JP3513443B2 - Rockfall monitoring system - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a rockfall monitoring system, and more specifically, it acquires an image of an area where rockfall is to be monitored and analyzes the image to determine the presence or absence of rockfall in the area. It is related to a rockfall monitoring system that can.

[0002]

2. Description of the Related Art Conventionally, as a device for monitoring whether or not a rock fall has occurred from a cliff, a device using a wire sensor,
A device using an extensometer was used.

An apparatus using a wire sensor is an apparatus configured to attach a wire to a place where there is a risk of falling rocks, and detect the cutting of the wire to monitor the occurrence of falling rocks.

A device using an extensometer is arranged such that both ends of a wire with pulleys are attached to an object (rock mass) that is at risk of falling and an absolutely stationary object (object that is absolutely stationary), The device is configured to monitor the occurrence of rockfall by measuring the amount of change in the rotation angle of the pulley.

[0005]

However, the device using the above wire sensor or extensometer has the following disadvantages.

(1) If a rock fall occurs, the device will be destroyed, so it is necessary to carry out re-installation work in order to continue operation.

(2) When it is attempted to monitor rockfall in a wide area, the apparatus becomes complicated and the scale becomes large, so that the installation cost increases.

(3) Since it is necessary to directly attach the device to a place where there is a risk of falling, installation work of the device is often difficult.

(4) In order to grasp the accurate situation of the monitoring site, it is necessary to directly check the site.

The present invention has been made to solve the above-mentioned inconvenience, and can be installed easily and at low cost, has excellent durability, and can remotely monitor a monitoring site. The purpose is to provide a simple rockfall monitoring system.

[0011]

A rockfall monitoring system according to the present invention analyzes an image acquired by an image acquiring means for acquiring an image of an area where a rockfall (rockfall phenomenon) is to be monitored, and an image acquired by the image acquiring means. and a image analyzing means for extracting a moving object, and determining means for determining the moving object extracted by the image analysis means whether falling stone (stone dropped by falling rocks), the said image analysis means Is
Two acquired at a time by the image acquisition means
Area of the moving object extracted by extracting the moving object from the image
From within, cut out a small area image consisting of a predetermined number of pixels,
In the two images over the entire area of the moving object
Correlation with the image acquired earlier, and the correlation is minimized.
The small area is identified, and the determination means further separates the time.
In the new image acquired by
An image of a small area at the same position and the identification of the moving object
Correlation with the image of the small area that has been
If there is, it can be determined that the moving object is a falling stone.
And (the invention according to claim 1). Therefore, it is possible to obtain a rockfall monitoring system that can be installed easily and at low cost and has excellent durability.

In this case, the image analysis means obtains a difference image from the two images acquired by the image acquisition means at a time interval, and further binarizes the difference image,
The moving object is extracted by labeling (the invention according to claim 2).

Further, the judging means judges whether or not the moving object is a falling stone based on whether or not the moving object is stopped for a predetermined time or longer (the invention according to claim 3).

[0014]

Further, the rockfall monitoring system has display processing means for displaying the judgment result of the judgment means, the image acquired by the image acquisition means, and the like (the invention according to claim 4 ). Therefore, the observer can accurately determine whether or not rockfall has occurred.

In this case, the image acquisition means, the image analysis means, and the determination means are installed at a plurality of monitoring sites and are connected to the display processing means via communication networks (claim 5 ). Invention). Therefore, the surveillance staff can remotely monitor the surveillance site.

[0017]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 shows the structure of a rockfall monitoring system 10 according to an embodiment of the present invention.

The rockfall monitoring system 10 includes rockfall monitoring devices 12 (i) (i = 1 to Ni) installed at each monitoring site.
It is installed in a monitoring office located away from each of the monitoring sites, and is composed of the rockfall monitoring device 12 (i) and a display processing device 16 connected via a wireless or wired communication network 14. There is.

FIG. 2 shows an example of the monitoring site where the rockfall monitoring device 12 (i) is installed. The monitoring target area at this monitoring site is adjacent to the cliff 18 where a rockfall (rockfall phenomenon) is expected to occur, and is directly below the cliff 18 where a falling stone (stone dropped by the rockfall) C may intrude. It is an area (predicted rockfall area) 20 on the road surface 19 located at.

The road surface 19 includes roads for automobiles and pedestrians, railroad tracks for railway vehicles, and the like. Further, as the rockfall prediction region 20, a region near the entrance / exit of the tunnel, a region expected to be affected by the debris flow, and the like are assumed. In this case, the falling stone C also contains earth and sand. Further, the cliff 18 itself may be used as the rockfall prediction region 20. That is, the rockfall monitoring system 10 can be used as a disaster prevention information system, a road information system, a debris flow monitoring system, or the like.

