JP2013117772A - Image monitoring device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately distinguish between a person and a small animal existing close to a device.SOLUTION: An image monitoring device 1 includes an image processing unit 4 including: feature quantity calculation means 4c which obtains a period (intra-region existence period) for which a lower end of a change region to be tracked by tracking means 4b and a lower end of a monitored image are close to each other so as to be regarded as almost contacting each other, within an appearance period from the appearance time of the change region to the determination time, obtains a temporal proportion of the intra-region existence period to an evaluation period and calculates such a feature quantity that the higher temporal proportion indicates that the change region is more likely to represent a person; and determination means 4d which determines whether the change region represents a person's image or not on the basis of the feature quantity.

Description

  The present invention relates to an image monitoring apparatus that sequentially acquires monitoring images obtained by imaging a monitoring space, and performs image processing on the sequentially acquired monitoring images to determine the presence or absence of a person in the monitoring space.

  Conventionally, a region indicating the characteristics of a moving object such as an intruder is detected from a change between an image input from an imaging device such as a camera and a past image, and the region is a target moving object (for example, an intruder). There is known an image monitoring apparatus that determines whether or not.

  As this type of image monitoring device, for example, an input image currently captured by the imaging device is compared with a past image stored in advance, and a difference between both images is extracted as a change region. A device that calculates an area, a moving speed, and the like and determines whether a target moving object exists in the monitoring space based on the calculation result is known. In Patent Document 1 below, as an image sensor used in this type of image monitoring device, a fluctuation region is extracted and extracted from a difference between a captured image of a monitoring space sequentially acquired by an imaging unit and a past image. What tracks the movement of an area is disclosed.

JP 2006-318064 A

  By the way, when detecting an intruder invading into the monitoring area using an image of the monitoring area, it is determined whether or not there is an intruder using the characteristics of the change area generated in the image. It is carried out. At this time, since changes occur in the image due to various factors other than the intruder, it is necessary to distinguish between these changes and changes due to the intruder. In particular, in addition to an intruder, an example of a factor that causes a change in an image is a small animal that enters a surveillance area. Therefore, in the conventional image monitoring apparatus, (1) the difference in the number of pixels in the change area in the image (intruder: many, small animals: few), (2) perspective conversion from the width of the circumscribed rectangle of the change area in the image (3) Differences in the real space estimated by perspective transformation from the height of the circumscribed rectangle of the change area in the image The difference in estimated height (intruder: high, small animal: low) is used to distinguish intruders from small animals.

  However, when an intruder is located in the immediate vicinity of an imaging device such as a camera, the whole body image of the intruder may not be reflected in the image. Specifically, as shown in FIG. 7A, this corresponds to the case where the lower body of the intruder is out of frame and only the upper body is shown. As described above, when the whole body image of the intruder is not displayed on the screen, the size of the portion extracted as the change area looks small on the image, and as shown in FIG. It is similar to the state where is shown. Thereby, the estimated height in the real space of the intruder and the small animal in the immediate vicinity of the imaging device is regarded as equivalent (represented by the arrows in FIGS. 7A and 7B), and in the above (1) to (3) It became difficult to distinguish an intruder from a small animal, and there was a risk of misjudging it as a small animal rather than an intruder.

  Therefore, the present invention has been made in view of the above problems, and the time ratio (frequency) that is substantially in contact with the lower end of the image of the change area is smaller than when the change area is a person. Focusing on the fact that there is a small feature, it is possible to accurately distinguish between a person and a small animal by effectively using the temporal ratio of the change area existing within the predetermined area from the bottom edge of the screen including the touch at the bottom edge of the screen An object of the present invention is to provide an image monitoring apparatus that can perform the above-described operation.

In order to achieve the above object, an image monitoring apparatus according to claim 1 of the present application sequentially acquires a monitoring image obtained by imaging a monitoring space, and performs image processing on the monitoring image to determine the presence or absence of a person in the monitoring space. A monitoring device,
An extraction unit that compares the monitoring image with a reference image of a monitoring space stored in advance and extracts a changed region from the reference image;
A tracking unit that temporally tracks the change region;
An in-region existence period is obtained in which a lower end of the change region and a lower end of the monitoring image can be regarded as being substantially in contact in a predetermined evaluation period from the appearance time of the change region to a determination time for determining the presence or absence of a person. A feature amount calculation unit that calculates a feature amount that the change region is likely to be a human image as the ratio of the in-region existence period to the evaluation period increases;
A determination unit that determines whether or not the change region is a person image using the feature amount;
It is characterized by having.

With this configuration, the feature amount calculation unit of the present invention has the lower end of the change area in the predetermined evaluation period (for example, the appearance period from the change area appearance time to the determination time) for the change area tracked by the tracking section. And the lower end of the monitoring image are so close that they can be considered to be in close contact (hereinafter referred to as “in-area existence period”). Then, the feature amount calculation unit of the present invention obtains a time ratio that the in-area existence period occupies in the evaluation period (that is, a ratio between the accumulated time in the in-area existence period and the accumulated time in the evaluation period), and the ratio is large. The feature amount that the change area is likely to be a person is calculated. And the determination part of this invention determines whether a change area | region is a person image based on the said feature-value.
As a result, when the moving object in the immediate vicinity of the apparatus is a person, the change area is within a predetermined area from the lower end of the image including contact with the lower end of the image (the lower end of the change area and the lower end of the monitoring image are substantially in contact with each other). It can be determined that the person seems to be a person because the time ratio (frequency) existing within the range that can be considered is large. On the other hand, when the moving object that is in the immediate vicinity of the device is a small animal, since the temporal ratio (frequency) that the change area exists in the predetermined area from the lower end of the image including contact with the lower end of the image is relatively small, Can be judged. In this way, by using an event that the change ratio of the small animal is smaller in time ratio that is substantially in contact with the lower end of the monitoring image than the change area of the person, the moving object that is present in the immediate vicinity of the image monitoring apparatus is Thus, it is possible to accurately determine whether or not the person is an intruder.

