HK1219127B - Thermal image sensor and air conditioner - Google Patents

Description

Thermal image sensor and air conditioner

Technical Field

The present invention relates to a thermal image sensor and an air conditioner capable of detecting a person at a height such as each part or posture of the person in a room.

Background

A conventional thermal image sensor is constituted by a thermopile arranged in a vertical direction, and the vertical sensor is scanned from left to right or from right to left at a constant cycle to acquire a thermal image of the entire room. At this time, the thermal image sensor acquires a thermal image of the entire room at the initial start-up, stores the thermal image as a background thermal image, calculates a difference between the entire thermal image and the background thermal image every time the entire thermal image is obtained again, and determines that a person is present at the corresponding pixel position when the difference value is equal to or greater than a human body detection threshold value (for example, patent document 1).

Patent document 1: japanese laid-open patent publication No. 2009-92282

Disclosure of Invention

In the related art thermal image sensor, a sensor in which a small number of inexpensive thermopiles are vertically arranged is scanned in a horizontal direction to generate a thermal image, and image processing is applied thereto. As a result, inexpensive human detection can be performed in the air conditioner which is required to be low in cost. By using this human detection function, air conditioning control can be performed such that air is actively blown to a person or not. However, in an air conditioner with higher comfort, a function of directly blowing air to a face of a person, a function of warming feet in winter, and the like are required to be avoided, and more accurate person detection capable of detecting a person is required.

In the related art, this high-precision person detection can be realized by increasing the number of thermopiles arranged in the vertical direction. However, the increase in the number of elements with the increase in the number of acquired data requires an increase in the memory capacity for storing and processing data and an increase in the CPU (Central processing unit) capacity for processing data, which leads to a problem that the cost of the sensor is significantly increased.

The present invention has been made to solve the above problems, and an object of the present invention is to provide a thermal image sensor and an air conditioner including: high-precision person detection can be performed without increasing the memory capacity or enhancing the CPU capability.

In the present invention, there is provided a thermal image sensor and an air conditioner, wherein the thermal image sensor scans a thermal image acquisition unit that acquires temperature data for each pixel arranged in a predetermined direction in a direction perpendicular to the predetermined direction in units of 1 step, and synthesizes 1-dimensional thermal images acquired in the respective steps to acquire a 2-dimensional thermal image, the thermal image sensor including: an effective pixel output unit that outputs a position of a detection effective pixel selected as an effective pixel from the full pixels included in the thermal image acquisition unit in the immediately preceding step; a scanning unit which performs 1-step scanning to acquire temperature data of effective pixels; a pixel weight calculation unit that generates weight values for all pixels based on the relative position of each pixel with respect to the effective pixel and temperature data; and a full-pixel classification unit that selects, from the full pixels, detection effective pixels to be used in the next step in the order of increasing weight values.

According to the present invention, it is possible to perform highly accurate person detection without increasing the memory capacity or enhancing the CPU capability.

Drawings

Fig. 1 is an explanatory diagram showing an example of the overall configuration of an air conditioner on which a thermal image sensor according to embodiment 1 of the present invention is mounted.

Fig. 2 is an explanatory diagram showing an example of the structure of the thermal image sensor according to embodiment 1 of the present invention.

Fig. 3 is a diagram illustrating a basic operation of the thermal image acquisition unit for acquiring a 2-dimensional thermal image according to embodiment 1 of the present invention.

Fig. 4 is a diagram for explaining an operation of the control unit according to embodiment 1 of the present invention.

Fig. 5 is a diagram for explaining effective pixels determined by the equal arrangement portion in embodiment 1 of the present invention.

Fig. 6 is a diagram for explaining a thermal image obtained by the thinned-out scanning unit according to embodiment 1 of the present invention.

Fig. 7 is an example of the human similarity function stored in the human similarity function storage unit according to embodiment 1 of the present invention.

Fig. 8 is a diagram for explaining the human similarity s obtained by the human similarity calculation unit according to embodiment 1 of the present invention.

Fig. 9 is a diagram for explaining a configuration example of the human similarity map D according to embodiment 1 of the present invention.

Fig. 10 is a diagram showing an example of the weighting function w (x) according to embodiment 1 of the present invention.

Fig. 11 is a diagram for explaining the pixel weight map W according to embodiment 1 of the present invention.

