JP2017183904A - Region detection device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem that it is difficult to detect a region generated by a puddle while distinguishing from a ground surface which is made wet with water and on which no puddle is formed.SOLUTION: Imaging means 20 has sensitivities at least for a light of an absorption wavelength band to be absorbed by water and a light of a predetermined non-absorption wavelength band different from the absorption wavelength band, successively images a predetermined space and outputs images. Change degree calculation means 40 compares images picked up before and after and calculates an absorption band change degree that is a degree of a change of a light intensity in the absorption wavelength band and a non-absorption band change degree that is a degree of a change of a light intensity in the non-absorption wavelength band for each of pixels. Water region detection means 41 detects a water region consisting of pixels for which a degree of a difference between the absorption band change degree and the non-absorption band change degree exceeds a predetermined reference.SELECTED DRAWING: Figure 2

Description

  The present invention is caused by an area detecting device that detects an area where water is present in an image of a predetermined space, an area detecting device that detects an area in which a water pool is dynamically formed, and an entity such as a person. The present invention relates to an area detection device that detects a changed area separately from an area generated by an image reflected in a puddle.

  For the purpose of crime prevention or the like, a device that detects an intruder from an image obtained by capturing a surveillance space with a surveillance camera is widely used. In such a device, a change area caused by an intruder is detected by features such as size, shape, and movement, but the change area is also caused by various disturbances having characteristics similar to those of the intruder. In addition, a change area due to disturbance may overlap with a change area caused by an intruder, making it difficult to detect the characteristics of the intruder. Therefore, research is being conducted to prevent false detections caused by these disturbances.

  For example, Patent Document 1 describes an intruding object detection device that prevents erroneous detection due to planting. In this intruding object detection apparatus, a planting area designated by an operator or the like is stored, or a changed area in which many edge pixels are frequently changed is stored as a planting area.

  And in this intruding object detection device, in order to accurately determine the changing area due to the intruding object moving before planting, the intruding object likelihood of the changing area overlapping the planting area overlaps with the planting area. It is determined whether or not the change area is caused by an intruding object by adjusting the lower change area.

JP 2012-88661 A

  By the way, an image of a person reflected in a puddle can be generated as a disturbance in the monitoring space. However, in the prior art, as described below, it is difficult to set or detect a puddle region and to prevent erroneous detection due to an image reflected in the puddle.

  That is, the puddle appears dynamically and disappears dynamically, so that it is difficult for an operator or the like to set a puddle area in advance. In addition, most of the puddles are caused by rain and accompanied by a change in which the ground gets wet. However, wet ground where no image is shown needs to be distinguished from a puddle. However, similar image changes occur when the dry ground gets wet and when a puddle occurs on the wet ground. Therefore, it is difficult to distinguish and detect a wet ground and a region where a puddle is generated.

  The image of the person reflected in the puddle has a size, shape, movement, and the like that are very similar to those of the intruder. For this reason, it is difficult to adjust the likelihood of an intruding object to be low with respect to the change area due to this image. As a result, if this image is an image of a person outside the monitoring space reflected in a puddle in the monitoring space, it will be erroneously detected as an intruder. In addition, when an intruder passes through a puddle, the image and substance reflected in the puddle are integrated, and the change area having the characteristics of a person is detected while changing its size rapidly, and can be traced as a single intruder. It becomes difficult.

  The present invention has been made in view of the above problems, and an object of the present invention is to provide an area detection device that can detect an area caused by a puddle or the like by distinguishing it from an area of wet ground. Another object of the present invention is to provide an area detection device capable of accurately detecting a puddle area that dynamically appears and disappears. Another object of the present invention is to provide an area detection device that can detect a change area generated by an entity such as a person by distinguishing the change area from an image reflected in a puddle.

  In order to solve such a problem, the region detection device of the present invention has sensitivity to at least light in an absorption wavelength band absorbed by water and light in a predetermined non-absorption wavelength band different from the absorption wavelength band, and a predetermined space. The image pickup means for sequentially picking up images and outputting the image is compared with the image picked up before and after, and the degree of change of the light intensity in the absorption wavelength band and the degree of change in the absorption wavelength band for each pixel in the non-absorption wavelength band A change degree calculating means for calculating a non-absorption band change degree that is a degree of change in light intensity, and detecting a water region composed of pixels in which a difference degree between the absorption band change degree and the non-absorption band change degree exceeds a predetermined reference A water region detecting means.

  Further, in the area detection device according to the present invention, the water area storage means for sequentially storing the water areas and the water areas detected by comparison before and after are extracted to extract a combination of water areas having an inclusion relationship. It is preferable to further comprise a puddle determination unit that determines a water region whose area change degree exceeds a predetermined area change threshold as a puddle region.

  In the region detection device according to the present invention, the imaging means further outputs an image having sensitivity in the visible light or near-infrared wavelength band, and further compares the images captured before and after. When the change area overlaps the puddle area, the change area detection means for detecting the change area in the visible light or near infrared wavelength band, the puddle area storage unit for storing the puddle area, The farthest position where the corresponding space position is farthest from the imaging means is set, and the farthest position is used to determine the unreal area by the image reflected in the puddle in the change area and the real object in the change area. It is preferable to include an entity region discriminating unit that discriminates the entity region.

  The area detection apparatus according to the present invention further includes object information storage means for storing the object size of the object moving in the space, and the entity area determination means has the change area overlapping the puddle area as the object size. If it is larger than the above, it is preferable that the partial area of the object size including the farthest position in the change area is determined as the substantial area.

  Further, in the area detection device according to the present invention, the actual area determination means determines that the change area is insubstantial when the farthest position of the change area overlapping the puddle area is on the approximate area boundary of the puddle area. It is preferable to distinguish the region.