The image monitoring apparatus according to claim 2 of the present application is the image monitoring apparatus of claim 1,
When calculating the feature amount, correction is performed such that the time length of the in-region existence period is reduced as the distance between the lower end of the change region and the lower end of the monitoring image increases.

With this configuration, when the feature amount calculation unit of the present invention calculates the feature amount after obtaining the in-region existence period of the change region tracked by the tracking unit, the distance between the lower end of the change region and the lower end of the monitoring image is Even if there is a change region in the distance that can be regarded as being substantially in contact, the time length of the in-region existence period is corrected to decrease as the distance increases, and the time ratio between the corrected in-region existence period and the evaluation period In other words, the feature amount is calculated in accordance with (that is, the ratio between the corrected accumulated time in the region existing period and the accumulated time in the evaluation period). And the determination part of this invention determines whether a change area | region is a person image based on the said feature-value.
For example, when the lower end of the change area is in contact with the lower end of the monitoring image (that is, when the change area includes all or part of the lower end area of the monitoring image), the change area is the most likely person. The feature amount is calculated so as to take a large value. Moreover, even if the lower end of the change area is not in contact with the lower end of the monitoring image, if the distance is relatively small, an error caused by the problem of extraction accuracy (for example, a person's clothes are very similar to the background) Since there is a strong suspicion of an error (because it is difficult to extract a location as a change area), a relatively large feature amount is calculated. On the other hand, even if there is a change area within the range of distance that can be considered to be almost in contact with the lower end of the monitoring image, the larger the distance, the less likely the error is due to the problem of extraction accuracy, and the possibility of small animals Therefore, a small feature amount is calculated as not being a person.
In this way, by calculating the human-like feature quantity in consideration of the distance between the lower end of the change area and the lower end of the monitoring image, the person and the small animal are accurately distinguished and determined regardless of the extraction accuracy of the change area. can do. And the misreporting by considering a person as a small animal can be reduced, and the misinformation by considering a small animal as a person can be reduced.

The image monitoring apparatus according to claim 3 of the present application is the image monitoring apparatus according to claim 1 or 2,
The feature amount calculation unit calculates only one or a plurality of periods during which the change region has moved by a predetermined amount or more as the predetermined evaluation period.

With such a configuration, the feature amount calculation unit of the present invention obtains the movement amount of the change area tracked by the tracking unit, sets only one or a plurality of periods in which the movement amount is equal to or greater than a predetermined value as a predetermined evaluation period, A feature amount corresponding to the ratio of the in-region existence period in the evaluation period to the evaluation period is calculated. And the determination part of this invention determines whether a change area | region is a person image based on the said feature-value.
If the small animal stays near the lower end of the monitoring image, the feature amount is calculated to be large, and as a result, the small animal may be erroneously determined to be a person. For this reason, if the change area is a small animal, a feature is used in which the change area is more likely to move away from the lower end of the monitoring image when compared with the case where the change area is a person. Thus, it is possible to distinguish a person and a small animal with high accuracy. In other words, by setting the period during which the moving amount of the change region is equal to or greater than the predetermined value as the evaluation period, the feature amount can be calculated by excluding the dwelling period of the small animal. A person and a small animal can be distinguished with high accuracy without determination.

  As described above, when the moving object present in the immediate vicinity of the apparatus is a person, the image monitoring apparatus of the present invention can be considered that the change area is substantially in contact with the lower end of the image (that is, including the contact of the change area). The time ratio (frequency) that exists in the predetermined area from the lower end of the monitoring image is large. On the other hand, if the moving object that is closest to the device is a small animal, the change area is substantially in contact with the lower end of the image. By actively utilizing the feature that the time ratio (frequency) that can be considered is relatively small, it is possible to accurately distinguish and determine whether or not the moving object present in the immediate vicinity of the image monitoring apparatus is an intruder. it can.

1 is a block configuration diagram showing an overall configuration of an image monitoring apparatus according to the present invention. (A) It is a figure which shows a mode when the tracking label area | region by the intruder who exists nearest to an apparatus moves. (B) It is a figure which shows a mode when the tracking label area | region by the small animal which exists in the immediate vicinity of an apparatus moves. It is explanatory drawing of each period defined in the image monitoring apparatus which concerns on this invention. It is a flowchart which shows the process sequence of the whole operation | movement by the image monitoring apparatus which concerns on this invention. It is a flowchart which shows the procedure of the specific process content of the feature-value calculation process of FIG. It is a figure which shows an example of the evaluation function at the time of calculating the number of frames existing in an area | region in the image monitoring apparatus which concerns on this invention. (A), (b) It is explanatory drawing which shows an example of the misreport by the conventional image monitoring apparatus.

  Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the accompanying drawings. It should be noted that the present invention is not limited to this embodiment, and all other forms, examples, operation techniques, etc. that can be implemented by those skilled in the art based on this embodiment are included in the scope of the present invention. .

