JPH0514889A - Image monitoring device - Google Patents

Abstract

PURPOSE:To prevent erroneous detection by dividing approaching plural objects which have intruded in so as to decide whether they are target intruding objects or not. CONSTITUTION:An image signal from an ITV camera 1 is converted into a digital signal by an A/D converter 21 and then it is stored in a memory 22 as an image data. A change area extracting part 23 excutes differential calculation between the image data taken into the memory 22 and that newly taken in at a specified time so as to extract the change area of images. A change area large dividing part 24 largely divides the change area extracted by the change area extracting part 23 so as to enclose them with a rectangle. A changing object extracting part 25 divides the large change area enclosed with a rectangle into a lot of sections and then separately extracts each changing object. A changing object deciding part 26 decides whether the changing object extracted by the changing object extracting part 25 is the panticular intruding object to be detected or not, and then activates an alarming device 6 depending on the deciding result.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image monitoring device for detecting an intruding object in a monitoring area.

[0002]

2. Description of the Related Art One example of this type of image monitoring apparatus is shown in FIG. That is, for example, I
The TV camera (imaging means) 1 captures an image in the monitoring area 7 and converts it into an electric signal. The image signal captured by the ITV camera 1 is sent to the processing device 3 and the display device 5 via the transmission line 2. The display device 5 displays the images captured by the ITV camera 1, and the processing device 3 continuously captures the images captured by the ITV camera 1 and performs image processing and determination for intruding object detection. As a result, when an intruding object is detected, an intruding object detection is displayed on the screen of the display device 5, and an alarm sound is emitted by the alarm device 6.

In the processing device 3, for example, the following processing is performed. That is, first from the ITV camera 1,
The images in the monitoring area 7 are continuously captured by the processing device 3. The processing device 3 performs analog-digital conversion on the captured image signal at regular intervals and stores it in an image memory (not shown).

Next, inter-pixel processing as described below is performed. That is, an image is captured at a certain time t = t0,
After that, this image data at time t = t0 and time t = t1
Then, the difference in brightness in pixel units from the image data captured at is calculated. Since this result includes the noise component of the image, it is binarized with a certain constant value (threshold of brightness) to extract the original change amount of the image. That is, the difference image is binarized with a predetermined threshold value. This is to set a pixel whose difference value is equal to or larger than a predetermined threshold value to "1" and to set a pixel whose difference value is equal to or smaller than the predetermined threshold value to "0". Is detected as ".

When the change area is extracted in this way, next,
A rectangle circumscribing this change area is obtained. That is, first, the binarized image is projected on the x-axis (horizontal axis). The area where the projection value is large is the change area. Therefore, next, a portion larger than the constant value is projected on the y-axis (vertical axis). An area where the projection value is equal to or larger than a certain value can be set as a change area. By connecting the change regions of the x and y axes, a rectangle surrounding the change regions can be obtained. If the size of the rectangle is larger than a certain value (threshold value of the number of pixels), the processing device 3 operates the alarm device 6 to detect an intruding object.

The threshold value of the brightness is set in the monitoring area 7
, And the threshold value for the number of pixels differs depending on the size of the intruding object to be detected, so it is necessary to set these two threshold values in advance. The threshold value is set, for example, by the light pen 4 and the display device 5.

[0007]

However, in the above-described conventional image monitoring apparatus, when a plurality of intruding objects approach each other or when they partially overlap each other, the plurality of intruding objects are collected. There is a problem in that it may be detected as one intruding object, which may make the recognition of the intruding object erroneous.

Further, since all intruding objects are approximated by rectangles, different objects may be regarded as the same object if they have the same size, which may cause misrecognition of the intruding object. There was a problem.

Therefore, the present invention provides an image monitoring apparatus capable of dividing a plurality of intruding objects even if they are close to each other to determine whether or not the intruding object is a target and preventing erroneous detection. The purpose is to do. Another object of the present invention is to provide an image monitoring apparatus capable of identifying an intruding object, determining whether or not the intruding object is a target, and preventing erroneous detection.