  According to the region detection apparatus of the present invention, it is possible to accurately detect a water region caused by a puddle or the like by distinguishing it from a wet ground region. Moreover, according to the area | region detection apparatus of this invention, the water pool area | region which is a water area | region which appears and disappears dynamically can be detected with a sufficient precision. Further, according to the region detection apparatus of the present invention, it is possible to detect a change region caused by an entity by distinguishing it from an image caused by an image reflected in a puddle.

It is a block diagram showing the structure of the intrusion detection apparatus by this embodiment. It is a block diagram showing the function of the area | region detection apparatus by this embodiment. It is a numerical example of the light intensity in an absorption band and a non-absorption band. It is a numerical example of the ratio (difference) of an absorption band change degree, a non-absorption band change degree, and a change degree. It is an image figure which concerns on the process example of a water pool area | region determination. It is an image figure which shows the positional relationship of a water pool area | region and a non-substance area | region. It is an image figure which concerns on the process example of an entity area | region discrimination | determination. It is an image figure which concerns on another example of a process of entity area | region discrimination | determination. It is a flowchart which shows the flow of the process which the intrusion detection apparatus by this embodiment performs. It is a flowchart which shows the flow of the water area | region detection process which the area | region detection apparatus by this embodiment performs. It is a flowchart which shows the flow of the water pool area | region determination process which the area | region detection apparatus by this embodiment performs. It is a flowchart which shows the flow of the actual area | region discrimination | determination process which the area | region detection apparatus by this embodiment performs.

  Hereinafter, an intrusion detection device in which an area detection device according to an embodiment of the present invention is mounted will be described. The intrusion detection device according to the present embodiment detects an intruder who has entered the monitoring space and outputs an alarm signal. The intrusion detection device sequentially inputs monitoring images obtained by photographing the monitoring space to the region detection device. The region detection device detects the water region, determines whether the detected water region is a puddle region, Detection and determination of an actual region portion that is not a non-substantive region based on an image reflected in a puddle among the detected change regions. The intrusion detection device outputs an alarm signal when an actual region by a person is detected from the monitoring image.

[Configuration of Intrusion Detection Device 1]
FIG. 1 is a block diagram showing a schematic configuration of an intrusion detection apparatus 1 according to the present embodiment. The intrusion detection device 1 includes an imaging unit 2, a storage unit 3, an image processing unit 4, and an output unit 5.

  The imaging unit 2 is a so-called multispectral camera, and includes an imaging element such as a CCD element or a C-MOS element, an optical system component, an A / D converter, and the like. The imaging unit 2 is connected to the image processing unit 4, sequentially captures a predetermined monitoring space, generates a monitoring image, and inputs each monitoring image to the image processing unit 4.

  The optical system component of the imaging unit 2 includes a near-infrared cut filter, a bandpass filter that passes only 970 nm, which is a wavelength band absorbed by water (absorption wavelength band, hereinafter referred to as absorption band), and an absorption wavelength band of water A non-absorption wavelength band (non-absorption wavelength band; hereinafter referred to as a non-absorption band), a band-pass filter that passes only 900 nm, and a filter switching unit that switches these at a predetermined period are included.

  For example, the imaging unit 2 captures a monitoring image with an imaging period of 1/5 second, and switches to a band-pass filter that passes only the absorption wavelength band at a rate of once every 30 minutes, thereby performing imaging. An image (hereinafter referred to as an absorption band image) is generated, and a non-absorption band image (hereinafter referred to as a non-absorption band image) is generated by switching to a bandpass filter that passes only the non-absorption band after 1/5 second. In other times, the image is switched to the near-infrared cut filter to capture an image in the visible light band (hereinafter referred to as a visible image). The absorption band image and the non-absorption band image are used for detection of the water region, and the visible image is used for detection of the real region.

  The imaging unit 2 is fixedly installed and captures an absorption band image, a non-absorption band image, and a visible image having the same size. Therefore, the corresponding pixels between these images are obtained by imaging the same position in the monitoring space.

  In addition to 970 nm, the absorption bands of 1.94 μm, 1.45 μm, and 1.19 μm are known, and when using an InGaAs imaging device or the like, these wavelength bands can also be used. Further, the non-absorption band can be appropriately selected and set from the range of 340 nm to 940 nm in addition to 900 nm.

  The switching cycle is an example, and a cycle shorter or longer than 30 minutes can be set as appropriate according to the purpose of use of the water area to be detected and the environment of the monitoring space (such as drainage).

  Further, instead of switching the filter, the illumination can be switched to generate an image in each wavelength band. For example, the imaging unit 2 includes an absorption band illuminating unit that irradiates only light in the absorption band, a non-absorption band illuminating unit that irradiates only light in the non-absorption band, and an illumination switching unit that switches between turning these lights on and off at a predetermined cycle. Each illumination means can be realized by LED illumination or the like. The imaging unit 2 generates an absorption band image by performing a difference process between the monitoring image obtained by turning on the absorption band illumination unit and imaging the monitoring space and the monitoring image obtained by turning off both illumination units and imaging the monitoring space. Further, the imaging unit 2 generates a non-absorption band image by performing a difference process between the monitoring image obtained by turning on the non-absorption band illumination unit and imaging the monitoring space and the monitoring image obtained by turning off both illumination units and imaging the monitoring space. Moreover, the imaging part 2 produces | generates a visible image by light-extinguishing both illumination means and imaging a monitoring space.

  The imaging unit 2 may capture a near-infrared image having sensitivity only to near-infrared light instead of a visible image. In this case, the imaging unit 2 includes a near-infrared illumination unit that irradiates only near-infrared light in addition to the above-described absorption band illumination unit and non-absorption band illumination unit, and the illumination switching unit turns off these three lights. Switch at a predetermined cycle. And the imaging part 2 produces | generates a near-infrared image by lighting a near-infrared illumination means and imaging a monitoring space.