(About the overall configuration of the image monitoring device)
The image monitoring apparatus according to the present invention sequentially acquires monitoring images obtained by imaging a monitoring space, and performs image processing on the sequentially acquired monitoring images to determine the presence or absence of a person in the monitoring space.

  In particular, according to the present invention, as shown in FIGS. 2A and 2B, the frequency at which the human body exists outside the screen at the lower end of the monitoring image and exists within a predetermined area from the lower end of the monitoring image. Focusing on the fact that the whole body is almost visible in the case of small animals, while the frequency (ratio) is small in the predetermined area from the lower end of the monitoring image, and the frequency (ratio) depends on the frequency (ratio). The calculated feature amount is calculated. That is, in a predetermined period (hereinafter referred to as “evaluation period”) from the appearance time of the change area extracted from the monitoring image to the determination time for determining whether the person is an intruder, the change area is the lower end of the monitoring screen. The ratio (hereinafter referred to as “intra-area occupation ratio”) of the period (the intra-area existence period) that can be regarded as being substantially in contact with the image is calculated as a feature amount. In this embodiment, the distance between the lower end of the change area and the lower end of the monitoring image is obtained, and when the distance is within a predetermined range (within 10 pixels) (including contact with the lower end of the monitoring image), “ The period in which the change area exists within the range is determined as the in-area existence period. Then, an image monitoring apparatus including a function for determining whether or not the change area is a person image based on the calculated occupation ratio in the area is provided.

  As shown in FIG. 1, the image monitoring apparatus 1 of this example is schematically configured to include an imaging unit 2, a storage unit 3, an image processing unit 4, and an output unit 5 in order to realize the above-described functions.

  The imaging unit 2 is configured by a two-dimensional image sensor having an optical system such as a lens and a two-dimensional array element such as a CCD or CMOS, for example, and an image processing unit 4 that captures a captured image (hereinafter referred to as an image) obtained by imaging a monitoring space. Is output. The imaging unit 2 is installed on the ceiling or wall and images a space obliquely below from above. More specifically, the imaging unit 2 using the two-dimensional image sensor forms an image of the light emitted from the monitoring space on the light receiving plane of the two-dimensional array element by an optical system such as a lens, and uses the light and darkness of the image as the amount of charge. It performs photoelectric conversion, sequentially reads out and converts it into an electrical signal, and outputs it to the image processing unit 4 as an image in which each pixel is expressed by a luminance value. At that time, an image of one frame is acquired at a predetermined time interval and output to the image processing unit 4.

  In addition, the imaging unit 2 includes, as imaging parameters, external parameters indicating the installation position and imaging direction, and internal parameters indicating focal length, angle of view, lens distortion, other lens characteristics, and the number of pixels of the imaging element. This imaging parameter can be obtained by actually performing measurement or the like, and is stored in the storage unit 3 in advance.

  Then, using this imaging parameter, the pixel position in the monitoring image can be coordinate-converted between coordinates (imaging plane coordinates) on the imaging plane of the imaging unit 2 and coordinates (real coordinates) in real space. . In this embodiment, both (1) conversion from imaging surface coordinates to real coordinates and (2) conversion from real coordinates to imaging surface coordinates using the imaging parameters are collectively referred to as perspective transformation.

  The imaging unit 2 includes an infrared detection element having predetermined pixels (for example, 320 × 320 pixels), detects infrared rays emitted from the monitoring region, and temperature distribution in the monitoring region according to the detected amount of infrared rays. Can be configured by a thermal image sensor that outputs the image as a monitoring image to the image processing unit 4.

  The storage unit 3 stores setting information related to the image monitoring apparatus 1 (for example, the installation height of the apparatus and the depression angle) and information used for various processes of the image processing unit 4. The setting information related to the image monitoring device 1 includes imaging parameters including the installation environment of the imaging unit 2 described above. The information used for various processes of the image processing unit 4 includes, for example, image data for each frame, a reference image for extracting a change area from the image data, a calculation formula used for the processing of this embodiment, Various parameters such as threshold values are included.

  The image processing unit 4 is configured by a computer having a CPU and the like, receives an input of a digitized image from the imaging unit 2, and includes a series of processes shown in FIGS. 4 and 5 described later as an extraction process, a labeling process, In order to execute label integration processing, noise removal processing, tracking (tracking of object candidate) processing, feature amount calculation processing, and intruder determination processing, an extraction means 4a, tracking means 4b, feature amount calculation means 4c, and determination means 4d are included. It is out.

  The extraction unit 4a performs an extraction process for extracting a changed area from the image acquired by the imaging unit 2 as a changed area. Further, the extracting unit 4a labels the extracted change area by labeling processing. Further, the extracting unit 4a performs a label integration process for integrating a plurality of labels in a predetermined range into one label. Further, the extraction unit 4a determines whether or not noise extraction is performed for each label, and performs noise removal processing to be removed when it is determined that noise extraction is performed.

  The tracking unit 4b temporally tracks the change region extracted by the extraction unit 4a, and the label region (change region) obtained by the labeling process by the extraction unit 4a is compared with the tracking label region of the previously acquired image. Corresponding.

  The tracking label area is a unique label (hereinafter referred to as tracking) that is different from the label in the labeling process by the extraction means 4a so that the label area by the same tracking object is always the same label in the images acquired so far. It refers to the label area that has been given a label again.

