JP2004208209A - Device and method for monitoring moving body - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a moving body monitoring device that can surely detect a moving body to be monitored only by separating the moving body from disturbances. <P>SOLUTION: This moving body monitoring device is provided with a changing area extracting means 21 which extracts a changing area by taking the difference between input images inputted from an image inputting means 1 in prescribed cycles and a background image, a target judging means 22 which judges the presence/absence of the moving body in the changing area extracted based on prescribed judging conditions, and a background image updating means 23 which updates the background image by using the image of the changing area after erasing the image of the moving body from the image of the changing area when it is judged that the moving body exists in the extracted changing area. This monitoring device is also provided with a target information calculating means 24 which calculates the information on the moving body, such as the location, size, etc., of the moving body when the presence of the moving body is judged by means of the judging means 22 and an outputting means 3 which outputs the calculated information on the moving body. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
According to the present invention, a digital image captured by an image capturing apparatus (for example, a high-sensitivity digital camera or the like) installed outdoors or indoors is subjected to image processing at a fixed cycle, and a target such as a vehicle or an aircraft existing in an image capturing area is obtained. TECHNICAL FIELD The present invention relates to a moving object monitoring device and a moving object monitoring method for detecting a moving object and outputting moving object information such as its position and size.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, a moving object detection device in an image (also referred to as an image sensor) detects a moving target such as a vehicle or an aircraft by a background difference method from a digital image captured by an ITV camera or the like.
In this moving object detection device, for example, a currently captured input image is compared with a background image in which no moving object exists, which is calculated and created by a computer in the detection device, and the difference between the two images is changed. (The image showing the difference between the two images is called a “difference image”).
If a moving object such as a vehicle or an aircraft is included in the input image, a moving object such as a vehicle or an aircraft not included in the background image is extracted as a change area in the difference image (the extracted area is referred to as an “extraction area”). ").
[0003]
With respect to this extraction region, calculation of features such as area (size), luminance distribution, and inertia axis is performed. If the extraction region has target characteristics, the extraction region is set to a target (that is, a monitoring target such as a vehicle or an aircraft). And the information on the position and size of the moving object is output.
In addition, if the change of the texture (texture: pattern, etc.) between the input image and the background image is small, the extraction area may be the shadow of a cloud projected on the road surface by sunlight or clouds flowing in the sky. And determined that the extraction area was not a moving target of a vehicle or an aircraft.
[0004]
That is, the conventional moving object detection device calculates various feature amounts such as size, brightness, shape, and texture of the extraction region, and indicates the characteristics of the target moving object for each of the extracted regions. It is determined based on a predetermined criterion whether a feature such as disturbance is present.If the extraction region indicates a target feature amount such as a vehicle or an aircraft, it is determined that a mobile target such as a vehicle or an aircraft exists in the extraction region. I was
[0005]
Also, for example, Patent Document 1 states, "When an image changes, the change is caused by a change in lighting conditions, a slow background change, or the appearance or passage of a moving object (that is, a moving object). Is detected by a statistical method for each small region, the background image of the small region is appropriately updated only in the case of illumination fluctuation or slow background fluctuation, and the difference processing and 2 A moving object (moving object) region extraction method based on background subtraction that extracts a moving object (moving object) region by performing a binarization process ”is shown.
[0006]
As a specific method of updating the background image in the invention of Patent Document 1, in paragraphs 0043 to 0044, "the luminance value of each small area is t ₀ During the (predetermined time interval), the number of frame images is obtained. Therefore, the small area a ₁ When the luminance value is plotted on the horizontal axis and the frequency of appearance of the same luminance value is plotted on the vertical axis, a frequency distribution is obtained. a ₁ There is no moving object in the small area, and it is the background. The luminance of the background changes mainly due to illumination variations, but t ₀ In the case where is a short time of about several seconds, since the illumination change is generally small, this distribution has a bell-like shape with a small spread around the luminance peak. That is, the statistical dispersion degree σ shows a small value. On the other hand, small area a ₂ Since the moving object is photographed, t ₀ In the meantime, the luminance value changes greatly, and the distribution shows a spread. That is, the degree of dispersion σ indicates a large value. Therefore, the degree of dispersion when there is a large illumination variation is σ ₀ The value is compared with the σ, and if σ is small, it is determined that a change in the luminance value due to illumination fluctuation is made, and the numerical value of the background image small area is updated. Also, σ becomes σ ₀ If it is larger, it is determined that the luminance change of this small area is caused by factors other than illumination fluctuation such as appearance of a moving object, and the numerical value of the background image small area is not updated. Thus, the small area a _K Is updated. Is described.