Fig. 12 is a diagram for visually explaining the dynamic effective pixel arrangement according to embodiment 1 of the present invention.

Fig. 13 is a diagram for explaining an operation of a control unit provided in the thermal image sensor according to embodiment 2 of the present invention.

Fig. 14 is a diagram showing an example of the prior whole pixel map stored in the prior whole pixel map storage unit according to embodiment 2 of the present invention.

Fig. 15 is a diagram showing an example of the prior whole pixel weight map W2 according to embodiment 2 of the present invention.

Description of the symbols

1: an air conditioner; 10: a thermal image acquisition section; 11: a thermal image sensor; 12: an indoor unit; 13: an outdoor unit; 14: a control unit; 21. 21a, 21b, 21 n: a vertical orientation thermal image; 22: an overall room thermal image; 110: an effective pixel output section; 111: an equal arrangement part; 112: an effective pixel selection section; 120: a thinning and scanning section; 130: a person similarity generating unit; 131: a person similarity function storage unit; 132: a person similarity calculation unit; 140: a pixel weight calculation unit; 141: a person similarity evaluation unit; 142: a person similarity map generation unit; 143: a pixel weight map generation unit; 150: a full-pixel classification section; 151: a mapping classification unit; 152: an effective pixel determination section; 160: a prior weight coefficient generation unit; 161: a prior whole pixel mapping storage unit; 162: a prior whole pixel weight map generation unit.

Detailed Description

Embodiment 1.

Fig. 1 is an explanatory diagram showing an example of the overall configuration of an air conditioner on which a thermal image sensor according to embodiment 1 of the present invention is mounted. The air conditioner 1 includes an indoor unit 12 and an outdoor unit 13. The indoor unit 12 is provided with a thermal image sensor 11.

Fig. 2 is an explanatory diagram showing an example of the structure of the thermal image sensor 11. The thermal image sensor 11 includes a thermal image acquisition unit 10 and a control unit 14. The thermal image acquisition unit 10 includes, for example, a thermopile arranged in a vertical direction, and can acquire a 1-dimensional thermal image. In the example shown in the figure, a 1-dimensional thermal image of N pixels (N is a natural number) can be acquired. The vertical sensor scans the room from left to right or from right to left at a constant cycle, thereby acquiring a thermal image of the room as a 2-dimensional image. The control unit 14 controls acquisition of the thermal image in the thermal image acquisition unit 10 including the scanning. The position of each pixel in the thermal image acquisition unit 10 is defined as being in the upward and downward direction as shown in the drawing, and the uppermost pixel position is 1 and the lowermost pixel position is N. In the present embodiment, the case where the thermal image acquisition unit 10 in which the pixels are arranged in the vertical direction as the predetermined direction is scanned in the horizontal direction perpendicular to the predetermined direction is described, but the present invention is not limited to this, and for example, the thermal image acquisition unit may be configured such that the pixels are arranged in the horizontal direction and the pixels are scanned in the vertical direction.

Fig. 3 is a diagram illustrating a basic operation of acquiring a 2-dimensional thermal image by the thermal image acquisition unit 10. The thermal image acquisition unit 10 acquires the vertical thermal image 21, which is a 1-dimensional thermal image in the vertical direction, in a time-sharing manner. By scanning the positions where the thermal images are acquired in the horizontal direction, the thermal images acquired at the respective positions are synthesized, thereby generating an entire room thermal image 22 which is a 2-dimensional thermal image of the entire room. For example, at time t0, the vertical-direction thermal image 21a is acquired as the vertical-direction thermal image 21. In addition, at time t1, the vertical-direction thermal image 21b is acquired as the vertical-direction thermal image 21. In addition, at time tn, the vertical-direction thermal image 21n is acquired as the vertical-direction thermal image 21. The thermal image acquisition unit 10 scans the position where the thermal image is acquired horizontally, and acquires the vertical thermal image 21 at each horizontal position. The thermal image acquisition unit 10 synthesizes the acquired plurality of vertical thermal images 21 to generate an overall room thermal image 22. The scanning in the horizontal direction can be realized by driving the thermopile with a stepping motor, for example.