As shown in FIG. 1, the rockfall monitoring device 12
(I) is provided with a CCD camera 22 as an image acquisition means, and this CCD camera 22 is installed so that its field of view covers the predicted rockfall region 20 shown in FIG. Then, the CCD camera 22 captures an image of the rockfall prediction area 20 every predetermined time (for example, 1 second), and the obtained image is displayed, for example, on NTSC (National Television Subcommit
A tee) type image signal Sa is output to an image transmission processing device 30 described later.

As shown in FIG. 1, the rockfall monitoring device 12
(I) includes an image transmission processing device 30.

FIG. 3 shows the configuration of the image transmission processing device 30. The image transmission processing device 30 includes an image input unit 3
2, a frame memory 34, a working memory 36,
Arithmetic control unit 40, image compression unit 42, I / O control unit 4
4 and.

The image input section 32 takes in the image signal Sa from the CCD camera 22 according to the control signal Sf from the I / O control section 44, and digitizes the image signal Sa into the frame memory 34. Supply.

The frame memory 34 stores the image data Da from the image input section 32 and outputs the image data Da to the memory 36, the arithmetic control section 40 and the image compression section 42.

The image compression unit 42 converts the image data Da from the frame memory 34 into, for example, JPEG (Joint Phot).
The image data Da ′ obtained by compression according to the ographic Experts Group) method is supplied to the arithmetic control unit 40. Note that this compression processing may be performed based on an instruction from the arithmetic control unit 40.

The image data Da from the frame memory 34 is stored in the memory 36 together with a template T which will be described later.
(J) etc. are accumulated.

The arithmetic and control unit 40 is a CPU (Central Proc
essing unit) and ROM (Read Only Memo) that stores system programs, application programs, etc.
ry) and RAM (Random Acc
ess Memory), timer for clocking, A / D converter, D /
It is configured by a microcomputer including an input / output interface such as an A converter.

As will be described later, the arithmetic control unit 40 analyzes the image data Da from the frame memory 34 to determine whether or not rockfall has occurred in the rockfall prediction area 20 (see FIG. 2). When it is determined that a rockfall has occurred, information indicating that, a code for identifying the rockfall monitoring device 12 (i) in which the rockfall has occurred, the image data Da ′ from the image compression unit 42, etc. Is supplied to the I / O control unit 44 as the notification data Db.

The I / O control unit 44 supplies the notification data Db from the arithmetic control unit 40 to the display processing device 16 via the communication network 14. Further, the I / O control unit 44 sends the image signal Sa from the CCD camera 22 to the image input unit 32.
For example, an instruction to take in a predetermined time (for example, 1 second)
It is output as the control signal Sf every time or according to the instruction from the arithmetic control unit 40.

When the display processing device 16 shown in FIG. 1 receives the notification data Db from the rockfall monitoring device 12 (i), it emits an alarm sound from a sounding unit (not shown) and displays the notification on the screen of a display (not shown). The image information of the rockfall prediction area 20 based on the image data Da ′ included in the data Db is displayed.

FIG. 4 shows a template T for individually recording information on a moving object including a falling stone C (see FIG. 2).
(J) (j = 0 to Nj-1) is shown. This template T (j) has a pos field, a size
It has a field, a point field, a cnt field, a flg field, and an image field.

The pos field is a field for recording the position of the moving object described later (upper left end coordinate of the circumscribing rectangle: pos obtained at the time of labeling described later).

The size field is a field for recording the size of the moving object (width and height of the circumscribing rectangle: size).

The point field is a field for recording the representative point coordinates (point) of the moving object.

The cnt field is a field for recording the counter value (cnt) corresponding to the time when the moving object is stationary.

The flg field is the template T
This is a field for recording a flag {flg (TRUE: valid, FALSE: invalid)} indicating the validity of (j).

In the image field, the coordinates of the representative point (p
It is a field for recording an original image (image) of a 16 × 16 square pixel area centering on (oint).

The template T (j) is set for each moving object and is stored in the memory 36 shown in FIG. As shown in FIG. 5, the templates T (j) are linked by a pointer p (j) (indicated by an arrow in FIG. 5) and are virtually arranged in one direction.
That is, the template T (j) has a unidirectional list structure. The number (j) of the template T (j) indicates the order of the templates T (j) arranged in one direction for convenience.

The templates T (j) are stored in the memory 36 in an arbitrary order (for example, the order in which they were created). That is, they are not arranged in the order of numbers (j). Therefore, the pointer p (j) indicates the next number (j +
By designating an address on the memory 36 in which the template T (j + 1) of 1) is stored, the templates T (j) are arranged in the order of numbers (j).