[0007]
Also, in paragraph 0074 of Patent Document 1, "the maximum value of df (time differential value of luminance value in a small area) when an object appears in advance and the maximum value of df when a sudden illumination change occurs" Boundary value m ₀ , The σ (variance of the frequency of appearance as a statistical feature) and σ ₀ (Dispersion degree of the appearance frequency distribution as a statistical feature amount resulting from the illumination variation, which has been obtained in advance), and t ₀ (Maximum values m and m of df between (predetermined time intervals)) ₀ By comparing with σ, σ becomes σ ₀ If the value is smaller than σ, the numerical value of the background image small area is updated, ₀ Greater than, m is m ₀ If the value is smaller than σ, the numerical value of the background image small area is ₀ Greater than, m is m ₀ If it is larger than this, it is determined that the luminance change of the small area is caused by factors other than illumination fluctuations such as the appearance of a moving object, and the numerical value of the small area is not updated. It is also described.
[0008]
Japanese Patent Application Laid-Open No. H11-157, “Patent Document 2 discloses that“ the extraction area is divided into a plurality of appropriately sized spatial areas, the degree of change in the texture of the image is calculated for each of the divided areas, and the area having the maximum degree of change is selected. The presence of an intruder (that is, a moving body) is determined based on the image information of the selected area, whereby the intruder is reliably detected even if the ratio of the intruder in the extraction area is small. Has been described.
[0009]
[Patent Document 1]
JP-A-7-302328 (FIG. 1, FIG. 2, FIG. 9, paragraph 0043, paragraph 0044, paragraph 0074)
[Patent Document 2]
JP-A-2002-175531 (FIG. 9, front page)
[0010]
[Problems to be solved by the invention]
In the above-described conventional moving object detection device, when a disturbance region such as a relatively large cloud shadow or sunlight overlaps with a target region such as a vehicle or an aircraft, and is mixed together in the extraction region, the detection region in the extraction region If the disturbance area occupies a large area in, the change in the overall texture of the extraction area becomes small, and the extraction area is determined to be a disturbance, so that a moving target such as a vehicle or an aircraft cannot be detected (target detection The inability to do so is called "lost").
In addition, even when the extraction region can be determined as the target, the disturbance and the target are mixed together in the extraction region, so that the accuracy of information such as the position and size of the target is deteriorated. was there.
[0011]
In the moving object region extraction method based on the background difference disclosed in Patent Literature 1, when a target smaller than the small region exists, the target and the background are mixed in the small region, and the target is detected. There is a problem that it becomes impossible.
Further, in the technique disclosed in Patent Document 2, the selected image of the “region having the largest degree of change” is an image including a background image, and a small movement occurs when the background image has a disturbance such as a cloud shadow. There is a problem that a body target cannot be reliably detected.
[0012]
The present invention has been made in order to solve such a problem, and is a situation where disturbances such as cloud shadows and sunshine are mixed with moving objects to be monitored such as vehicles and aircraft. A moving object monitoring device and a moving object monitoring method capable of accurately detecting a moving object and accurately calculating and outputting information such as the position and size of a moving target by reliably detecting only a moving object. The purpose is to do.