[0010]

An image monitoring apparatus according to a first aspect of the present invention continuously captures images captured by the image capturing means and an image capturing means for capturing an image in a monitoring area, and captures the captured images. By processing the image, a change area extracting means for extracting and extracting the change area group of the image in the monitoring area in the global area, and the global change area extracted by the change area extracting means are divided into individual change objects. The changing object extracting means for extracting the changing object and the judging means for judging whether or not the changing object extracted by the changing object extracting means is an intruding object of a purpose to be originally detected.

An image monitoring apparatus according to a second aspect of the present invention comprises an image pickup means for picking up an image in a monitoring area, an image picked up by the image pickup means, and a processed image. , A change area extracting means for extracting and extracting a change area of an image in the monitoring area, and a change area extracted by the change area extracting means is divided into a plurality of parts to represent a change object as a set of a plurality of figures. By the changing object dividing means and the changing object represented by the changing object dividing means as a set of a plurality of figures with a reference pattern registered in advance in the dictionary, the purpose which the changing object should originally detect And a determining means for determining whether the object is an intruding object.

[0012]

According to the first aspect of the present invention, the change region group in the monitoring region is cut out and extracted in the global region, and the extracted global change region is divided into individual change objects to be extracted. Even if multiple intruding objects are close to each other, it is determined whether or not the changing object is the intruding object of the intended purpose by determining whether or not the intruding object is the intended intruding object. Therefore, false detection can be prevented.

According to the second aspect of the present invention, the change object extracted from the monitoring area is divided into a plurality of pieces to represent the change object as a set of a plurality of figures, and the change object represented as a set of the plurality of figures. By comparing with the reference pattern registered in the dictionary in advance, it is possible to identify the intruding object by determining whether the changing object is the intruding object of the intended purpose to be detected. Since it can be determined whether or not it is an object, false detection can be prevented.

[0014]

Embodiments of the present invention will be described below with reference to the drawings. Note that the same parts as those in FIG.

FIG. 1 schematically shows the structure of an image monitoring apparatus according to the first embodiment. In the figure, the image signal from the ITV camera 1 is converted into a digital signal by the A / D converter 21, stored in the memory 22 as image data, and sent to the change area extraction unit 23. The changed area extraction unit 23 extracts a changed area of the image by performing a difference operation between the image data captured in the memory 22 and the image data newly captured after a predetermined time.

The change area global division unit 24 divides the change area group extracted by the change area extraction unit 23 into a large area and encloses it in a rectangle. The changing object extracting unit 25 individually extracts each changing object by finely dividing the global changing area surrounded by the rectangle. The change object determination unit 26 determines whether or not the change object extracted by the change object extraction unit 25 is an intruding object that is a target to be originally detected, and operates the alarm device 6 according to the determination result.

2 to 6 show examples of display screens of the display device 5 in this embodiment. Figure 2 shows the normal state, I
The image captured by the TV camera 1 is usually displayed on the display device 5 in real time. FIG. 3 shows the monitoring area 7
There is an intruding object 11 inside, and that is an image in which the image changes. FIG. 4 is an image in which this changed area is extracted, and the changed pixels are extracted. Figure 5
It is an image in which this changing region group is divided into large regions and surrounded by a rectangle 14. FIG. 6 shows a case where a changing object is separated and cut out, and an approaching object is separated and cut out, and each is surrounded by a rectangle. As a result, based on the size of the rectangle, it is determined whether or not the extracted change object is an intruding object that is a target to be originally detected.
Is operated.

The intruding object detection process of this embodiment will be described in detail below with reference to the flowchart shown in FIG.
First, as a background image, the image signal from the ITV camera 1 is A / D converted by the A / D converter 21 and stored in the memory 22 as image data (see FIG. 17). Next, the image data (see FIG. 18) after a certain time is taken in. Then, the change area is extracted by comparing the two image data. After that, images are continuously captured at regular intervals and extraction processing is performed.