  Further, the imaging unit 2 may capture and output a visible image during the day and a near-infrared image during the night as images used for detection of the substantial area.

  As described above, the imaging unit 2 has sensitivity to at least light in an absorption wavelength band that is absorbed by water and light in a predetermined non-absorption wavelength band different from the absorption wavelength band, and sequentially captures an image by imaging a predetermined space. Is output. Further, the imaging unit 2 may further output the image with sensitivity in the visible light or near-infrared wavelength band.

  The storage unit 3 is a storage device such as a ROM or a RAM. The storage unit 3 stores various programs and various data used in the image processing unit 4, and inputs / outputs such information to / from the image processing unit 4.

  The image processing unit 4 is configured using an arithmetic device such as a CPU, DSP, or MCU, and is connected to the imaging unit 2, the storage unit 3, and the output unit 5. The image processing unit 4 functions as each unit to be described later by reading and executing a program from the storage unit 3. The image processing unit 4 processes the monitoring image from the imaging unit 2 and outputs an alarm signal to the output unit 5 when an intruder is detected from the monitoring image.

  The output unit 5 is a communication interface circuit that connects the image processing unit 4 and an external device. For example, the output unit 5 is a communication device that communicates with a server in the monitoring center, and transmits an alarm signal input from the image processing unit 4 to the server.

  FIG. 2 is a schematic functional block diagram of the region detection apparatus according to the present embodiment. The imaging unit 2 functions as the imaging unit 20. The storage unit 3 functions as a background image storage unit 30, a water region storage unit 31, a puddle region storage unit 32, and an object information storage unit 33. The image processing unit 4 functions as a degree-of-change calculating unit 40, a water region detecting unit 41, a water pool determining unit 42, a changed region detecting unit 43, and a substance region determining unit 44.

  The imaging unit 20 images the monitoring space, inputs the absorption band image and the non-absorption band image to the change degree calculation unit 40, and inputs the visible image to the change area detection unit 43.

  The background image storage unit 30 stores an absorption band image, a non-absorption band image, and a visible image captured in the past from the current time as a background image. Hereinafter, the background image of the absorption band is referred to as an absorption band background image, the background image of the non-absorption band is referred to as a non-absorption band background image, and the background image of the visible light band is referred to as a visible background image.

  The degree-of-change calculating means 40 calculates a degree of change indicating the degree of change in the luminance value in each pixel for each of the absorption band and the non-absorption band, and uses the current time image for each of the absorption band and the non-absorption band. The background image in the image storage means 30 is updated.

  Specifically, the degree-of-change calculating unit 40 reads the absorption band background image from the background image storage unit 30, and for each corresponding pixel, the luminance value Bw of the read absorption band background image and the current value input from the imaging unit 20. The ratio (Bw / Iw) of the luminance value Iw of the absorption band image at the time is calculated, and the ratio calculated for each pixel is output to the water region detection means 41. A ratio in each pixel is a change degree Vw in the absorption band in the pixel (hereinafter referred to as an absorption band change degree).

  The degree-of-change calculation means 40 reads the non-absorption band background image from the background image storage means 30 and, for each corresponding pixel, the luminance value Bn of the read non-absorption band background image and the current time input from the imaging means 20. The ratio (Bn / In) of the luminance value In of the non-absorption band image is calculated, and the ratio calculated for each pixel is output to the water region detection means 41. A ratio in each pixel is a change degree Vn in the non-absorption band in the pixel (hereinafter referred to as a non-absorption band change degree).

  In this way, the degree-of-change calculating means 40 compares the images captured before and after, and for each pixel, the degree of change in the light intensity in the absorption wavelength band of water and the degree of change in the absorption band and the non-absorption wavelength band. A non-absorption band change degree that is a degree of change in light intensity is calculated, and the absorption band change degree and the non-absorption band change degree are output to the water region detection means 41.

  The degree-of-change calculating unit 40 replaces the absorption band background image and the non-absorption band background image stored in the background image storage unit 30 with the absorption band image at the current time and the non-absorption band image at the current time, respectively. Update the background image.

  The water region detection unit 41 detects a water region where water is present in the monitoring space using the absorption band variation and the non-absorption band variation input from the variation calculation unit 40. The storage unit 31 sequentially stores the water areas detected by the water area detection unit 41.

  Specifically, the water region detection unit 41 calculates the ratio R (= Vw / Vn) of the absorption band change degree Vw and the non-absorption band change degree Vn for each corresponding pixel, and calculates the calculated ratio R of each pixel. Pixels that exceed a predetermined threshold value Tr are extracted. In this case, pixels whose ratio R is equal to or less than the threshold value Tr are not extracted. Then, the water area detecting means 41 detects each group of neighboring pixels among the extracted pixels as a water area, outputs information of the detected water area to the water pool determining means 42, and detects the detected water area. The area information is added to the water area storage means 31.

  With reference to FIG. 3 and FIG. 4, the principle of detecting a pixel of interest capturing a water pool as a pixel in the water region will be described.

  A numerical example 200 in FIG. 3 shows the light intensity 201 in the absorption band in the target pixel and the light intensity 202 in the non-absorption band in the target pixel. The three numerical values shown for each of the light intensity 201 and the light intensity 202 are, in order from the top, the light intensity at the time when the position in the monitoring space corresponding to the target pixel is dry ground, and the same position is wet. However, the light intensity at the time when the puddle was not formed (it was wet ground) and the light intensity at the time when the puddle was formed at the same position are shown.