Fig. 4 is a diagram for explaining the operation of the control unit 14. The control unit 14 determines effective pixels for acquiring a thermal image from all pixels in the vertical direction provided in the thermal image acquisition unit 10 based on the use resources (the number of pixels that can be used) permitted in the air conditioner 1 in the equalization arrangement unit 111, and generates an effective pixel index i for specifying each effective pixel. Here, the number of effective pixels is smaller than the number of full pixels. When Ns denotes the number of all pixels included in the thermal image acquisition unit 10 and Nmax denotes the number of determined effective pixels, the position xi of the effective pixel corresponding to the effective pixel index i is obtained by the following equation (1). At this time, the effective pixel index i becomes an integer of 1 to Nmax.

[ mathematical formula 1]

Fig. 5 is a diagram for explaining effective pixels determined by the equalization arrangement unit 111. In the figure, squares respectively represent pixels. In the figure, the effective pixels are denoted by Pe and are filled with gray. Further, S1 in the figure indicates the scanning direction when the thermal image is acquired. In the example shown in the figure, the number of all pixels included in the thermal image acquisition unit 10 is twice the number of effective pixels. The effective pixel corresponding to the effective pixel index i in the uniform arrangement is set as an initial effective pixel, and the position xi of the initial effective pixel is set as an initial effective pixel position. The equalization arrangement section 111 outputs the effective pixel index and the initial effective pixel position.

In the present embodiment, the equalization arrangement portion 111 functions as an initial arrangement portion for determining the arrangement of the initial effective pixels. Although the initial effective pixels are arranged uniformly in all the pixels in the equalization arrangement unit 111, the initial effective pixels do not necessarily need to be completely equalized, and may be arranged in an initial arrangement such that the initial effective pixels are arranged in a region where the possibility of detecting a person is considered to be high.

Next, when the value of the effective pixel selection flag is 0, the effective pixel selection unit 112 selects and outputs the initial effective pixel position determined by the equalization arrangement unit 111. When the value of the effective pixel selection flag is 1, the detected effective pixel position determined by the effective pixel determination unit 152 described later is selected and output. The value of the effective pixel selection flag at the time of initial setting is 0, and the initial effective pixel position determined by the equalization arrangement unit 111 is selected in the first operation. The equalization arrangement unit 111 and the effective pixel selection unit 112 operate as an effective pixel output unit 110 that outputs the positions of the effective pixels selected from all the pixels.

The thinning-out scanning unit 120 as a scanning unit performs thermal image scanning of 1 step in the left-right direction using only the effective pixels defined at the effective pixel positions selected by the effective pixel selection unit 112, and acquires temperature data of the effective pixels. The thermal image was acquired with the position at which the thermal image was acquired changed by 1 step amount, with the minimum unit of thermal image scanning set to 1 step. When the thermal image scanning of 1 step in the horizontal direction is ended, a 1-dimensional thermal image as temperature data acquired by each effective pixel is obtained.

Fig. 6 is a diagram for explaining a thermal image obtained by the thinning-out scanning unit 120. The thermal image is composed of temperature data Dt obtained by each effective pixel corresponding to the effective pixel index i. The temperature data Dt is expressed in degrees celsius, for example. The number of pixels of the thermal image obtained by the thinned scanning unit 120 is Nmax.

The human similarity function storage unit 131 stores a preset human similarity function. The human similarity calculation unit 132 converts the value of each effective pixel of the thermal image acquired by the thinned-out scanning unit 120 into a human similarity s using a human similarity function. The human similarity function storage unit 131 and the human similarity calculation unit 132 operate as the human similarity generation unit 130 that obtains the human similarity for the effective pixels.

Fig. 7 shows an example of the human similarity function stored in the human similarity function storage unit 131. The human similarity function represents a relationship between the acquired temperature data Dt and the human similarity s. The human similarity is an index for estimating the possibility of the presence of a human being at the position of the corresponding pixel, and is set so that the greater the value in the range of 0 to 1, the higher the possibility of the presence of a human being. The radiant heat of a person also depends on the measurement site, which is approximately around 27 degrees. However, it is considered that the human figure similarity is set to be present even on the low temperature side when there is a human figure in the region of only the effective pixel portion and when the radiation heat is attenuated by wearing clothes. The human similarity function can be defined by a formula or a table format. In fig. 7, P1 is a pixel where a person exists, and P2 is a pixel where a person exists only in a part of the pixels. P2 is called a partially fit pixel, and the person similarity s is determined according to the proportion of persons in the pixel and has a value between 0 and 1.