[0013]
[Means for Solving the Problems]
A moving object monitoring apparatus according to the present invention includes: an image input means for inputting an image of a shooting target area taken by a digital camera at a predetermined cycle; and an input image input from the image input means at the predetermined cycle and a monitoring target. A change region extracting means for extracting a change region in the image by taking a difference from a background image in which no moving object exists; and a monitoring target in a change region extracted by the change region extraction device based on a predetermined determination condition. Target determining means for determining whether or not a moving object exists, and background image updating means for updating a background image at the predetermined period, wherein the moving object of the monitoring target is located in a change area extracted by the change area extracting means. If the target determination means determines that a moving object exists, the background image is updated using an image obtained by deleting the image of the moving object from the image of the change area determined to have the moving object. Background image updating means, target information calculating means for calculating mobile object information such as the position and size of the monitoring target mobile object whose existence has been determined by the target determining means, and the target information calculating means calculating And output means for outputting moving body information.
[0014]
In addition, the moving object monitoring method according to the present invention includes a step of inputting an image of an imaging target area picked up by a digital camera at a predetermined constant cycle, and the step of: Extracting a change region in the image by taking a difference from the background image that does not exist, and determining whether or not the monitoring target moving object exists in the change region extracted based on a predetermined determination condition. And updating the background image at the predetermined period. If it is determined that the moving object to be monitored is present in the extracted change area, the background image is moved from the image of the change area determined to include the moving object. Updating the background image using the image from which the body image has been deleted, and calculating the moving object information such as the position and size of the moving object of the monitoring target determined to be present And steps are those having the step of outputting the mobile body information calculated.
[0015]
BEST MODE FOR CARRYING OUT THE INVENTION
Embodiment 1 FIG.
FIG. 1 is a block diagram showing a basic configuration of the mobile monitoring apparatus according to the first embodiment.
The moving object monitoring device according to the present embodiment is applied to, for example, an airport surface detecting device (ASDE: Airport Surface Detecting Equipment), and reliably detects a moving object that is a moving target such as an aircraft that lands or takes off. This is for obtaining the position, size, and the like.
In the figure, reference numeral 1 denotes an image input unit for inputting image information (image) of an area to be photographed, which is photographed by a high-sensitivity digital camera (not shown) installed outdoors, as digital image information at a fixed period, and 2 denotes an image. From the image information (image) input from the input unit 1, a monitoring target (for example, a monitoring target such as an aircraft) existing in the shooting area is detected, and the background image is updated based on the detection result. The image processing unit 3 calculates and outputs information such as the position and size of the detected moving object, and the output 3 outputs the information of the moving object (aircraft) calculated by the image processing unit 2 to a subsequent device (not shown). Department.
[0016]
The image processing unit 2 calculates a difference between image information input from the image input unit 1 at regular intervals and a background image in a state where no moving object is present, thereby extracting a changed region in the image. Of a moving object (for example, an aircraft) that is a monitoring target based on the “size”, “brightness”, “shape”, and “change in texture (texture: pattern, etc.)” of the change area extracted by the change area extraction unit 21. Target determining means 22 for determining the presence or absence of the object, background image updating means 23 for updating the background image based on the detection result of the moving object (aircraft), and information such as the position and size of the detected moving object (aircraft) Is calculated from target information calculating means 24 for calculating.
[0017]
FIG. 2 is a flowchart illustrating a processing procedure in the image processing unit of the mobile monitoring device according to the first embodiment.
FIG. 3 is a diagram for explaining a flow of detecting a moving object (aircraft) in a situation where a moving object (aircraft) as a monitoring target and a cloud shadow area as a disturbance exist in an input image.
In FIG. 3, reference numeral 4 denotes a target area of a moving object (aircraft) as a monitoring target, 5 denotes a cloud shadow area as a disturbance, and 6 denotes an aircraft cylinder target area 4 and a cloud shadow area 5 mixed. This is the area when there is.
The input image is composed of, for example, 360 × 240 pixels.
[0018]
The operation (contents of processing) of the mobile monitoring apparatus according to the present embodiment will be described with reference to FIGS.
The processing shown in the flowchart of FIG. 2 is performed at predetermined time intervals (for example, every one second).
First, the operation (processing procedure) of the mobile monitoring apparatus according to the present embodiment will be described using the flowchart of FIG.
In step S1, a digital image obtained by photographing a photographing target area with a digital camera from the image input unit 1 is imported as an input image.