  As shown in FIG. 5, the feature amount calculation unit 4 c is configured to calculate the circumscribed rectangular coordinates of the tracking label area, the perspective conversion process, the movement amount calculation process of the tracking label area, the in-area occupation ratio calculation process, and the humanity closest to the apparatus. Processing, feature amount calculation processing including humanity calculation processing other than the closest to the device, and humanity calculation processing is performed, and a feature amount for determining an intruder is calculated for the tracking label area obtained by the tracking unit 4b. Yes.

  The feature amount calculated by the feature amount calculation process includes the size in the real space (estimated width / estimated height) estimated by the perspective transformation from the size of the changed region on the image, the moving amount, and the above-described region. In-occupation ratio etc. are included. Details of the feature amount calculation processing will be described later.

  The determination unit 4d executes an intruder determination process, which will be described later, and the target tracking label region (change region) enters based on the feature amount calculated by the feature amount calculation unit 4c in accordance with a predetermined determination condition. It is judged whether it is a person.

  The output unit 5 outputs a determination signal indicating that an intruder has been determined by the determination unit 4d of the image processing unit 4 to the outside. The output unit 5 includes, for example, a display or a buzzer. When a determination signal indicating that there is an intruder is input from the determination unit 4d of the image processing unit 4, the output unit 5 drives a display or a buzzer to notify that there is an intruder.

  The output unit 5 is connected to a security device (not shown) or a remote monitoring center via a communication line, and outputs an intruder determination signal input from the determination unit 4d of the image processing unit 4 to the communication line. It can also be configured as a communication I / F.

(Regarding the presence / absence determination process by the image monitoring apparatus 1)
Next, the processing operation of the image processing unit 4 in the case where the presence or absence of a person in the monitoring space is determined using the image monitoring apparatus 1 having the above configuration will be described with reference to FIG.

  When the image monitoring apparatus 1 is powered on, the image processing unit 4 executes setting information acquisition processing (ST1). In the setting information acquisition process, setting information related to the image monitoring apparatus 1 set in advance and information used for various processes of the image processing unit 4 are acquired, and the acquired information is stored in the storage unit 3. The information referred to here includes the installation height and depression angle of the imaging unit 2, the number of pixels in the vertical and horizontal directions of the image obtained from the imaging unit 2, the vertical angle of view and horizontal angle of view, the reference image, various threshold values, and the like.

  Next, the image processing unit 4 executes an input image acquisition process (ST2). In the input image acquisition process, the imaging unit 2 acquires an image obtained by imaging the monitoring range of the monitoring space.

  Next, extraction processing is executed by the extraction means 4a of the image processing unit 4 (ST3). In the extraction process, the change area is extracted from the image acquired in the input image acquisition process. In this embodiment, the change area is extracted by taking the difference between the monitoring image acquired this time and the reference image of the monitoring space stored in advance. At this time, a background image of the monitoring space or a monitoring image acquired in the past can be appropriately selected and used as the reference image.

  In addition, when acquiring a thermal image as an image from the imaging part 2, the area | region where temperature change was extracted from the acquired thermal image as a change area | region. For example, a past thermal image is stored in advance in the storage unit 3 as a reference image, a region where the difference between the current thermal image and the reference image is equal to or greater than a threshold value is determined to have a temperature change, and the region having this temperature change is determined. Extract as a change area.

  Next, a labeling process is executed by the extraction means 4a of the image processing unit 4 (ST4). In the labeling process, the changed area obtained in the extraction process of ST3 is labeled. For example, it is possible to use a technique in which extracted pixels adjacent around the pixel of interest are grouped into a label area. In the labeling process, a label integration process for integrating a plurality of labels in a predetermined range into one label is performed. For example, for each label extraction area, the installation information obtained by the setting information acquisition process (for example, information such as the installation height and depression angle of the imaging unit 2) is used to determine the position of the area on the screen based on the assumed size of the person. Calculate the size. When the calculated region includes an extraction region to which another label is added, the label is reassigned so that the same label is obtained. On the other hand, if there is only one label in the screen, no processing is performed.

  Next, noise removal processing is executed by the extraction means 4a of the image processing unit 4 (ST5). In the noise removal process, it is determined whether or not each label obtained in the labeling process of ST4 is noise extraction. If it is determined that noise extraction is performed, the label is removed. For example, when the number of pixels of the label is equal to or less than the set value, the label is removed as noise extraction. Note that the set value at that time is stored in the storage unit 3 in advance.

  Next, tracking processing is executed by the tracking means 4b of the image processing unit 4 (ST6). In the tracking process, the label area that has been subjected to the labeling process of ST4 and the noise removal process of ST5 is associated with the tracking label area of the previously acquired image. Specifically, it is determined whether the tracking label area of the previously acquired image and the label area of the currently processed image are based on the same tracking object based on the positional relationship in the image. If it is determined that the tracking object is the same tracking object, the same tracking label as the tracking label area of the previously acquired image is applied again. For a label area that appears in the image currently being processed and is not associated with any of the previous tracking label areas, tracking is started as a newly appearing tracking label area, and a new Give a tracking label. Further, for tracking label areas that existed in the previous time and that are not associated with any of the label areas that appear in the currently processed image, tracking is terminated and the label is discarded.