Fig. 8 is a diagram for explaining the human similarity s obtained by the human similarity calculation unit 132. The human similarity s is generated in correspondence with the effective pixel index i. The human similarity s shown in fig. 8 is an example of a case where the human similarity function of fig. 7 is applied to the thermal image of fig. 6.

The human similarity evaluation unit 141 determines the human similarity s of each effective pixel obtained by using the human similarity function by using the magnitude of the value of the threshold th1 of the human similarity, thereby calculating the human flag f (i) indicating the presence or absence of a human. When the human similarity s is larger than the threshold th1, it is determined that there is a person and a human character f (i) such as "1" is calculated, and when it is smaller than the threshold th1, it is determined that there is no person and a human character f (i) such as "0" is calculated. The human similarity map generating unit 142 generates a human similarity map D including the effective pixel index i and the human character f (i).

Fig. 9 is a diagram for explaining a configuration example of the human similarity map D. The human similarity map D is a structure in which the effective pixel list i is associated with the human character flag f (i) in the effective pixel. Note that the person similarity map D shown in fig. 9 is an example of a case where the threshold th1 for the person similarity s in fig. 8 is set to 0.3. The human similarity threshold th1 is a value determined at the time of design, and is a value determined appropriately by experiments or the like. In the air conditioner, a plurality of operation modes can be set, and different values can be used for each operation mode. This makes it possible to change the characteristic of detecting a person for each operation mode.

In the present embodiment, the human figure similarity map D is configured by associating the effective pixel index i with the human figure flag f (i) in the effective pixel, but the present invention is not limited to this, and for example, it may be configured by associating the effective pixel index i with the human figure similarity s in the effective pixel. In this case, the possibility that a person exists in the corresponding effective pixel is represented not by two values but by more gradation values. In this case, the process of calculating the human character flag f (i) by the human similarity evaluation unit 141 is not necessary. That is, the human similarity map D may numerically indicate the possibility that a human is present in each of the effective pixels.

Next, the pixel weight map generating unit 143 obtains a weight value g (x) from the weight function W (x) defined by equation (2), the effective pixel index i, the position xi of the effective pixel corresponding to the effective pixel index i, and the person mark f (i) by equation (3), and generates a pixel weight map W including the pixel position x and the weight value g (x) of the pixel position. By the equation (3), weight values are generated for all pixels including pixels other than the effective pixels from the human character mark obtained for only the effective pixels.

[ mathematical formula 2]

[ mathematical formula 3]

Fig. 10 is a diagram showing an example of the weight function w (x). The weight function w (x) defines a weight coefficient at a pixel position x opposite to the reference pixel position, and the weight coefficient decreases as the distance from the reference pixel position increases. In the formula (3), the relative pixel position with reference to the position xi of the effective pixel is represented by (x-xi).

Fig. 11 is a diagram for explaining the pixel weight map W. The pixel weight map W is configured to associate the pixel position x of all pixels included in the thermal image acquisition unit 10 with the weight value g (x) at that pixel position.

As described above, the human similarity evaluation unit 141, the human similarity map generation unit 142, and the pixel weight map generation unit 143 function as the pixel weight calculation unit 140 that generates the weight values for all pixels based on the positions of the respective pixels with respect to the effective pixels and the human similarities.

The map classification unit 151 compares the sum of the weight values g (x) in the pixel weight map W with a predetermined threshold value th2, and sets the value of the effective pixel selection flag according to the comparison result. The effective pixel selection flag is set to 1 when the sum of the weight values g (x) is greater than the threshold th2, and is set to 0, which is the value at the time of initial setting, when the sum of the weight values g (x) is equal to or less than the threshold th 2. The threshold th2 is a value determined at the time of design, and is a value determined appropriately by an experiment or the like. In the air conditioner, a plurality of operation modes can be set, and different values can be used for each operation mode. This makes it possible to change the characteristic of detecting a person for each operation mode.

The map classification unit 151 classifies the pixel weight map W according to the following conditions in the order of condition 1 and condition 2, and generates an effective pixel map W'. In addition, the condition 2 is a condition for detection that places emphasis on the head compared to the feet of the person, but is not a necessary condition. By adding the condition 2, the face and the head sensitive to wind can be detected with higher accuracy.