The final template T (Nj-1)
Terminal pointer (NULL) indicating that there is no more template T (j) in the pointer p (Nj-1) of
Is set.

In the memory 36, the template T
A management structure Ta for managing the list structure of (j) is stored. In this management structure Ta, a pointer p (head) to the first template T (0) and a pointer p (t to the last template T (Nj-1).
ail) is recorded. Therefore, this management structure T
A virtual array of the template T (j) can be formed by the pointer p (head) and the pointer p (tail) obtained from a.

"Rockfall Monitoring Process" Next, a process of the image transmission processing device 30 of FIG. 3, mainly for monitoring a rockfall in the arithmetic control unit 40, will be described with reference to the flowchart of FIG.

In this process, with the occurrence of falling rocks, the falling rock C invades the road surface 19 at a certain point,
After that, the presence or absence of rockfall is detected by utilizing the characteristic that all or part of it remains stationary on the road surface 19 (see FIG. 2).

First, in step S11, the arithmetic control unit 40 performs a process of initializing the template T (j). Specifically, the terminal symbol (NULL) is written to the pointer p (head) and the pointer p (tail) of the management structure Ta shown in FIG. As a result, the registered template T (j) does not exist.

In the following step S12, the arithmetic control unit 40 takes in the latest image data Da captured by the CCD camera 22 and stored in the frame memory 34. Then, this image data Da is used as a background image (BKG).
(BKG ← Da).

In the next step S13, the arithmetic control unit 40 takes in the latest image data Da newly captured by the CCD camera 22 and stored in the frame memory 34. Then, this image data Da is stored as a comparison target image (IMG) (IMG ← Da).

Following this, in step S14,
A process for detecting the presence or absence of a rock fall {a process for judging whether the determination result B is TRUE (fall rock is present) or FALSE (no rock fall)} is performed (the process of step S14 will be described later). .

Then, in the following step S15, a process for confirming the presence or absence of rockfall is performed. Specifically, it is confirmed whether the determination result B obtained in step S14 is B = TRUE or B = FALSE, and if B = TRUE, the process proceeds to the next step S16. ,on the other hand,
If B = FALSE, the process proceeds to step S17, which is the latter stage of step S16.

In step S16, the arithmetic control unit 40
Captures the image data Da ′ from the image compression unit 42, and determines the result B (information indicating that rockfall has occurred) and the rockfall monitoring device 12 (i) that rockfall has occurred.
Is output as notification data Db together with a code for identifying the.

This notification data Db is supplied to the display processing device 16 via the communication network 14, and accordingly, in the display processing device 16, the processing for issuing an alarm sound, the rockfall prediction area 20.
Display processing of the image information and the like is performed. Even if the occurrence of rockfall is not detected in step S14,
For example, the notification data Db may be supplied to the display processing device 16 every predetermined time.

In step S17, the arithmetic control unit 40
Performs a process of updating the background image (BKG). Specifically, the background image (BKG) is replaced with the comparison target image (IMG) obtained in step S13 and stored as a new background image (BKG) (BKG ← IMG).

After the process of step S17, the process returns to step S13.

"Rockfall Detection Processing" Next, the processing of step S14 (rockfall detection processing) will be described with reference to FIG.

In step S14, a rock fall determination process (step S21: FIG. 8), a moving object detection process (step S22: FIG. 9) and a template update process (step S23: FIG. 10) are sequentially performed.

"Rockfall determination process" First, the rockfall determination process of step S21 will be described with reference to FIG. The rockfall determination process is a process of determining whether or not the moving object corresponding to the already registered template T (j) is the falling stone C, and constitutes the determination means of the arithmetic control unit 40.

First, in step S31, the arithmetic control unit 40 performs a process of setting the determination result B to the initial value FALSE (no falling rock) (initialization process of the determination result B) (B ← FALSE).

Then, the determination process for each template T (j) is started. Specifically, first, step S3
2, the template with the smallest number (j) among the registered templates T (j) {that is,
A template designated by the pointer p (head) of the management structure Ta shown in FIG. 5} T (0) is selected and stored as a selected template (tmp) {tmp ← T.
(0)}. If the pointer p (head) is the terminal symbol (NULL), the rockfall determination process of step S21 ends.

In the next steps S33 and S34, the arithmetic control unit 40 selects the template (t) selected in step S32 or step S40 described later.
mp) is executed to determine whether or not the moving object registered (moving) is moving.