In this change area extraction processing routine, as shown in FIG.
Pixels are processed according to FIGS.
That is, the captured image at a certain time t = t0 is shown in FIG.
Then, this image becomes the background image, and the image data obtained by A / D converting this image and storing it is shown in FIG. And t =
At time t1, an intruding object 11 enters the surveillance area 7,
FIG. 17 shows the image data stored by A / D converting the captured image when the image changes. The difference between the brightness of each of these image data is calculated for each pixel, and this is performed for all pixels to obtain a difference image as shown in FIG. However, since this difference image contains noise components, it is binarized with a certain fixed value (threshold of brightness),
Extract the original image change. In FIG. 19, the difference image of FIG. 18 is binarized with “threshold = 4”, for example. That is, pixels of "4" or more are set to "1", and pixels of "4" or less are set to "0". As a result, the pixel 12 in the original change area is detected as "1".

Next, the projection of this binarized image (see FIG. 19) is calculated. That is, first, the projection in the x-axis direction is calculated by adding the binarized values for each x-coordinate of the image. This projection value is indicated by the symbol a in FIG. Here, an area having a small projection value is considered to be a noise component, and therefore, when the portion above the noise component is extracted, the x component of the actual change area is cut out. The projection in the y-axis direction is calculated for this change area in the x-axis direction. This projection value is shown by the symbol b in FIG. Here, when a portion having a certain value or more is extracted, a rectangle 14 representing the large area of the change region is obtained as shown in FIG.

By the processing up to this point, the rough shape of the change area is extracted. However, if a plurality of intruding objects are close to each other as shown in FIG. 3, they are collectively extracted as one object. . Therefore, in the present embodiment, the rough rectangle is divided into more detailed parts to extract the changing object independently. The processing method is shown in FIG.

That is, the globally divided rectangle is divided in the y-axis direction, and the projection is calculated in the x-axis and Y-axis directions in the same manner as before to generate a rectangle. In the present embodiment, the width of the division is the y-axis direction length O1 of the object. In FIG. 6A, the projection of the upper part M11 in the x-axis direction is shown above, and the lower part M11 is shown.
Twelve x-axis projections are shown below. Here, as shown in FIG. 6B, the projection of the global rectangle in the y-axis direction is calculated for the large part of the projection in the x-axis direction. Then, the changing object is individually cut out from the rough rectangle to obtain a fine rectangle.

When the small rectangle indicating the changing object is close to the size of the reference rectangle set in advance, for example, it is determined that the changing object is an intruding object which is a target to be detected, and the alarm device 6 is activated. Activate. Otherwise,
Acquire the next image without activating the alarm device 6,
Perform new processing.

As described above, according to the first embodiment, when there is an intruding object in the surveillance area imaged by the ITV camera 1 and it is extracted as a change area of the image, the change area group is set as a large area. Even if multiple intruding objects are close to each other, they are divided and surrounded by a rectangle, and the rectangle is finely divided to extract the changing objects independently. Since it can be determined whether or not it is possible to prevent erroneous detection.

FIG. 8 schematically shows the structure of the image monitoring apparatus according to the second embodiment. The same parts as those of the first embodiment shown in FIG. In the figure, the image signal from the ITV camera 1 is represented by the A / D converter 2
At 1, the signal is converted into a digital signal, stored in the memory 22 as image data, and sent to the change area extraction unit 23. The changed area extraction unit 23 extracts a changed area of the image by performing a difference operation between the image data captured in the memory 22 and the image data newly captured after a predetermined time.

The changing object extracting section 27 is a changing area extracting section 2
The change area extracted in 3 is surrounded by a rectangle. The changing object dividing unit 28 finely divides the changing area surrounded by the rectangle to represent the changing object as a set of a plurality of small rectangles. The change object determination unit 29 compares the change object represented by the set of the plurality of small rectangles with the reference pattern registered in the dictionary memory 30 in advance, so that the change object extraction unit 27 extracts the change. It is determined whether or not the object is an intruding object for the purpose to be originally detected, and the alarm device 6 is operated according to the result of the determination.