Condition 1: large weight value g (x)

Condition 2: the pixel position x being small (giving priority to the upper pixel)

Next, the effective pixel determination unit 152 selects Nmax pixels, which are the number of effective pixels determined by the available resources, in the order of arrangement of the effective pixel map W 'using the effective pixel map W', determines these as detection effective pixels, which are effective pixels used for the next scan, and outputs the pixel positions of the effective pixels corresponding to the effective pixel indexes.

As described above, the map classification unit 151 and the effective pixel determination unit 152 function as the all-pixel classification unit 150 that selects the detection effective pixels used in the next step from all pixels in descending order of weight value.

As described above, the effective pixel selection unit 112 selects and outputs the initial effective pixel position determined by the equalization arrangement unit 111 when the value of the effective pixel selection flag is 0, and selects and outputs the detected effective pixel position determined by the effective pixel determination unit 152 described later when the value of the effective pixel selection flag is 1.

The thinning-out scanning unit 120 performs the next 1-step scanning in the left-right direction using the effective pixels at the selected pixel positions. Thereafter, until the horizontal scanning reaches either one of the left and right sides, the processes of the human similarity calculation unit 132, the human similarity evaluation unit 141, the human similarity map generation unit 142, the pixel weight map generation unit 143, the map classification unit 151, and the effective pixel determination unit 152 are repeated. When the horizontal scanning reaches either one of the left and right sides, the control unit 14 sets the effective pixel selection flag to 0, which is the value at the time of initial setting, and restarts the operation from the equalization arrangement unit 111. At this time, the scanning position is also temporarily reset.

By assigning pixels preferentially to pixel positions where there is a high possibility of a person being present in this way, dynamic effective pixel arrangement is performed that concentrates on a person-present location. When the horizontal scanning reaches either one of the left and right, a 2-dimensional thermal image to which pixels are preferentially assigned at pixel positions where the possibility of having a person is high is generated.

In addition, in a case where horizontal scanning is performed in the reverse direction from the position after the horizontal scanning reaches one of the left and right ends, for example, in a case where horizontal scanning is performed in the right direction and then horizontal scanning is continuously performed in the left direction from the right end, since the scanning positions are close to each other, it is not always necessary to set the effective pixel selection flag to 0 which is the value at the time of initial setting in the first 1-step scanning. On the other hand, in the case where the scanning position is temporarily reset after the horizontal scanning reaches one of the left and right ends and then the horizontal scanning is performed in the same direction, for example, in the case where the scanning position is temporarily returned to the left end after the horizontal scanning is performed in the right direction and then the horizontal scanning in the right direction is started again, the scanning position is not close, and therefore, it is preferable that the effective pixel selection flag is set to 0 which is the value at the time of initial setting in the scanning of the first 1 step. In the thermal image sensor according to the present embodiment, the scanning position is once reset after the horizontal scanning reaches one of the left and right ends, and then the horizontal scanning is performed in the same direction.

Fig. 12 is a diagram for visually explaining a dynamic effective pixel arrangement. In the figure, the squares represent pixels, where pixels filled in gray represent active pixels that are dynamically assigned. Since a large number of effective pixels are allocated to an area where there is a high possibility of a person as shown in the figure, a thermal image with high resolution can be acquired, and highly accurate person detection can be performed to the extent that part detection is possible. A thermal image based on a dynamic effective pixel configuration can be obtained from the thermal image sensor as described above. In the obtained thermal image, since the temperature data is not acquired for the pixels other than the effective pixels, the temperature data acquired by the effective pixels can be recognized by setting the temperature data to a predetermined value.

In the first 1 step of horizontal scanning, the thermal image sensor according to the present embodiment determines effective pixels equally spaced from all pixels in the vertical direction based on the number of pixels that can be used, which is determined from the resources that can be used, and performs 1 step of data acquisition in the vertical direction of the thermal image using only the effective pixels. The thermal image sensor further includes a human similarity function storage unit as a database for associating temperature data based on a human radiative thermal model with human similarity, and when 1-step scanning in the horizontal direction is completed, the thermal image sensor determines whether or not a human is present at the effective pixel position by calculating the human similarity from the detected temperature data of the effective pixels and performing threshold processing on the human similarity, and performs rearrangement of the effective pixels by weighting so that a part of the human includes a large number of effective pixels, and uses the result in the next step.