  Next, a feature amount calculation process is executed by the feature amount calculation means 4c of the image processing unit 4 (ST7). In the feature amount calculation process, a feature amount for determining an intruder is calculated for the tracking label area obtained in the tracking process of ST6. Specifically, the humanity of the tracking label area is obtained by using the width / height in the real space estimated by the perspective transformation, the movement amount, and the like.

  Next, intruder determination processing is executed by the determination means 4d of the image processing unit 4 (ST8). In the intruder determination process, the feature amount for determining the intruder obtained in the feature amount calculation process in ST7 is used to determine whether or not the tracking label area is an intruder according to a predetermined determination condition. For example, the humanity of the tracking label area is calculated for each frame by a feature amount calculation process to be described later, and as a determination condition, when the cumulative value of humanity from the appearance time is equal to or greater than a threshold and the movement amount is equal to or greater than the threshold It is determined that the tracking label area is an intruder. The threshold value at that time is stored in the storage unit 3 in advance.

  In the flowchart of FIG. 4, loop 1 means that each process of ST2 to ST8 is executed every time one frame of image is acquired, and loop 2 executes each process of ST7 to ST8 by the number of label areas. Means.

(Details of feature value calculation processing)
Next, the feature amount calculation processing (ST7) executed by the feature amount calculation means 4c of the image processing unit 4 in the flowchart of FIG. 4 will be described with reference to FIG.

  The feature amount calculation means 4c of the image processing unit 4 first performs circumscribed rectangular coordinate calculation processing of the tracking label area (ST11). In the circumscribed rectangle coordinate calculation process of the tracking label area, the coordinate values of the circumscribed rectangle vertices circumscribing the tracking label area and the width and height of the circumscribed rectangle are calculated for the tracking label area obtained in the tracking process of ST6.

  Next, a perspective transformation process is executed (ST12). In the perspective transformation process, the coordinate values, widths and heights of circumscribed rectangle vertices obtained in the circumscribed rectangle coordinate calculation process of the tracking label area in ST11, the setting information stored in the storage unit 3, the setting of the image monitoring apparatus 1 is installed. Using the height and depression angle, the estimated width and height of the tracking label area in real space are calculated by perspective transformation.

  Next, the tracking label area movement amount calculation process is executed (ST13). In the tracking label area movement amount calculation process, the movement amount of the tracking label area calculated in the tracking process of ST6 is calculated. For example, the center-of-gravity movement distance between frames is determined from the center-of-gravity coordinates of the tracking label area of the previous frame and the center-of-gravity coordinates of the current tracking label area, and the value of this center-of-gravity movement distance is used as the movement amount.

  In the tracking process of ST6, when the target tracking label area is a newly appearing tracking label area, since the previous tracking label area does not exist, the movement amount is set to 0, and an area described later The occupation ratio is 0.

When the movement amount calculation processing of the tracking label area in ST13 is executed, it is first determined whether or not the movement amount of the tracking label is equal to or larger than a predetermined minimum movement amount as the in-area occupation ratio processing (ST14). Here, the minimum movement amount refers to the minimum movement amount that can be regarded as the tracking label area moving. In this embodiment, the minimum movement amount is specifically set in advance in the storage unit 3 as 2 pixels. Suppose that If it is determined in ST14 that the movement amount of the tracking label is equal to or greater than the minimum movement amount, the movement period is incremented (+1) (ST15). Here, the movement period means that the movement amount of the tracking label area is not less than a predetermined movement amount (minimum movement amount) in the period from the appearance time when the label area appears until the determination time for determining whether or not the intruder is present. In this embodiment, this is calculated as the cumulative number of frames added in ST15 from the frame in which the label area appears to the determination frame. On the other hand, if it is determined that the amount of movement of the tracking label is not equal to or greater than the minimum amount of movement, the in-area occupation ratio calculation process is terminated and the in-area occupation ratio is not calculated (updated), but the apparatus closest to ST22, which will be described later. The process shifts to humanity calculation processing.
As described above, in this embodiment, the in-area occupation ratio is calculated only when the movement amount of the label area in the predetermined time is greater than or equal to the predetermined movement amount (minimum movement amount). That is, it is equivalent to calculating the occupation ratio in the area in the evaluation period (movement period) with the movement period as the evaluation period in the present invention. As described above, if the label area is a small animal, the label area is easily moved away from the lower end of the monitoring image when the label area is moving. By using it, a person and a small animal can be distinguished with high accuracy. That is, when a small animal stays near the lower end of the monitoring image, if the above processing is not performed, the feature amount (occupation ratio in the region) is calculated to be large, so that the small animal is erroneously determined to be a person. There is a fear. However, by calculating the occupancy ratio in the evaluation period as the evaluation period, the period (movement period) in which the movement amount of the label area is greater than or equal to the predetermined movement amount (minimum movement amount) as in the above process, Since it is possible to calculate by excluding the residence period of the small animal from the evaluation period, it is possible to discriminate between the person and the small animal with high accuracy without erroneously discriminating the staying small animal as a person.
In ST14, for the processing in ST21 described later, the frame number and the movement determination result here are stored in the storage unit 3 in association with each other.