  A numerical example 300 in FIG. 4 is calculated from the numerical example 200, the degree of change 301 in the absorption band in the target pixel, the degree of change 302 in the non-absorption band in the target pixel, the degree of change 301, and the degree of change 302 calculated from A change ratio 303 in the pixel is shown. The two numerical values shown for each of the degree of change 301 and the degree of change 302 are, in order from the top, the degree of change when the position in the monitoring space corresponding to the target pixel changes from dry ground to wet ground, and the same The degree of change when the position changes from a wet ground to a puddle is shown. Further, the two numerical values shown in the change ratio 303 are the change ratio when the position in the monitoring space corresponding to the target pixel is changed from the dry ground to the wet ground, and the same position in order from the top. The ratio of the degree of change when changing from wet ground to a puddle is shown.

  In the degree of change 301 in the absorption band, when the two degrees of change are compared, the ratio of 1.64 and 1.75 is 0.94, and there is no significant difference between the two. Therefore, it can be seen that it is difficult to discriminate whether it has changed from a dry ground to a wet ground or from a wet ground to a puddle using only the light intensity in the absorption band.

  On the other hand, when the two ratios are compared, the ratio of 1.18 and 1.58 is 1.35, which is significantly different from 0.94 of the degree of change 301. Therefore, if the ratio of change is used, it becomes easy to discriminate whether it has changed from a dry ground to a wet ground or from a wet ground to a puddle.

  The threshold value Tr for discriminating these can be set to 1.25, for example. The pixel of interest is not extracted as a pixel in the water region because the ratio of the degree of change is 1.18 when it changes from dry ground to wet ground, and it is less than 1.25, and changes from wet ground to a puddle. At that time, the ratio of the degree of change is 1.58, which exceeds 1.25 and is extracted as a pixel in the water region.

  Further, a water region that has directly changed from a dry state to a puddle without going through a wet state can be detected by the ratio of the degree of change. Incidentally, referring to the numerical examples in FIG. 3 and FIG. 4, when the target pixel changes from a dry state to a puddle, the ratio of the degree of change is 1.85 and exceeds 1.25. Are extracted as pixels.

  The ratio of change may be the ratio (Vn / Vw) of the non-absorption band change degree Vn and the absorption band change degree Vw by exchanging the numerator and the denominator. Pixels whose degree ratio falls below the threshold value 1 / 1.25 are extracted as pixels in the water region.

  Further, instead of the ratio of the degree of change, a difference in degree of change or an absolute value of the difference in degree of change can be used. When using the difference (vw−vn) in the degree of change, the water region detection unit 41 extracts pixels in which the difference in degree of change exceeds the threshold and exceeds the threshold value as pixels in the water region. When the difference in change degree (vn−vw) is used, the water area detection unit 41 extracts pixels whose difference in change degree falls below the threshold value as pixels in the water area. When the absolute value of the difference in change is used, the water area detection unit 41 extracts pixels whose absolute value of the difference in change exceeds the threshold value as a pixel in the water area.

  In other words, the ratio, difference, or absolute value of the difference is the difference indicating the difference between the absorption band change and the non-absorption band change, and the water region detection means 41 has a difference exceeding a predetermined standard. Pixels are extracted as pixels in the water region.

  In this way, a water region composed of pixels in which the degree of difference between the absorption band change degree and the non-absorption band change degree exceeds a predetermined reference is detected, and the water area is output to the puddle determination means 42 and the water area is It is stored in the area storage means 31.

  As described above, the area detecting device including the degree-of-change calculating means 40, the water area detecting means 41, and the background image storage means 30 has the difference between the absorption band change degree and the non-absorption band change degree. By detecting the water area based on the above, it is possible to detect the water area where the object image can be reflected with high accuracy without misdetecting the area where the image of the object is not reflected due to being wet.

  Here, in an outdoor monitoring space, there is a case where vegetation containing a lot of moisture is detected as a water region. The puddle judging means 42 pays attention to the difference that the vegetation can shake but the puddle does not move, and the difference that when the vegetation does not shake, the size does not change, but the puddle gradually increases or decreases. The detection of such plants and the like is prevented by the processing described below.

  That is, the puddle determination unit 42 compares the water region at the current time input from the water region detection unit 41 with the water region detected in the water region storage unit 31 and has a containment relationship. Extract a combination of regions. Then, the water pool determination means 42 determines that the water area that does not have the inclusion relation with the detected water area among the water areas at the current time is not a water pool. Furthermore, the puddle determination means 42 calculates the degree of change in the area between the water regions within the extracted combination, and determines that the water region in which the degree of change in the area exceeds a predetermined area change threshold is a puddle region. Then, it is determined that a water region whose area change degree does not exceed the area change threshold value is not a puddle region.

  In order to accommodate both a gradually increasing and gradually decreasing pool, the change in area can be defined as the ratio of the smaller area to the larger water area. The area change threshold is 1.1, for example, and when the area ratio exceeds 1.1 and exceeds, it can be determined as a puddle region. Also, the magnitude relationship in calculating the area ratio may be reversed, in which case the area change threshold is, for example, 1 / 1.1, when the area ratio is below 1/1. It can be determined as a water pool region. The above threshold is an example, and can be determined as appropriate according to the value of the water region detection period, the resolution, and the like.

  Instead of the area ratio, the absolute value of the area difference may be calculated. In this case, normalization such as dividing the absolute value of the difference by the smaller area is preferably performed, and the area change threshold can be set to 0.1, for example. The size relationship of areas used for normalization may be determined in reverse. In that case, the area change threshold can be set to 0.1 / 1.1 or the like. Alternatively, instead of the size of the area, the ratio of the areas or the absolute value of the difference between the areas may be calculated in order of time, and the gradually increasing pool and the gradually decreasing pool may be distinguished and detected.

  In this way, the water pool determination means 42 compares the water areas detected at the preceding and following times, extracts a combination of water areas having an inclusion relationship, and the degree of change in area of the extracted combinations is a predetermined area. A water region exceeding the change threshold is determined as a puddle region.

