position of the heat source region is less than a threshold, the temperature of the heat source region is higher than the threshold, and the size of the heat source region is less than the threshold, the heat source region The monitoring device according to any one of claims 1 to 7, wherein the monitoring device is determined to represent a heating appliance. The second determination means, based on the attribute information of the heat source region extracted for each far-infrared image of the time series data, the variation amount of the position of the heat source region is less than a threshold, and The temperature of the heat source area in the daytime is higher than the temperature of the heat source area at night and the size of the heat source area is larger than a threshold value, and it is determined that the heat source area represents a window. Item 9. The monitoring device according to any one of items 8. The second determination means, based on the attribute information of the heat source region extracted for each far-infrared image of the time series data, the variation amount of the position of the heat source region is less than a threshold, and When the size of the heat source region is larger than the threshold and the far-infrared image at the same time including the heat source region is determined by the first determination means that an object like a person exists, the heat source region The monitoring device according to any one of claims 1 to 9, wherein when the temperature is higher than a threshold value, the heat source region is determined to represent a television. The second determination means, based on the attribute information of the heat source region extracted for each far-infrared image of the time series data, the variation amount of the position of the heat source region is less than a threshold, and When the size of the heat source region is smaller than a threshold and the far-infrared image at the same time including the heat source region is determined by the first determination means that an object like a person exists, the heat source region The monitoring device according to any one of claims 1 to 10, wherein when the temperature is higher than a threshold value, the heat source region is determined to represent an ashtray. The monitoring device according to claim 1, further comprising an output unit that outputs an alarm when the second determination unit determines that the heat source region represents a person. The monitoring apparatus according to any one of claims 1 to 12, wherein the background image is updated every time a far-infrared image is received by the receiving unit. Computer
Receiving means for receiving time-series data of a far-infrared image captured by a far-infrared camera that captures a predetermined range;
A heat source region is extracted from the far-infrared image by comparing each of the far-infrared images of the time-series data received by the receiving unit with a background image generated in advance based on the far-infrared image. Extraction means,
Based on the attribute information of the heat source area extracted by the extraction means, first determination means for determining person-likeness, and the first heat source area extracted for each of the far-infrared images of the time-series data. Second determination means for determining whether the heat source region represents a person based on the determination result determined by the determination means;
Program to function as.