In step S2, the changed area extracting unit 21 obtains a difference between the background image, which is an image when no moving object is present, and the input image captured in step S1, and obtains the changed area as an extracted area.
Note that, as described above, the background image is an image created from a past image in a state where no moving object such as an aircraft exists in the shooting target area.
[0019]
Next, in step S3, the target determination unit 22 calculates the size of the extracted region and the variance of the brightness.
Further, the degree of change in the texture of the extraction area is calculated by comparing the input image with the background image.
As the degree of change, for example, a correlation value based on a normalized correlation is used.
[0020]
In step S4, it is determined whether or not the extraction region has the characteristics of the moving object (aircraft) with respect to the size of the extraction region, the variance of the brightness of the extraction region, and the degree of change in the texture of the extraction region C obtained in step 3. I do.
For the determination, the following conditions (1) to (3) are applied to the size of the extraction region, the variance of the brightness, and the degree of change of the texture.
(1) Target minimum allowable size ≤ size of extraction area ≤ target maximum allowable size
(2) Minimum value of variance of target brightness ≦ variance of brightness of extraction area
Here, the target minimum allowable size and the target maximum allowable size are given by calculating values at respective positions in the image based on parameter values.
A predetermined value is given as a parameter to the minimum value of the variance of the target brightness.
[0021]
In the present embodiment, a vertical length and a horizontal length when a monitoring target (that is, a monitoring target moving body) is approximated by a rectangle in an image are used as indices of the target size.
For the perspective, a moving object (aircraft) whose size is different depending on the position of the image (smaller as the image is higher, larger as the image is lower, and substantially proportional to the y-coordinate of the image) is targeted. The standard size (length x width) of the moving object (aircraft) is set at the two points, and the standard size of the moving object (aircraft) between the two points is determined by linear interpolation from the values of these two points. Ask.
The settable values (the image y coordinate of the setting position, the vertical length, and the horizontal length) are called parameters.
[0022]
(3) Minimum degree of target texture change ≤ degree of texture change in extraction area
Here, the condition (3) is performed when the extraction area has a predetermined area (for example, 10 pixels) or more. In order to obtain a correlation with high reliability, the number of samples needs to be about 10 or more.
By making a determination about a predetermined area or more, it is possible to avoid erroneously determining a target as a disturbance such as a change in illumination due to a low degree of correlation with a target having a small area.
Here, the “area” refers to the number of pixels (number of pixels, number of dots) occupied by an object in an image.
If the extraction area does not have a predetermined area, the condition (3) is assumed to be satisfied unconditionally.
In addition, a predetermined value of the minimum degree of texture change is given by a predetermined value as a parameter.
Extraction regions that do not satisfy the above conditions can be regarded as disturbance regions such as cloud shadows and sunshine in the image.
[0023]
If both of the above conditions (1) and (2) are satisfied, the process proceeds to step S5 to make the determination of (3).
On the other hand, if any of the above conditions is not satisfied, the extraction region is regarded as a region where no target exists.
Then, the process proceeds to step S6, where the background image is updated using the image of the area where the change area is not extracted and the image of the area where the target moving object does not exist.
In step S5, when the degree of texture change of the extraction region obtained in step S3 satisfies the determination condition (3), the extraction region is regarded as the target region.
Updating of the background image in the target area is performed in step S7. In step S7, when the target determination unit 22 determines that the monitoring target exists in the change area extracted by the change area extraction unit 21, A background image is obtained by using an image in which no image of the moving object is present from the image of the changing region determined to be present (i.e., an image obtained by deleting the image of the moving object from the image in the changing region determined that the moving object is present). To update.
On the other hand, when the determination condition of (3) is not satisfied, the extraction region is regarded as a disturbance region 9 such as a cloud shadow or light, and the process proceeds to step S8 to update the background image.
[0024]
In step S6, the background image is updated in an area where no target exists.
For the update, the following Expression 1 is applied to each pixel in a region where no target exists in the background image.