  The puddle area storage means 32 is a means for storing the puddle area. The puddle determination unit 42 stores the puddle region in the puddle region storage unit 32 for use in the determination by the substance region determination unit 44.

  FIG. 5 illustrates the state of the water pool region determination process.

  In the vicinity of time ta, a puddle is formed due to rain or the like, and it gradually increases. At time ta, the water area detection means 41 detects the water area 400 and stores it in the water area storage means 31. At the subsequent time (ta + 1), the water area 401 is detected by the water area detecting means 41 and input to the water pool determining means 42. Since the water region 400 included in the water region 401 is stored in the water region storage unit 31, the water pool determination unit 42 extracts a combination of the water region 401 and the water region 400 (area of the water region 401). / (Area of the water region 400) is larger than 1.1, so the water region 401 is determined as a puddle region.

  Moreover, it rains near the time tb, and the puddle gradually decreases. At time tb, the water area 402 is detected by the water area detection means 41 and stored in the water area storage means 31. At the subsequent time (tb + 1), the water region detection unit 41 detects the water region 403 and inputs this to the water pool determination unit 42. Since the water region 402 including the water region 403 is stored in the water region storage unit 31, the water pool determination unit 42 extracts a combination of the water region 402 and the water region 403, and (area of the water region 402) / Since (the area of the water region 403) is greater than 1.1, the water region 403 is determined as a puddle region.

  On the other hand, even if a vegetation area is stored in the water area storage means 31 as a water area, a combination of water areas having an inclusive relationship is not extracted if the vegetation shakes, so the vegetation area is not determined to be a puddle area. Further, even if the vegetation does not shake, the size of the vegetation does not change in about 30 minutes, so the area ratio is 1.1 or less, and the area of the vegetation is not determined as a puddle area.

  As described above, the region detection device including the degree-of-change calculation unit 40, the water region detection unit 41, the water pool determination unit 42, the background image storage unit 30, and the water region storage unit 31 has an inclusion relationship. By determining the puddle area based on the determination of the degree of change in the area, the water puddle area where the image of the object can be reflected by collecting water can be detected without erroneously detecting the water area due to water or other plants. It can be detected with high accuracy.

  The change area detection unit 43 performs background difference processing between the visible image at the current time input from the imaging unit 20 and the visible background image past from the current time stored in the background image storage unit 30, and A change area in the visible image is detected, and the detected change area is output to the real area determination means 44. Further, the visible background image is updated using the visible image at the current time. A background correlation process may be performed instead of the background difference process. Further, as described above, a near infrared image can be used instead of the visible image.

  That is, the change area detection unit 43 compares images taken at different times, detects a change area in the wavelength band of visible light or near-infrared light, and outputs the change area to the real area determination unit 44.

  The change area detected in this manner includes a change area caused by an entity such as an intruder (hereinafter referred to as an entity area), and an image reflected in a puddle, that is, a change area caused by an entity that is not present (hereinafter referred to as an intangible area). Can be included.

  Here, as shown in FIG. 6, in the image captured by the imaging unit 20, even though a non-substance region due to the image 502 reflected in the puddle 501 between the entity 500 and the imaging unit 20 may appear, A non-substance area does not appear in the area of the water pool 503 that is not between 500 and the imaging means 20. That is, the non-substance region always appears on the side closer to the position corresponding to the imaging unit 20 on the image captured by the imaging unit 20.

  The entity area discriminating means 44 discriminates the entity area and the entity area by a process that pays attention to the property that the entity area overlaps the water pool area and appears closer to the position corresponding to the imaging means 20.

  That is, when the change area input from the change area detection means 43 overlaps with the puddle area stored in the puddle area storage means 32, the entity area determination means 44 determines the corresponding monitoring space in the change area. The farthest position is set to the farthest position from the imaging means 20, and the farthest position is used to determine the non-substance area in the change area and the substantive area in the change area.

  If the imaging means 20 is a normal wide-angle camera, the entity area determination means 44 sets the position of the uppermost pixel (the position of the pixel with the smallest Y coordinate) among the pixels in the change area as the farthest position. . Alternatively, if the imaging unit 20 is a fish-eye camera, the entity region determination unit 44 sets the position of the pixel farthest from the image center among the pixels in the change region as the farthest position.

  On the other hand, the entity area discriminating means 44 discriminates a change area that does not overlap with the water pool area as an entity area. Then, the entity area discriminating unit 44 stores the width and height of the change area determined as the entity area in the object information storage unit 33, and the object information storage unit 33 sets the width and height of the change area, that is, the monitoring space. The size of the object moving inside (hereinafter referred to as object size) is stored. The object information storage means 33 stores a standard size of a person as an initial value of the object size.

  The object size can be expressed in terms of the number of pixels when the monitoring space is relatively narrow, but it is preferable that the object size be an actual size. For this purpose, the storage unit 3 stores the camera parameters of the image pickup unit 2 in advance, and the real area discrimination means 44 performs reverse perspective transformation on the coordinate values of the circumscribed rectangle of the change area using the camera parameters to calculate the actual size object size. . The standard size in actual size can be set to 170 cm × 60 cm, for example.

  In addition, in order to refer to the object size detected when it does not overlap with the puddle area at the time when it overlaps with the puddle area, the entity area discriminating means 44 detects the change area detected at the preceding and following times. A tracking process for associating the change regions whose positions are closest to each other is performed, and the same identification code is given to the associated change regions and the object size detected from the change region. In addition, the entity area determination unit 44 stores tracking information in which the identification code assigned to the coordinate value of the circumscribed rectangle of the change area determined to be the entity area is stored in the object information storage unit 33, and the change area determined to be the entity area The object information storage means 33 stores the identification code assigned to the object size in association with each other.