Next, when it is determined in ST14 that the movement amount of the tracking label is equal to or greater than the predetermined minimum movement amount, and the movement period is incremented (+1) in ST15, the tracking label area is within the predetermined area from the lower end of the monitoring image. (ST16). That is, it is determined whether or not the lower end y coordinate of the circumscribed rectangle obtained by the circumscribed rectangle coordinate calculation process of the tracking label area is located within a predetermined y coordinate (within 10 pixels) from the lower end of the screen.
If it is determined in ST16 that the lower end of the tracking label exists within the predetermined area from the lower end of the monitoring image, the in-area existence period is incremented (+1) (ST17). On the other hand, if it is determined that the lower end of the tracking label does not exist within the predetermined area from the lower end of the monitoring image, the process proceeds to ST18. In ST17, for the processing in ST21 described later, the frame number and the determination result of the presence / absence of the area are stored in the storage unit 3 in association with each other.

Next, if it is determined in ST16 that the tracking label area exists within the predetermined area from the lower end of the monitoring image, is the moving period of the change area equal to or longer than a predetermined minimum evaluation period (for example, a fixed period of 10 frames)? It is determined whether or not (ST18). If it is determined in ST18 that the movement period is equal to or longer than the minimum evaluation period, the process proceeds to ST19, and the process for calculating (updating) the area occupation ratio in ST19 to ST21 is continued. On the other hand, if it is determined that the movement period is not equal to or shorter than the minimum evaluation period, the process shifts to the personality calculation process closest to the apparatus in ST22, which will be described later, without calculating (updating) the in-area occupation ratio.
As described above, when the movement period corresponding to the evaluation period in the present invention is equal to or longer than a predetermined period (minimum evaluation period), the feature amount calculating unit 4c calculates the feature amount (occupation ratio in the region). It is possible to avoid calculating and determining a feature amount (occupation ratio in the region) with a short evaluation period. That is, the feature amount (occupation ratio in the region) calculated by a short evaluation period is likely to appear as an extreme value, and an error is likely to occur in the determination result. Therefore, the evaluation period is equal to or longer than a predetermined period (minimum evaluation period). A highly reliable determination result can be obtained by calculating the feature amount (occupation ratio in the region) for the first time after elapse.

Next, when it is determined in ST18 that the moving period of the change area is equal to or longer than a predetermined minimum evaluation period, the moving period is a predetermined past period (a period retroactive to the past by a predetermined period from the determination time). For example, it is determined whether or not it is longer than a fixed period of 20 frames (ST19). If it is determined that the movement period is not equal to or longer than the past period, the in-area occupation ratio is calculated by “in-area existing period in movement period ÷ movement period” (ST20). That is, in ST20, the occupation ratio in the area is calculated using the movement period as the evaluation period in the present invention.
On the other hand, if it is determined that the movement period is equal to or longer than the past period, a period of the movement period that is retroactive from the determination time by a predetermined period of time (hereinafter referred to as “the past period within the movement period”). The intra-region existence period in the past period within the movement period is calculated. Specifically, first, information indicating the association between the frame number stored in the storage unit 3 and the movement determination result in ST14 is read, and the frame number determined to be moving is determined from the determination time. By extracting the frame number that goes back by the number of frames in the past period (here, 20 frames), the frame number corresponding to the past period in the movement period is obtained. Then, in ST17, information indicating the association between the frame number stored in the storage unit 3 and the determination result of the presence / absence of the area is read, and the frame number corresponding to the past period within the moving period exists in the predetermined area. The frame number determined to have been extracted is extracted, and the number of frames is calculated as “the existing period in the area in the past period in the movement period”. The past period within the movement period is 20 frames. Subsequently, the occupation ratio in the area is calculated by “the existing period in the area in the past period in the movement period ÷ the past period in the movement period” (ST21). That is, in ST21, the in-area occupation ratio is calculated using the past period in the movement period as the evaluation period in the present invention.
Thus, by switching the evaluation period for calculating the in-region occupation ratio based on whether or not the movement period is equal to or longer than the past period, it is possible to distinguish the person and the small animal with high accuracy. In other words, if the moving object is in the monitoring range for a long time while moving (when the moving period is long), calculating the occupation ratio in the area with the moving period as the evaluation period will increase the evaluation period of the parameter. In some cases, it is difficult to make a clear difference in the feature amount (occupation ratio in the region) for discriminating between a person and a small animal. Therefore, in this case, by calculating the feature amount (occupation ratio in the region) with the period (the past period in the movement period) that is back by the time of the past period determined in advance from the determination time in the movement period as the evaluation period, It is possible to distinguish and determine a person and a small animal with relatively high accuracy.

  Note that the minimum movement amount, the minimum evaluation period, and the past period are stored in the storage unit 3 in advance. Further, the “movement period” and “in-area existence period” are stored in the storage unit 3 in association with each tracking label area. Further, the area occupation ratio calculated in ST20 or ST21 is held for each tracking label area until the next area occupation ratio is calculated and updated, and the area occupation ratio is calculated by the area occupation ratio calculation process. Even when the occupation ratio is not calculated (that is, in the case of No in ST14 or ST18), it can be used in the process after the humanity calculation process closest to the apparatus described later.