As a result, a large number of effective pixels are dynamically allocated to a region with a high probability of human presence while maintaining the number of effective pixels used for the entire region constant, and therefore, it is possible to perform highly accurate human detection using the same resources as those used for low pixels. By applying the present thermal image sensor, it is possible to realize an air conditioner having a function of detecting the position of a human face, hands and feet, etc., for example, and blowing air while avoiding the human face or warming the feet with emphasis on the temperature.

As described above, in the thermal image sensor and the air conditioner according to the present invention, the thermal image sensor scans the thermal image acquiring unit that acquires temperature data for each pixel arranged in the vertical direction in units of 1 step in the horizontal direction, and synthesizes 1-dimensional thermal images acquired in each step to acquire a 2-dimensional thermal image, and the thermal image sensor includes: an effective pixel output unit that outputs a position of a detection effective pixel selected as an effective pixel from the full pixels included in the thermal image acquisition unit in the immediately preceding step; a scanning unit for performing 1-step scanning to acquire temperature data of the effective pixels; a human figure similarity generating unit that obtains a human figure similarity for the effective pixels using a relationship between a human figure similarity indicating a possibility of a human being and the temperature data; a pixel weight calculation unit that generates weight values for all pixels based on the relative positions of the respective pixels with respect to the effective pixels and the human similarity; and a full-pixel classification unit that selects detection effective pixels to be used in the next step from the full pixels in descending order of weight value, so that highly accurate person detection can be performed without increasing the memory capacity and enhancing the CPU capability.

Embodiment mode 2

Fig. 13 is a diagram for explaining the operation of the control unit 14 included in the thermal image sensor according to embodiment 2 of the present invention, and is different from the control unit according to embodiment 1 in that the pixel weight map generation unit 143 operates, and in that the pre-weight coefficient generation unit 160 includes a pre-entire pixel map storage unit 161 and a pre-entire pixel weight map generation unit 162. The thermal image sensor of the present embodiment has the same configuration as that of the thermal image sensor of embodiment 1, and includes a thermal image acquisition unit 10 and a control unit 14. The thermal image acquisition unit 10 is the same as that in embodiment 1 described above.

Fig. 14 is a diagram showing an example of the prior whole pixel map stored in the prior whole pixel map storage unit 161. The prior whole pixel map M is a 2-dimensional map indicating the position of a pixel having a high frequency selected as an effective pixel in a predetermined time period until the end of the latest one-way scan in the left-right direction, and is a map in which the position of each pixel is associated with a pixel value M (x, y). The pixel with a high frequency selected as the effective pixel has a value of m (x, y) 1, and the other pixels have a value of m (x, y) 0. Here, x is a position in the vertical direction of the pixel, and y is a position in the horizontal direction of the pixel.

The pre-whole pixel map storage unit 161 stores the number of times each pixel is used as an effective pixel in a predetermined period. The number of times is compared with a predetermined threshold th3, and the pixel position with the high frequency of being selected as the effective pixel is determined, thereby generating the prior whole pixel map M. The predetermined period and the threshold th3 are values determined at the time of design, and are values determined appropriately by experiments or the like. In the air conditioner, a plurality of operation modes can be set, and different values can be used for each operation mode. This makes it possible to change the characteristic of detecting a person for each operation mode. When only the immediately preceding one-way scanning period is set to a predetermined period, the position of the effective pixel in the scanning is a pixel position with a high selection frequency.

The prior whole pixel weight map generation unit 162 obtains the prior weight coefficient g2(x, y) by equation (4) using the prior whole pixel map M, and calculates the prior whole pixel weight map W2 including the pixel position and the prior weight coefficient g2(x, y) of the pixel position.

[ mathematical formula 4]

Fig. 15 is a diagram showing an example of the prior whole pixel weight map W2. The prior whole pixel weight map W2 is a 2-dimensional map having a value of 1 or 0.5 in accordance with the position of each pixel. As described above, the prior whole pixel map storage unit 161 and the prior whole pixel weight map generation unit 162 operate as the prior weight coefficient generation unit 160, and the prior weight coefficient generation unit 160 generates the prior weight coefficient having a larger value for the pixel selected as the higher frequency of the effective pixel in the past predetermined period.