  With reference to FIGS. 7 and 8, the determination process performed by the entity region determination unit 44 will be described in detail.

  FIG. 7 is a diagram for explaining a case where a change area in which a real area and a non-real area are integrated is detected.

  As in the case (A) shown in FIG. 7, when the change area 601 overlapping the puddle area 600 is input and the size is larger than the object size 602 to which the same identification code as the change area is given, The discriminating means 44 discriminates the partial area 604 of the object size 602 including the farthest position 603 in the change area 601 as a real area, discriminates the remaining partial area 605 as a non-substantive area, and outputs information of the real area 604. To do. In the size comparison, there is a margin in consideration of the detection error of the change area. For example, if the width and height of the change area 601 are Wi and Hi, the width and height stored as the object size 602 are Wr and Hr, respectively, and the margin is 5%, If (Wi × Hi) / (Wr × Hr)> 1.05, it is determined that the size of the change area 601 is larger than the object size 602, and (Wi × Hi) / (Wr × Hr) ≦ 1.05. If so, it is determined that the size of the change area 601 is smaller than the object size 602.

  Similarly, as in the case (B) shown in FIG. 7, when the change area 611 that overlaps the puddle area 610 is larger than the object size 612 corresponding to the area, the entity area determination unit 44 uses the change area 611. Among them, the partial area 614 having the object size 612 including the farthest position 613 is determined as a substantial area, the remaining partial area 615 is determined as a non-realistic area, and information on the substantial area 614 is output.

  Similarly, as in the case (C) shown in FIG. 7, when the change area 621 that overlaps the puddle area 620 is larger than the object size 622 corresponding to the area, the entity area determination unit 44 determines that the change area 621 Among these, the partial area 624 having the object size 622 including the farthest position 623 is determined as a substantial area, the remaining partial area 625 is determined as a non-realistic area, and information on the substantial area 624 is output.

  On the other hand, FIG. 8 is a diagram for explaining a case where a change area composed only of a non-substance area is detected. When the entity 700 exits the monitoring space and does not appear in the visible image range 701, the non-substance region 702 may be mistracked as the entity when the image is reflected in a puddle in the monitoring space.

  With respect to this case, the entity region discriminating means 44 determines that the change region 702 that overlaps the puddle region 703 is smaller than the object size 704 corresponding to the region, and that the farthest position 705 is substantially on the boundary line of the puddle region 703. If there is, the change area 702 is determined as a non-substance area. Note that “substantially on the boundary line” means that the distance from the boundary line to the farthest position is not more than a predetermined distance threshold in consideration of the detection error of the change region. The distance threshold may be a value of about several pixels.

  As described above, when each of the change areas detected by the change area detection means 43 overlaps with the puddle area, the entity area determination unit 44 determines the position in the monitoring space corresponding to the change area. The farthest position farthest from the imaging means 20 is set, and the farthest position is used to discriminate between the non-substance area by the image reflected in the puddle in the change area and the real area by the real object in the change area. .

  More specifically, when the change area that overlaps the puddle area is larger than the object size, the substance area determination unit 44 determines the object size including the farthest position set in the change area among the change areas. The partial area is determined to be an actual area. Further, when the farthest position of the change area overlapping the puddle area is on the substantial area boundary of the puddle area, the real area determination unit 44 determines the change area as a non-substance area.

  As described above, the area includes the change area detection means 43, the substance area determination means 44, the background image storage means 30, and the puddle area storage means 32, and includes the object information storage means 33 as necessary. The detection device can discriminate the substance area of the change area with high accuracy based on the water pool area and the farthest position.

[Operation of Intrusion Detection Device 1]
The operation of the intrusion detection device 1 according to the present embodiment will be described with reference to the flowcharts of FIGS.

  After the intrusion detection device 1 starts operation, the imaging unit 2 functions as the imaging unit 20 and images the monitoring space at a predetermined imaging cycle and outputs an image. The process shown in FIG. 9 is repeated in this imaging cycle.

  First, the image processing unit 4 confirms whether or not the water region detection cycle has arrived (S1). This confirmation can be performed by counting time or counting the number of image inputs.

  When the water region detection cycle arrives (YES in step S1), the image processing unit 4 executes a water region detection process (S2). Hereinafter, the water region detection processing in step S2 will be described with reference to FIG.

  In the water region detection process, first, the image processing unit 4 functions as the degree-of-change calculation unit 40, and the storage unit 3 functions as the background image storage unit 30.

  The degree-of-change calculating unit 40 acquires the absorption band image from the imaging unit 20 (S20), reads the absorption band background image from the background image storage unit 30, and the luminance value Bw of the absorption band background image and the absorption band image of each pixel. The ratio of the luminance value Iw is calculated as the absorption band change degree Vw of the pixel (S21). However, in the first detection process in which no absorption band background image is stored, the degree-of-change calculation unit 40 omits step S21.

  Next, the degree-of-change calculating unit 40 obtains a non-absorption band image from the imaging unit 20 (S22), reads out the non-absorption band background image from the background image storage unit 30, and the luminance value Bn of the non-absorption band background image in each pixel. And the luminance value In of the non-absorption band image is calculated as the non-absorption band change degree Vn of the pixel (S23). However, in the first detection process in which the non-absorption band background image is not stored, the degree-of-change calculation unit 40 omits step S23.

  Note that the order of steps S20 to S23 can be appropriately changed as long as S21 is executed after S20 and S24 is executed after S22.

  The image processing unit 4 that has finished calculating the degree of change functions as the water region detection unit 41, and the storage unit 3 functions as the water region storage unit 31.