9 to 12 show examples of display screens of the display device 5 in this embodiment. FIG. 9 is an image in which there is an intruding object 11 in the monitoring area 7 and the image changes. FIG. 10 is an image in which this changed area is extracted, and the changed pixels are extracted. Figure 1
Reference numeral 1 is an image in which this changing object is cut out and surrounded by a rectangle 17. FIG. 12 is a result of representing the changing object in detail by a set of a plurality of small rectangles. As a result, by comparing the changing object represented by this set of small rectangles with the reference pattern, it is determined whether or not the extracted changing object is the intruding object of the intended purpose to be detected. If the object is determined to be an object, the alarm device 6 is activated.

The intruding object detection processing of this embodiment will be described in detail below with reference to the flowchart shown in FIG. Note that the processing from the capturing of the background image to the acquisition of the binarized image of FIG. 19 is the same as that in the first embodiment described above, so a description thereof will be omitted and the subsequent processing will be described.

When a binarized image is obtained as described above, the projection of this binarized image is calculated. That is, first, the projection in the x-axis direction is calculated by adding the binarized values for each x-coordinate of the image. This projection value is indicated by reference sign c in FIG. Here, an area having a small projection value is considered to be a noise component, and therefore, when the portion above the noise component is extracted, the x component of the actual change area is cut out. This x
The projection in the y-axis direction is calculated for the change area in the axial direction. This projection value is shown by reference numerals d, e, f in FIG. Here, when a portion having a certain value or more is extracted, a rectangle 17 representing a changing object is obtained as shown in FIG.

Although the rough shape of the changing region is extracted by the processing up to this point, it is difficult to identify the changing object. Therefore, in this embodiment, this rough rectangle is divided into a plurality of small rectangles in more detail. The processing method is shown in FIG.

That is, FIG. 12A is a rectangle obtained by cutting out the change area and the area. The y axis is divided into the length of the y axis of the target object, and the projection is taken in the x axis direction to form a rectangle. Split. In this embodiment, the rectangular area is divided into four equal parts to obtain the projection. FIG. 12B shows the small rectangles M21 to M2 of FIG.
4 shows an x-axis projection of 4. Then, when a portion equal to or larger than a predetermined threshold is cut out, the rough rectangle is divided into fine small rectangles. This state is shown in FIG. 12
What is shown in (d) is the four small rectangles generated by this embodiment, and the matching process with the dictionary pattern is next performed.

This matching processing method is shown in FIG.
The set of small rectangles generated as described above is matched with the dictionary pattern. For example, in this embodiment, after first classifying by the size of the changing object, secondly normalizing the rectangle to obtain the area of the difference from the reference pattern in the dictionary memory 30. In the dictionary memory 30, a certain target object is ideally fetched, a rectangular set is generated as described above, this is normalized to a certain size, and registered as a reference pattern. A plurality of these patterns are registered, and an example of the reference pattern is shown in FIG.

FIG. 13A shows a set of small rectangles generated from the change object extracted as described above and normalized. Both of them are overlapped as shown in FIG. 13C to perform matching. That is, FIG. 13 (a) and FIG.
3 (b) and P1 to P4 in this order. However, in order for humans to respond to movements of objects such as moving their necks and legs,
If there is a slight misalignment in the x-axis direction, correction is performed so that the difference between the changing object and the reference pattern is reduced. P4
For this, this correction is performed.

Then, by finding the reference pattern having the smallest total area of the difference D (the hatched portion in FIG. 13c) between the pattern obtained by overlapping the two and the reference pattern,
The changing object at that time can be determined to be the reference pattern. As a result, when it is determined that the changing object is an intruding object that is the original object to be detected, the alarm device 6 is activated. In other cases, the next image is captured without activating the alarm device 6 and a new process is performed.

As described above, according to the second embodiment, when an intruding object exists in the surveillance area imaged by the ITV camera 1 and is extracted as a change area of the image, the change area is rectangular. By enclosing, and further dividing the rectangle into, the changing object is expressed by a set of a plurality of small rectangles, and by matching with the reference pattern, the intruding object can be specified and it can be determined whether or not the intruding object is the target. Therefore, false detection can be prevented. Needless to say, the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the spirit of the present invention.