Updated pixel value = current pixel value * weight 1 + input pixel value * weight 2 ... Equation 1
Here, the input pixel value refers to the value of the pixel of the input image corresponding to the pixel of interest in the background image.
The weight 1 and the weight 2 give predetermined values such that the sum of the two values becomes 1. For example, 0.9 is used for weight 1 and 0.1 is used for weight 2.
In addition, as the input pixel value of Expression 1, a pixel value having the largest number of occurrences in a predetermined area of the input image (hereinafter, referred to as a “mode value”) may be applied.
[0025]
In step S7, the background image is updated in the target area.
For updating, Equation 1 is applied to each pixel in the extraction region in the background image in the same manner as in step S6.
Here, the weights 1 and 2 are given predetermined values so that the weight 1 is sufficiently larger than the weight 2 and the weight 1 is larger than the weight 1 in step S6. For example, 0.99 is used for weight 1 and 0.01 is used for weight 2.
[0026]
In step S8, the background image is updated for the shadow or light area.
To this end, as in step S5, equation 1 is applied to each pixel in the extraction region of the background image that has been regarded as a disturbance region in step S7.
Here, the weight 2 applied to the input pixel value is sufficiently large, and a predetermined value in which the sum of the weights 1 and 2 is 1 is given.
For example, 0.99 is used for weight 1 and 0.01 is used for weight 2.
Hereinafter, the method of updating the background image when the weight of the input pixel value is sufficiently large is referred to as “immediate update”.
Further, this immediate update may be applied not only to the extraction region but also to a region near the extraction region.
For example, the immediate update is applied to the outer peripheral area corresponding to the outer peripheral two pixels of the extracted area in addition to the extracted area.
[0027]
In updating the background image in step S8, Expression 1 above may simply substitute the input pixel value for the updated pixel value.
Lastly, in step S9, information about the position and size of the moving object, which is the target calculated by the target information calculating means 24, is transmitted to the output unit 3, and a device (not shown) connected to the subsequent stage performs a target tracking process. A monitoring operation is performed.
[0028]
The processing in each step shown in the flowchart of FIG. 2 described above will be described using specific numerical values with reference to FIG.
FIG. 3 shows a flow of an image processing situation when the processing of FIG. 2 is performed from time t seconds to time t + 1 seconds.
Here, the result of extracting the changed area between the input image A and the background image B captured in step S1 at time t seconds in step S2 is shown in an extracted area C.
In this example, in step S3, the degree of change of the texture is obtained by the normalized correlation, and the disturbance area 5 due to the shadow of the cloud projected on the road surface has a correlation value of 0.9 close to 1.
On the other hand, the target area 4 of the moving body (aircraft) takes a relatively small value 0.3 because of a change in texture.
[0029]
In the mobile monitoring apparatus according to the present embodiment, as described above, the target determination unit 22 obtains the normalized correlation between the background image and the input image in the change area, and changes the texture based on the magnitude of the correlation value. Seeking degree.
In this case, in a region where the luminance fluctuates, such as a cloud shadow region, the degree of change in the texture is small regardless of the magnitude of the luminance change, so that the correlation value is high.
On the other hand, in the area of a moving object such as an aircraft, the degree of change in the texture is large regardless of the magnitude of the luminance change, and therefore, the correlation value is low.
For this reason, in the moving object monitoring apparatus according to the present embodiment, even when the luminance change is abrupt, if the degree of change of the texture is small, it can be regarded as a disturbance, and the extracted change area is the movement of the monitoring target. It is possible to reliably determine whether the object is a body or a disturbance such as a cloud shadow.
[0030]
Assuming that both the disturbance area 5 such as a cloud shadow and the target area 4 of the moving object (aircraft) have target features in the dispersion of the size and brightness, the condition is satisfied in step S4. Then, the process proceeds to step S5.
In step S5, an aircraft target having a large degree of change in texture is regarded as a target because it satisfies the condition, and the process proceeds to step S7. Update the image.
[0031]
In step S7, since the weight 1 of the equation 1 is sufficiently larger than the weight 2, the background of the target area is hardly updated.
On the other hand, in step S8, since the background image is replaced with the input image, the disturbance area is directly reflected on the background image.