  The water region detection means 41 extracts pixels whose ratio R between the absorption band change degree Vw and the non-absorption band change degree Vn is equal to or greater than the threshold value Tr, and detects each group of extracted pixels adjacent to each other as a water area. (S25). However, in the first detection process in which the calculation of each degree of change is omitted, the water area detecting means 41 processes for the sake of convenience that no water area has been detected.

  The water area detection means 41 that has performed the detection process updates the detection result to be stored in the water area storage means 31 for the next detection (S25). That is, the water area detection unit 41 additionally stores a binary image in which the detected pixel value in the water area is different from the pixel value outside the water area in the water area storage unit 31 as water area information. Incidentally, in the binary image when the water region is not detected, all pixel values are pixel values outside the water region.

  When no water region is detected (YES in step S26), the degree-of-change calculating unit 40 stores the absorption band image acquired in step S20 in the background image storage unit 30 as the next absorption band background image. (S27) The non-absorption band image acquired in step S22 is stored in the background image storage means 30 as the next non-absorption band background image (S28).

  After completing the above processing, the image processing unit 4 advances the processing to step S3 in FIG.

  Returning to FIG. 9, the image processing unit 4 executes a puddle area determination process (S3). Hereinafter, with reference to FIG. 11, the water pool region determination processing in step S <b> 3 will be described.

  In the puddle area determination process, the image processing unit 4 functions as a puddle determination unit 42, and the storage unit 3 functions as a water area storage unit 31 and a puddle area storage unit 32.

  The puddle determination means 42 sequentially sets the water area detected in step S25 of FIG. 10 as the water area of interest and executes the loop process of steps S30 to S34.

  In the loop processing of the water region, first, the puddle determination unit 42 reads the previous water region detected one cycle before from the water region storage unit 31, and superimposes the previous water region and the water region of interest to obtain the water region of interest. It is confirmed whether or not there is an inclusion relationship with any of the previous water areas (S31).

  If there is an inclusion relationship with any of the previous water regions, the pool determination means 42 calculates the ratio of the areas by dividing the smaller one of the previous water region and the attention water region in the inclusion relationship by the larger one. Then, by comparing with the threshold value Ts, it is confirmed whether or not there is an area change exceeding the threshold value (S32).

  When there is an area change exceeding the threshold value in the water area of interest, that is, when the area ratio is larger than the threshold value Ts (YES in step S32), the water pool determination means 42 determines that the water area of interest is a water pool area. (S33).

  On the other hand, an attention water region that is not in an inclusive relationship with any of the previous water regions (NO in step S31), or an attention water region that is in an inclusive relationship with the previous water region but has no area change exceeding the threshold (NO in step S32) ) Is determined not to be a puddle region.

  The puddle determination means 42 that has finished the processes of steps S31 to S33 for the water of interest checks whether or not all the water areas detected at the current time have been processed (S34), and the untreated water area If so (NO in step S34), the process returns to step S31 to proceed with the process on the untreated water area.

  On the other hand, when all the water areas have been processed (YES in step S34), the puddle determination means 42 stores the determination result for the next determination and for the entity determination performed in the subsequent stage. The update stored in the means 32 is performed (S35). That is, the puddle determination unit 42 overwrites and stores the binary image in which the pixel value in the water region determined as the puddle region is different from the other pixel values in the puddle region storage unit 32 as the puddle region information. Incidentally, when there is no water area determined to be a puddle area, all pixel values of the binary image are pixel values outside the puddle area.

  After completing the above processing, the image processing unit 4 returns the processing to step S1 in FIG.

  Returning to FIG. 9, when the water region detection cycle has not yet arrived (NO in S1), the image processing unit 4 first functions as the change region detecting means 43, and the storage unit 3 stores the background image. It functions as the storage means 30.

  The change area detection unit 43 acquires the visible image at the current time from the imaging unit 20 (S4), reads the visible background image from the background image storage unit 30, performs the background difference process between the visible image and the visible background image, and changes the change. A region is detected (S5). However, for the first time in the detection process, the change area detecting means 43 processes as if no change area was detected for convenience.

  If no change area is detected (NO in step S5), the visible background image in the background image storage means 30 is updated using the visible image at the current time (S11), and the process returns to step S1.

  On the other hand, when one or more change areas are detected (YES in step S5), the image processing unit 4 executes a substantial area determination process (S7). Hereinafter, with reference to FIG. 12, the real area determination process in step S7 will be described.

  In the actual area determination process, the image processing unit 4 functions as the actual area determination unit 44, and the storage unit 3 functions as the puddle area storage unit 32 and the object information storage unit 33.

  First, the entity area discriminating means 44 performs a change area tracking process for associating the change area detected in step S5 with the tracking information stored in the object information storage means 33 (S70). The same identification code as the tracking information is assigned to the change area associated with the tracking information, and a new identification code is assigned to the change area not associated with the tracking information. Further, the tracking information that is not associated with the change area is deleted from the object information storage means 33.

  Next, the entity area discriminating means 44 sequentially sets the change area as the attention change area and performs the loop processing of steps S71 to S80.

  In the loop process of the change area, the entity area determination unit 44 first reads out the information on the puddle area from the puddle area storage unit 32 and superimposes it with the change area of interest, so that the change of interest area overlaps the puddle area. It is confirmed whether or not (S72).

  When the attention change area does not overlap with the puddle area (NO in step S72), the entity area determination unit 44 determines that the attention change area is the entity area (S73), and changes the attention change area to the attention change area. The object size of the object indicated by the identification code given to is obtained and stored in the object information storage unit 33 (S74), and the attention change area is stored in the object information storage unit 33 as tracking information of the object.

  On the other hand, when the attention change area overlaps with the puddle area (YES in step S72), the entity area determination unit 44 performs the following determination that the attention change area may include a non-substance area.