[0036]

As described in detail above, according to the present invention, even when a plurality of intruding objects are approaching, it is possible to divide them and determine whether or not the object is a target intruding object, and prevent erroneous detection. It is possible to provide an image monitoring device capable of performing the above. Further, according to the present invention, it is possible to provide an image monitoring apparatus capable of identifying an intruding object, determining whether or not the intruding object is a target, and preventing erroneous detection.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of an image monitoring apparatus according to a first embodiment of the present invention.

FIG. 2 is a diagram showing an example of a display screen during normal monitoring.

FIG. 3 is a diagram showing an image example in which there is an intruding object and the image is changed.

FIG. 4 is a diagram showing an example of an image in which a changed area is extracted.

FIG. 5 is a diagram showing an example of an image in which a changing region group is divided into large regions and surrounded by a rectangle.

FIG. 6 is a diagram illustrating a processing method of dividing and cutting out a changing object.

FIG. 7 is a flowchart illustrating intruding object detection processing.

FIG. 8 is a block diagram showing a schematic configuration of an image monitoring apparatus according to a second embodiment of the present invention.

FIG. 9 is a diagram showing an example of an image in which an image has changed due to an intruding object.

FIG. 10 is a diagram showing an example of an image in which a change area is extracted.

FIG. 11 is a diagram showing an example of an image in which a change area is surrounded by a rectangle.

FIG. 12 is a diagram illustrating a processing method in which a changing object is divided and represented by a set of a plurality of small rectangles.

FIG. 13 is a diagram illustrating a matching processing method for a reference pattern and a changing object represented by a set of a plurality of small rectangles.

FIG. 14 is a flowchart illustrating an intruding object detection process.

FIG. 15 is a configuration diagram showing an example of an image monitoring device.

FIG. 16 is a diagram showing an example of an image after capture A / D conversion at time t = t0.

FIG. 17 is a diagram showing an example of an image after the acquisition A / D conversion at time t = t1.

18 is a diagram showing an example of a difference image between the image of FIG. 16 and the image of FIG.

FIG. 19 is a diagram showing an example of a binarized image obtained by binarizing the image of FIG. 18 with a predetermined threshold value.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... ITV camera (imaging means), 3 ... Processing device, 5 ... Display device, 6 ... Alarm device, 7 ... Monitoring area, 21 ... A / D converter, 22 ... Memory, 23 ... Change area extraction part, 24 ... Change region global division unit, 25, 27 ... Change object extraction unit, 2
6, 29 ... Changing object judging unit, 28 ... Changing object dividing unit, 3
0 ... Dictionary memory.

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Shozo Abe             70 Yanagicho, Saiwai-ku, Kawasaki City, Kanagawa Prefecture             Toshiba Yanagimachi Factory

Claims (2)

[Claims] 1. A change in an image in the monitoring area by continuously capturing the images captured by the image capturing means and an image capturing means for capturing an image in the monitoring area, and processing the captured image. A change area extracting means for extracting a region group by cutting it out in a large area, a change object extracting means for dividing and extracting the global change area extracted by the change area extracting means into a single change object, and this change object extracting means An image monitoring apparatus, comprising: a determination unit that determines whether or not the extracted change object is an intruding object of a purpose to be originally detected. 2. A change in the image in the monitoring area by continuously capturing the images captured by the image capturing means and an image capturing means for capturing an image in the monitoring area, and processing the captured image. A change area extracting unit that cuts out and extracts an area, a change object dividing unit that expresses a change object as a set of a plurality of figures by dividing the change area extracted by the change area extracting unit, and the change object. By comparing the changing object represented by the dividing means as a set of a plurality of figures with a reference pattern registered in advance in the dictionary, whether or not the changing object is an intruding object of the intended purpose should be detected. An image monitoring apparatus comprising: a determination unit that makes a determination.