Subsequently, in the target area 4, information such as the position and size is calculated by the target information calculating means 24 in step S9, and the calculated information is transmitted to the output unit.
Here, the updated background image E is used by the change region extracting means in the next cycle of t + 1 seconds.
[0032]
In this way, disturbance areas such as cloud shadows and sun light in the image were found, and these disturbance areas were treated as the background in the near future, so in the near future, disturbance areas such as cloud shadows and sun light Even if the target and the target (moving body) regions coexist, the disturbance region is not extracted as a changed region.
That is, as shown in FIG. 3, at the time t + 1 seconds, the input image D (the image in which the disturbance region and the target region are mixed) and the updated background image E (the background image in which only the disturbance region exists) are moved from the target (the moving object). ) Is extracted, and a target (moving body) is extracted from the extracted region F.
Therefore, in the cycle of t + 1 seconds, as a result of step S2, only the target (aircraft) is separated and extracted from the disturbance region such as cloud shadows and sunlight even in a situation where the disturbance region and the target coexist. .
For this reason, in step S9, it is possible to calculate information such as the position and size of the target (moving body) to be monitored with high accuracy and output the information to the output unit.
[0033]
As described above, the moving object monitoring apparatus according to the present embodiment extracts an integrated change region from a difference image between a background image and an input image, and changes the change region in size, brightness, shape, and texture. , The presence of a moving target is determined, and information such as the position and size of the detected target is output.
Further, the background image is updated based on the target detection result.
Further, the background image updating means replaces the background image with the input image only for the area determined to be the disturbance area based on the determination result by the target determining means.
This makes it possible to extract a moving target with little disturbance and no movement.
Further, in the moving object monitoring apparatus according to the present embodiment, since the background image is replaced with the input image in the region regarded as a disturbance (that is, it is immediately updated), the disturbance region is directly reflected on the background region.
For this reason, if the cloud shadow continues to exist, the disturbance area is not extracted as a changed area in the next cycle.
[0034]
Further, even when the disturbance area is still present in the input image, the disturbance area already existing in the background image is extracted as a changed area by replacing the background image in the disturbance area with the input image. Therefore, it is possible to prevent a target area (moving body) newly appearing in the extraction area in the near future and an existing disturbance area from being mixed.
[0035]
In the first embodiment, a case where the present invention is applied to an airport surface detecting device has been described as an example. However, a road monitoring device or another monitoring device may be used.
[0036]
【The invention's effect】
A moving object monitoring apparatus according to the present invention includes: an image input means for inputting an image of a shooting target area picked up by a digital camera at a predetermined cycle; an input image input from the image input means at a predetermined cycle; A change area extracting means for extracting a change area in the image by taking a difference from a non-existent background image, and a moving object of a monitoring target is set in a change area extracted by the change area extraction means based on a predetermined determination condition. A target determining unit for determining whether or not the moving object is present; and a background image updating unit for updating the background image at a predetermined cycle, wherein the target determining unit determines that the moving object of the monitoring target exists in the change area extracted by the change area extracting means. When the means determines, the background image updating means updates a background image using an image obtained by erasing the image of the moving object from the image of the change area determined to have the moving object. A target information calculating means for calculating moving body information such as a position and a size of the monitoring target moving body whose presence is determined by the target determining means, and an output for outputting moving body information calculated by the target information calculating means. With this arrangement, it is possible to reliably find a disturbance area such as a cloud shadow in an image, and treat this area as a background in subsequent processing.
Therefore, in the near future after updating the background image that also includes the disturbance area as the background, even if the disturbance area and the monitoring target (mobile body) are mixed, only the moving body that is the monitoring target is included in the disturbance area. As a result, it is possible to provide a moving object monitoring device capable of outputting information such as the position and size of the moving object with high accuracy.