As described above, in this embodiment, as one of the feature amounts for determining whether or not a person is a person, an “area occupation ratio” that is a ratio of the area existing period to the evaluation period is calculated. Yes. That is, the occupation ratio in the area is calculated as “in-area existing period in evaluation period ÷ evaluation period”.
At this time, in this embodiment, the “evaluation period” is set to “a period during which an amount of movement greater than or equal to a predetermined amount in the appearance period (movement period)” or “a past period that is predetermined from the determination time in the movement period”. The period occupied in the area is calculated as the ratio of the period existing in the area and the evaluation period in the evaluation period.
However, the present invention is not limited to this, and the “evaluation period” may be “appearance period” or “past period”, and the in-area occupation ratio may be calculated as the ratio between the in-area existence period and the evaluation period in the evaluation period. That is, this corresponds to calculating the occupation ratio in the region without determining the movement amount of the tracking label in ST14.
Therefore, as shown in FIG. 3, the in-area occupation ratio in the present invention is (1) the ratio of the in-area existence period in the appearance period to the appearance period, and (2) the in-area existence period in the past period in the past period. It can be calculated by the ratio, (3) the ratio of the in-area existence period in the movement period to the movement period, and (4) the ratio of the in-area existence period in the past period in the movement period to the past period.
Note that the above-described formula for calculating the occupation ratio in the area is stored in the storage unit 3 in advance.

By the way, as shown in FIG. 6, the pixel existence distance between the lower end y coordinate of the screen and the lower end y coordinate of the label circumscribing rectangle is calculated (the image lower end y coordinate−the lower end y coordinate of the label circumscribing rectangle). It is also possible to prepare an evaluation function in accordance with) and use the output value (corresponding to the in-region existence period). For example, when the target label circumscribing rectangle exists at a position shifted by 2 pixels from the lower end of the screen, the in-region existence period is set to +0.8 based on the evaluation function of FIG.
Thereby, when the lower end of the label circumscribing rectangle and the lower end of the monitoring image are in contact with each other, the label area is most likely to be a person, and a large in-area occupation ratio is calculated. Even if the lower end of the circumscribed rectangle of the label is not in contact with the lower end of the monitoring image, if the distance is relatively small, an error caused by the extraction accuracy problem (for example, the person's clothes are very similar to the background) Since there is a strong suspicion of an error (because it is difficult to extract a certain location as a change area), a relatively large area occupation ratio is calculated. On the other hand, even if there is a label area within the range of distance that can be regarded as being substantially in contact with the lower end of the monitoring image, the larger the distance, the less likely the error is due to the problem of extraction accuracy. If it does not exist, the occupation ratio in the smaller area is calculated. In this way, by calculating the occupancy ratio in the area in consideration of the distance between the lower end of the label area and the lower end of the monitoring image, the person and the small animal are accurately distinguished and determined regardless of the extraction accuracy of the label area. be able to. And the misreporting by considering a person as a small animal can be reduced, and the misinformation by considering a small animal as a person can be reduced.
Note that the error due to the above-described extraction accuracy appears more prominently when a thermal image sensor is used as the imaging unit 2 because the difference between the clothing temperature of the human image and the background temperature does not appear clearly. Therefore, when a thermal image sensor is used as the imaging unit 2, the above processing is utilized to function more effectively, and a person and a small animal can be distinguished and determined with high accuracy.

  When the area occupancy ratio is calculated in the area occupancy ratio calculation process in ST14 to ST21, the process proceeds to the humanity calculation process closest to the apparatus (ST22).

  Next, in the humanity calculation process closest to the apparatus in ST22, the lower end y coordinate of the circumscribed rectangle obtained in the circumscribed rectangle coordinate calculation process of the tracking label area in ST11, and the tracking label area circumscribed rectangle obtained in the perspective transformation process in ST12 Using the estimated width / estimated height in real space and the above-mentioned area occupation ratio as parameters, the humanity closest to the apparatus is calculated. The humanity obtained here is used as information (feature amount) for determining whether or not the target tracking label area is an intruder in the intruder determination process of ST8.

  It should be noted that the personality calculation process closest to the apparatus in ST22 sets the humanity closest to the apparatus to 1 only when the above-described parameters satisfy the respective conditions, and 0 otherwise. Calculate as follows. Specifically, (1) the lower end y-coordinate of the circumscribed rectangle is within a predetermined range from the lower end of the monitoring screen, (2) the estimated width in real space is in the range of 20 cm to 100 cm, (3) the actual The estimated height in space is within the range of 70 cm to 160 cm, (4) the condition of each parameter is defined as the area occupation ratio is within the range of 0.9 to 1.0, and (1) to (4) Only when all the conditions are satisfied, the personality close to the apparatus is set to 1, and when any one of the conditions (1) to (4) is not satisfied, the personality close to the apparatus is set to 0.

  Subsequently, a humanity calculation process other than the closest to the apparatus is executed (ST23). In the humanity calculation process other than the apparatus closest to the apparatus, the humanity for distinguishing the intruder and the small animal at a place away from the image monitoring apparatus 1 (hereinafter referred to as “personality other than the apparatus closest to the apparatus”) is calculated. Specifically, for the target label area, (1) the estimated width in the real space is 60 cm to 100 cm, (2) the estimated height in the real space is 140 cm to 200 cm, (3) the area is occupied The condition that the ratio is in the range of 0 to 0.1 is defined, and only when all the conditions (1) to (3) are satisfied, the humanity other than the closest to the apparatus is set to 1, and the conditions (1) to (3) If one of the above is not satisfied, the humanity other than the closest to the device is set to zero.

  Next, humanity calculation processing is executed (ST24). In this humanity calculation process, the humanity closest to the apparatus obtained in the above-described personality calculation process in ST22 and the humanity other than the apparatus immediate obtained in the humanity calculation process other than the apparatus immediate in ST23. Based on the above, the humanity of the target tracking label area is calculated. Specifically, the humanity closest to the apparatus is compared with the humanity other than the apparatus immediate, and the larger value is adopted as the humanity value.