Specifically, first, in step S33, from the comparison target image (IMG) obtained in step S13 (see FIG. 6), the representative point coordinates (point) in the selection template (tmp) (step described later). The image of the area of 16 × 16 square pixels centering on (S73) is cut out, and this image and the selected template (tm
p) original image (image) (step S7 described later)
Obtained in 3. ) Are compared to obtain a normalized correlation coefficient (representing the similarity of images and taking a value of 0 to 1 depending on the degree of similarity). That is, the normalized correlation coefficient is obtained by comparing images included in the same coordinate range corresponding to a specific moving object in two images obtained at time intervals.

Then, in step S34, a process of comparing the value of the normalized correlation coefficient with a predetermined value (for example, 0.99) is performed. If the value of the normalized correlation coefficient is 0.99 or more, It is determined that the moving object is stationary (NO), and if the value of the normalized correlation coefficient is less than 0.99, it is determined that the moving object is moving (YES).

The processes of steps S33 and S34 will be described by taking images A1 to A3 (see FIGS. 11 to 13) as examples of images actually acquired by the CCD camera 22. Image A1 shown in FIG. The original image (im
If the image A2 shown in FIG. 12 is the comparison target image (IMG), the vehicle M in the image A1 is determined to be moving.

The image A2 shown in FIG. 12 is converted into the original image (i
image) and an image A3 shown in FIG.
Is the comparison target image (IMG), the vehicle M in the image A2 is determined to be moving, whereas the falling stone C and the cliff E are determined to be stationary.

The determination result in step S34 is YES.
If (moving object is moving), step S3
Go to 5. In step S35, the operation control unit 40 sets the flag (f in the selection template (tmp).
lg) is set to FALSE (invalid) (flg
← FALSE). Then, after step S35, the process proceeds to step S39. The initial value of the flag (flg) is TRUE (valid).

On the other hand, if the determination result in step S34 is NO (the moving object is stationary), the process proceeds to step S36. In step S36, the arithmetic control unit 40 performs a process of counting up the counter value (cnt) in the selection template (tmp) (that is,
(Time processing) is performed (cnt ← cnt + 1). The initial value of the counter value (cnt) is “0”.

In the following step S37, the arithmetic control unit 40 determines whether the counter value (cnt) has reached a predetermined judgment value (for example, the counter value Tmax corresponding to 30 seconds: Tmax) (that is, the moving object is 30 The process of determining whether or not it has been stationary for more than a second is performed.

In step S37, the counter value (cnt) has reached the judgment value (Tmax) (YE).
If it is determined to be S), in the next step S38, the determination result B is set to TRUE (falling rocks) (B ← TR
UE), and then proceeds to step S39. On the other hand, when it is determined that the counter value (cnt) has not reached the determination value (Tmax) (NO), the process of step S38 is not performed and the process proceeds to step S39.

That is, if at least one of the moving objects registered in the template T (j) has a stationary time of 30 seconds or more, the determination result B is TRUE (falling rocks), and Output processing of the notification data Db in step S16 (see FIG. 6) is performed.

In step S39, the arithmetic control unit 40
Determines whether or not to end the process of step S21 (rockfall determination process). Specifically, the template T (j) set as the selection template (tmp)
Is the last template T (Nj-1) {that is, a terminal symbol (NULL) is set to the pointer p (j).
Is set, and the pointer p
If (j) is the terminal symbol (NULL) (YES), the process of step S21 ends. On the other hand, the number (j) of the selection template (tmp) is the terminal symbol (NUL
When it is not L) (NO), the process proceeds to the next step S40.

In step S40, the template having the next largest number (j) after the template T (j) set as the selected template (tmp) up to that point (ie, the template pointed to by the pointer p (j)}. Select T (j + 1) and store it as a new selection template (tmp) {tmp ← T (j +
1)}. At this time, the original selection template (tm
The storage process of p) {writing process to the linked template T (j)} is also performed. And this step S4
When the process of 0 is completed, the process returns to the process of step S33, and the process for the new selection template (tmp) is performed.

[Moving Object Detection Processing] Next, the moving object detection processing of step S22 in FIG. 7 will be described with reference to FIG. This moving object detection processing is processing for detecting a new moving object and registering it in the template T (j), and constitutes the image analysis means of the arithmetic control unit 40.

First, in step S51, the arithmetic and control unit 40 performs the above step S12 (or step S1).
7) and the difference between the background image (BKG) and the comparison target image (IMG) obtained in step S13 (see FIG. 6) respectively (that is, the difference between the two images obtained at a predetermined time interval). The value is obtained as a difference image (DIF) (DIF ← | IMG-BKG |). In the difference image (DIF) thus obtained, there are high-luminance block regions at the current position and the past position of the moving object.