[0037]
Also, the moving object monitoring method according to the present invention includes a step of inputting an image of a shooting target area picked up by a digital camera at a predetermined cycle, and an input image input at a predetermined cycle and a background image having no monitoring target moving object. Extracting a change area in the image by taking a difference from the predetermined area, determining whether or not a monitoring target moving object exists in the change area extracted based on a predetermined determination condition; In the step of updating the background image, when it is determined that the monitoring target of the moving body is present in the extracted change area, the image of the moving body is determined from the image of the change area determined to have the moving body. Updating the background image using the erased image; and calculating moving object information such as the position and size of the moving object of the monitoring target whose existence has been determined. Since a step of outputting the mobile body information calculated, locate the disturbance regions, such as the shadow of a cloud in the image reliably, can be treated as background in the subsequent processing of this region.
Therefore, in the near future after updating the background image that also includes the disturbance area as the background, even if the disturbance area and the monitoring target (mobile body) are mixed, only the moving body that is the monitoring target is included in the disturbance area. As a result, it is possible to provide a moving object monitoring method capable of outputting information such as the position and size of the moving object with high accuracy.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a basic configuration of a mobile monitoring apparatus according to a first embodiment.
FIG. 2 is a flowchart for explaining a processing procedure in an image processing unit of the mobile monitoring device according to the first embodiment;
FIG. 3 is a diagram for explaining a flow of detecting a moving object that is a monitoring target when disturbance (such as a cloud shadow) exists in the moving object monitoring device according to the first embodiment;
[Explanation of symbols]
1: Image input unit 2: Image processing unit input unit
3: Output unit 4: Target area
5: disturbance area
6: Area where target area and disturbance area overlap
21: Change area extraction means
22: target determination means
23: background image updating means
24: target information calculation means

Claims (9)

Image input means for inputting an image of a shooting target area shot by a digital camera at a predetermined cycle,
A change region extraction unit that extracts a change region in the image by taking a difference between the input image input in the predetermined period from the image input unit and a background image in which the moving object of the monitoring target does not exist;
Based on a predetermined determination condition, target determination means for determining whether or not a monitoring target moving body is present in the change area extracted by the change area extraction means;
A background image updating unit that updates a background image at the predetermined cycle, wherein when the target determination unit determines that a monitoring target mobile object is present in a change area extracted by the change area extraction means, Background image updating means for updating a background image using an image obtained by deleting an image of a moving object from an image of a change area determined to be present,
Target information calculating means for calculating moving body information such as the position and size of the monitoring target moving body whose presence has been determined by the target determining means,
Output means for outputting mobile object information calculated by the target information calculation means. The mobile object monitoring device according to claim 1, wherein the target determination unit performs the determination when the change area is equal to or more than a predetermined number of pixels. 3. The moving object monitoring device according to claim 1, wherein the target determination unit performs the determination using a correlation value based on a normalized correlation between the background image and the input image in the change area. The mobile object monitoring device according to any one of claims 1 to 3, wherein the background image updating means updates the weighted average of the background image and the input image. The moving object monitoring apparatus according to any one of claims 1 to 4, wherein the background image updating means immediately updates the background image with the input image in a change area where it is determined that the moving object does not exist. apparatus. The mobile object monitoring device according to claim 5, wherein the background image is immediately updated with the input image even in a region near the periphery of the extracted change region. Inputting an image of a shooting target area shot by a digital camera at a predetermined cycle;
Extracting a change area in the image by taking a difference between the input image input in the predetermined cycle and a background image in which the moving object of the monitoring target does not exist;
Based on a predetermined determination condition, a step of determining whether there is a monitoring target moving body in the extracted change area,
In the step of updating the background image at the predetermined cycle, when it is determined that the monitoring target moving body is present in the extracted change area, the moving body is determined from the image of the change area determined to have the moving body. Updating the background image using the image from which the image has been deleted; and
A step of calculating moving body information such as the position and size of the moving body of the monitoring target whose existence has been determined;
Outputting the calculated moving object information. 8. The moving object monitoring method according to claim 7, wherein the background image is updated by a weighted average of the background image and the input image. 8. The moving object monitoring method according to claim 7, wherein the background image is immediately updated with the input image in the changing region in which the degree of texture change in the changing region is small and the moving object does not exist.

Images