  The entity area discriminating means 44 sets the uppermost end of the attention changing area at the farthest position (S75), calculates the size of the attention changing area, and is given the same identification code as the attention changing area from the object information storage means 33. The read object size is read out, and the size of the attention change area is compared with the read object size (S76). However, when the attention change area is a newly appearing object, the standard size is read instead of the object size to which the same identification code is assigned.

  When the size of the attention change area is larger than the read object size (YES in step S76), the entity area determination unit 44 assumes that the part of the object size on the farthest position side in the attention change area is the entity area. The change area is corrected to the real area (S77). Incidentally, it is determined that the portion other than the portion designated as the entity region is a non-substance region.

  On the other hand, when the size of the attention change area is equal to or smaller than the read object size (NO in step S76), the entity area determination unit 44 determines whether or not the farthest position of the attention change area is substantially on the boundary line of the puddle area. Is confirmed (S78). If substantially on the boundary line (YES in step S78), it is determined that the region of interest change is an insubstantial region (S79). As a result of the confirmation, if the farthest position is not on the boundary line (NO in step S78), the entity area determination unit 44 considers the object size obtained from the area of interest, assuming that the area of interest change is an entity area (S73). The object information storage means 33 stores the object size of the object indicated by the identification code assigned to the change area (S74).

  The entity area discriminating means 44 that has finished the processes of steps S72 to S79 for the attention change area confirms whether or not all the change areas detected at the current time have been processed (S80), and the unprocessed change area is determined. If there is (NO in step S80), the process returns to step S71, and the discrimination for the unprocessed change area is advanced.

  On the other hand, when all the changed areas have been processed (YES in step S80), substance area discriminating means 44 advances the process to step S7 in FIG.

  Returning to FIG. 9, the image processing unit 4 confirms whether or not there is a change area determined as the substantial area in step S <b> 7 (S <b> 8). YES), it is determined whether or not each corresponding change area, that is, each real area is by a person (S9). For example, the image processing unit 4 performs the determination by examining the size and shape of the substantial area, the movement based on the tracking information of the substantial area, and the like.

  If there are one or more physical regions determined to be human (YES in step S9), the image processing unit 4 outputs an alarm signal to the output unit 5 because there is an intruder in the monitoring space, and the output unit 5 Transmits the alarm signal to the server of the monitoring center (S10).

  On the other hand, if there is no change area determined to be an entity area (NO in step S8), or if there is no entity area determined to be an entity area but determined to be a person (in step S9) NO), the process of step S10 is omitted.

  After completing the above processing, the image processing unit 4 returns the processing to step S1 to process the monitoring image at the next time.

<Modification>
In the above embodiment, an example has been shown in which the degree-of-change calculating means 40 calculates the ratio between the luminance value of the background image and the luminance value of the image at the current time as the degree of change. In another embodiment, the degree-of-change calculating means 40 calculates the absolute value of the difference between the luminance value of the background image and the luminance value of the image at the current time as the degree of change. In this case, for example, the absorption band variation is | Iw−Bw |, and the non-absorption band variation is | In−Bn |. However, since it is necessary to normalize the sensitivity between the wavelength bands in either the calculation of the degree of change or the calculation of the degree of difference, when the degree-of-change calculating unit 40 calculates the degree of change based on the difference, the water region detecting unit 41 needs to calculate the ratio of the degree of change as the degree of difference of the degree of change.

DESCRIPTION OF SYMBOLS 1 ... Intrusion detection apparatus 2 ... Imaging part 3 ... Memory | storage part 4 ... Image processing part 5 ... Output part 20 ... Imaging means 30 ... Background image memory | storage means 31 ... Water area storage means 32 ... Puddle area storage means 33 ... Object information storage means 40 ... Change degree calculation means 41 ... Water area detection means 42 ... Puddle determination means 43 ... Change Area detection means 44... Real area discrimination means

Claims (5)

An imaging means having sensitivity to at least light in an absorption wavelength band absorbed by water and light in a predetermined non-absorption wavelength band different from the absorption wavelength band, and sequentially imaging a predetermined space and outputting an image;
Comparing the images captured before and after, for each pixel, the degree of change of the light intensity in the absorption wavelength band and the degree of light intensity change in the non-absorption wavelength band A change degree calculating means for calculating a belt change degree;
A water region detecting means for detecting a water region composed of pixels in which the degree of difference between the absorption band change degree and the non-absorption band change degree exceeds a predetermined reference;
An area detection apparatus comprising: Water area storage means for sequentially storing the water areas;
A combination of the water regions having an inclusion relationship is extracted by comparing the water regions detected before and after, and a water region in which the degree of area change exceeds a predetermined area change threshold is determined as a puddle region. A puddle judging means to perform,
The region detection apparatus according to claim 1, further comprising: The imaging means further outputs the image with sensitivity in the visible light or near infrared wavelength band,
further,
A change area detection means for detecting a change area in a visible light or near infrared wavelength band by comparing the images captured before and after;
A puddle area storage unit for storing the puddle area;
When the change area overlaps the water pool area, the position in the space corresponding to the change area is set to the farthest position farthest from the imaging unit, and the change area is set using the farthest position. A substantive region discriminating means for discriminating a substantive region by an image reflected in a puddle inside and a substantive region by a real object in the change region;
The region detection apparatus according to claim 2, further comprising: Object information storage means for storing an object size of an object moving in the space;
Further comprising
When the change area overlapping the puddle area is larger than the object size, the substance area determination unit determines a partial area of the object size including the farthest position from the change area as the substance area. The image processing apparatus according to claim 3, wherein: The substantial area determination means determines that the change area is the non-substance area when the farthest position of the change area overlapping the puddle area is on an approximate area boundary of the puddle area. The region detection apparatus according to claim 3 or 4, characterized in that