  By the way, in the processing shown so far, if the target label area is from an intruder closest to the device, the person-likeness of the device is high, and if the intruder is away from the device, a person other than the device closest to the device Conditioning is performed to increase the appearance. For this reason, by adopting as the larger value of the humanity closest to the device and the humanity other than the device closest, the position of the intruder is either near the device or away from the device, It is always possible to increase humanity.

  The present invention is not limited to the above embodiments, and may be implemented in various different embodiments within the scope of the technical idea described in the claims. Moreover, the effect described in the Example is not limited to this.

In the above embodiment, as the feature quantity for discriminating between a person and a small animal in the feature quantity calculation process in the feature quantity calculation unit 4c, the “area” represents the ratio of the change area existing within the predetermined area from the lower end of the monitoring screen. The “occupancy ratio” is calculated.
However, the present invention is not limited to this, and the feature quantity calculation means 4c uses the time period during which the lower end of the change area (the circumscribed rectangle) is in contact with the lower end of the monitoring screen as the feature quantity instead of the “occupation ratio in area”. The “lower end contact ratio” representing the target ratio may be calculated. In this case, the “in-area existence period” in the above-described embodiment may be calculated by replacing the “lower end contact period” that represents the period in which the lower end of the change area (the circumscribed rectangle) is in contact with the lower end of the monitoring image. That is, in the determination in ST16 of FIG. 5, it is determined whether or not the tracking label is in contact with the lower end of the monitoring image. In the increment process in ST17 of FIG. 5, the “in-area existence period” is replaced with the “lower end contact period”. The calculation formula of the lower end contact ratio in ST20 of FIG. 5 is “lower end contact ratio = lower end contact period in moving period ÷ movement period”, and the lower end contact ratio calculation formula in ST21 is “lower end contact ratio = past period in moving period”. The lower end contact period in ÷ the past period ”.
Thus, the lower end contact period indicating the period during which the change area in the monitoring image is in contact with the lower end of the image is obtained, and the change area is a person image based on the feature amount corresponding to the ratio of the lower end contact period to the evaluation period. In order to discriminate between a person and a small animal by determining whether or not the above-described change area calculates a feature amount according to the ratio of the change area existing in the predetermined area from the lower end of the monitoring image. It is possible to make a determination with higher accuracy by capturing a more characteristic event (an event in which a change area by a person is more frequently in contact with the lower end of a monitoring image than a change area of a small animal). That is, even when the small animal is moving laterally along the vicinity of the lower end of the monitoring image, it is possible to discriminate between the person image and the small animal. Furthermore, compared to the case where the above-described change area calculates the feature amount according to the ratio of the change area existing in the predetermined area from the lower end of the monitoring image, the information amount is reduced and the human body and the small animal in the vicinity of the lower end of the image are reduced. It can be distinguished and determined.

In the above-described embodiment, the feature amount calculation unit 4c uses the feature space to estimate the size in the real space (estimated width / estimated height) estimated by the perspective transformation from the size of the change region on the monitoring image, and the occupation ratio in the region ( Or the lower end contact ratio), etc., and the humanity (personality close to the apparatus, humanity other than the apparatus closeness) is calculated based on these feature amounts, and the determination means 4d determines the change region based on these humanities. Judging whether a person or a small animal.
However, the present invention is not limited to this, and the feature amount calculation unit 4c calculates only the area occupation ratio (or the lower end contact ratio) as the feature amount. Based on the feature amount, the determination unit 4d determines whether the change region is a person. You may determine whether it is a small animal.
That is, a series of processes for calculating humanity in ST22 to ST24 is omitted, and as a determination condition in ST8, for example, “when the area occupation ratio (or lower end contact ratio) is equal to or higher than a predetermined value (or a predetermined value) When the above occupancy ratio in the area continues for a predetermined frame or more), the tracking label area is determined to be an intruder, and based on the determination condition, it is determined whether the determination means 4d is a person or a small animal. May be.

DESCRIPTION OF SYMBOLS 1 Image monitoring apparatus 2 Imaging part 3 Memory | storage part 4 Image processing part 4a Extraction means 4b Tracking means 4c Feature-value calculation means 4d Determination means 5 Output part

Claims (3)

An image monitoring apparatus that sequentially acquires monitoring images obtained by imaging a monitoring space, and performs image processing on the monitoring image to determine the presence or absence of a person in the monitoring space,
An extraction unit that compares the monitoring image with a reference image of a monitoring space stored in advance and extracts a changed region from the reference image;
A tracking unit that temporally tracks the change region;
An in-region existence period is obtained in which a lower end of the change region and a lower end of the monitoring image can be regarded as being substantially in contact in a predetermined evaluation period from the appearance time of the change region to a determination time for determining the presence or absence of a person. A feature amount calculation unit that calculates a feature amount that the change region is likely to be a human image as the ratio of the in-region existence period to the evaluation period increases;
A determination unit that determines whether or not the change region is a person image using the feature amount;
An image monitoring apparatus comprising: The feature amount calculation unit, when calculating the feature amount, corrects the time length of the in-region existence period to be shorter as the distance between the lower end of the change region and the lower end of the monitoring image becomes larger. The image monitoring apparatus described. 3. The image monitoring apparatus according to claim 1, wherein the feature amount calculation unit calculates only one or a plurality of periods in which the change region has moved by a predetermined amount or more as the predetermined evaluation period.

Images