The pixel weight map generating unit 143 generates the weight values g (x) for all pixels in the up-down direction at the respective scanning positions y0 in the left-right direction, as in the pixel weight map generating unit in embodiment 1. Further, a new weight value g3(x) is obtained by the following expression (5) using the weight value g (x) and the prior weight coefficient g2(x, y0) at the scanning position y0, and a pixel weight map W including the pixel position x in the vertical direction and the weight value g3(x) at the pixel position is generated. The subsequent operations are the same as those in embodiment 1.

[ math figure 5]

g3(x)＝g(x)*g2(x，y0) (5)

As described above, the thermal image sensor and the air conditioner according to the present embodiment include the prior weight coefficient generation unit that generates the prior weight coefficient, and the pixel weight calculation unit further generates the weight value using the prior weight coefficient that becomes a larger value for the pixel having the higher frequency of being selected as the effective pixel in the past predetermined period. By determining the effective pixels for acquiring the thermal image by reflecting the selection result of the effective pixels in the past predetermined period, the detection accuracy can be further improved. In most cases, there is regularity in the range of motion of a person. For example, in the case where a sofa is placed in a living room, there is a high possibility that a person is present at the position where the sofa is placed. Therefore, by using the accumulation of the past selection results, it is possible to preferentially acquire a thermal image at a pixel position where there is a higher probability of a human being, and the detection accuracy is further improved. Further, the same effects as those of the thermal image sensor and the air conditioner in embodiment 1 are obtained.

Claims (11)

1. A thermal image sensor that acquires a 2-dimensional thermal image by scanning a thermal image acquisition unit that acquires temperature data for each pixel arranged in a predetermined direction in units of 1 step in a direction perpendicular to the predetermined direction and synthesizing 1-dimensional thermal images acquired in the respective steps, the thermal image sensor comprising:

an effective pixel output unit that outputs a position of a detection effective pixel selected as an effective pixel from all pixels included in the thermal image acquisition unit in an immediately preceding step;

a scanning unit which performs the scanning of 1 step to acquire the temperature data of the effective pixel;

a pixel weight calculation unit that generates a weight value for each of the full pixels based on the temperature data and a relative position of each of the pixels with respect to the effective pixel; and

and a full-pixel classification unit that selects the detection effective pixels used in the next scanning step from the full pixels in descending order of the weight values.

2. The thermal image sensor of claim 1,

the thermal image sensor includes a human similarity generating unit that obtains the human similarity for the effective pixels using a relationship between the human similarity, which is an index indicating a possibility of a human being, and the temperature data,

the pixel weight calculation unit generates weight values for the whole pixels based on the relative positions of the respective pixels with respect to the effective pixels and the human figure similarity.

3. The thermal image sensor of claim 2,

the human similarity generation unit includes:

a human similarity function storage unit that stores a human similarity function indicating a relationship between the human similarity and the temperature data; and

and a human similarity calculation unit configured to calculate the human similarity for the effective pixels using the temperature data of the effective pixels and the human similarity function.

4. The thermal image sensor of claim 3,

the human similarity function storage unit stores a relationship between the human similarity and the temperature data in a table format.

5. The thermal image sensor according to any one of claims 1 to 4,

the thermal image sensor includes a prior weight coefficient generation unit that generates a prior weight coefficient having a larger value for a pixel selected as the effective pixel with a higher frequency in a past predetermined period,

the pixel weight calculation unit further generates the weight value using the prior weight coefficient.

6. The thermal image sensor of claim 1,

the prescribed direction is a vertical direction,

when the weight values are the same value, the all-pixel classification unit selects pixels located above in order.

7. The thermal image sensor of claim 1,

the effective pixel output unit includes:

an initial arrangement unit configured to select initial effective pixels from the full pixels so that the initial effective pixels are in a predetermined initial arrangement; and

and an effective pixel selection unit that outputs the initial effective pixels as effective pixels at the time of initial setting, and outputs the detection effective pixels as effective pixels at a time other than the time of initial setting.

8. The thermal image sensor of claim 7,

the initial arrangement section is an equal arrangement section that selects the initial effective pixels so that the initial effective pixels are arranged equally among the all pixels.

9. The thermal image sensor according to claim 7 or 8,

the initial setting is performed when the sum of the weight values is equal to or less than a predetermined threshold value.

10. The thermal image sensor of claim 7,

the initial setting is set when the scanning is performed in the first 1 step of the scanning.

11. An air-conditioning apparatus is provided, in which,

a thermal image sensor according to claim 1.