In the next step S52, the arithmetic control unit 40 selects one of the registered templates T (j).
The template with the smallest number (j) {that is, FIG.
The template designated by the pointer p (head) of the management structure Ta shown in (1)} T (0) is selected and stored as the selected template (tmp) {tmp ← T (0)}.
The pointer p (head) is a terminal symbol (NULL).
If so, the processes of the next steps S53 to S55 are not performed.

In the following step S53, an area (movement) on the difference image (DIF) defined by the position (pos) and the size (size) in the selection template (tmp) selected in the step S52 or step S55 described later. The past position of the object) is specified, and the value of the difference image (DIF) included in this area is set to a zero value (0). In this case, in the difference image (DIF), a high-intensity blocky region exists only at the current position of the moving object.

The difference image (DI
F), images A1 to A4 as examples of images actually acquired by the CCD camera 22 (see FIGS. 11 to 14).
It will be described based on.

The image A1 shown in FIG. 11 is converted into the background image (BK
G) and the image A2 shown in FIG. 12 is the comparison target image (IM
G), the difference image (DI
In image F4, as shown in image A4 in FIG. 14, vehicle M, falling stones C and cliff E in image A2 in FIG. 12 are highlighted as a lump. The vehicle M in the image A1 in FIG. 11 has been deleted in step S53.

On the other hand, when the image A2 shown in FIG. 12 is the background image (BKG) and the image A3 shown in FIG. 13 is the comparison target image (IMG), the difference image (DIF) is emphasized as a block. Does not exist. That is,
In the image A3 of FIG. 13, there is no object extracted as a new moving object.

Next, in step S54, the arithmetic control unit 40 sets the pointer p of the selection template (tmp).
It is determined whether (j) is the terminal symbol (NULL) (YES) or not the terminal symbol (NULL), and if the determination result is YES, step S5.
The process moves to 6 and the process of extracting the moving object is started. On the other hand, if the determination result is NO, the next step S
Move to 55.

In step S55, the template having the next largest number (j) after the template T (j) set as the selected template (tmp) up to that point (ie, the template pointed to by the pointer p (j)}. Select T (j + 1) and set it as a new selection template (tmp) {tmp ← T (j +
1)}. After the process of step S55, the process returns to the process of step S53.

In step S56, the arithmetic control unit 40
Performs a process of binarizing the difference image (DIF).

Further, in step S57, labeling processing is performed on the binarized difference image (DIF). That is, a circumscribed rectangle corresponding to each moving object in the binarized difference image (DIF) is obtained, and the circumscribed rectangle is labeled with a label number k (k = 1 to Nk). By this process, moving objects (for example, the vehicle M, each falling stone C and the cliff E in the image A4 of FIG. 14) are extracted as one lump.

Subsequently, the arithmetic and control unit 40 carries out a process for judging validity / invalidity of each moving object. First, step S5
In step 8, the label number k is set to 1 (k ← 1).

Next, in step S59, the moving object with the label number k set in step S58 or step S61 described later is evaluated based on its area. Specifically, the area S of the lump (circumscribing rectangle) extracted as the moving object in step S57 is calculated, and this area S is larger than a predetermined reference area Sth (S
≧ Sth) is determined. The reference area Sth is a value determined depending on the installation conditions of the CCD camera 22 and the like, for example, Sth = 50 square pixels.

If it is determined in step S59 that the area S of the block is equal to or larger than the reference area Sth (YES: valid), the moving object corresponding to this block is template T (j) in step S60. Perform the process to register to.

Specifically, the position (pos) and size (size) of the circumscribed rectangle obtained in step S57.
Are respectively written in the pos field and the size field to create a template T (j), and this template T (j) is connected to the end of the array of the already existing template T (j).

That is, the existing end template T
(Nj-1) {pointer p of the management structure Ta shown in FIG.
Template T (j) specified by (tail)}
Of the newly created template T (Nj) is set in the pointer p (Nj−1) of the template T (Nj−1), and a terminal symbol (NULL) is set in the pointer p (Nj) of the template T (Nj). The template T is used as the pointer p (tail) of the management structure Ta.
Register the address of (Nj).

After this step S60, step S61
Move to. When it is determined in step S59 that the area S of the lump is less than the reference area Sth (NO: invalid), the process of step S60 is not performed and the process proceeds to step S61. In this case, the moving object is processed as a noise image.

In step S61, the arithmetic control unit 40
Performs a process of adding 1 to the label number k (k ← k +
1). Then, in a succeeding step S62, it is determined whether or not there is a moving object corresponding to the newly set label number k. When there is a moving object (YES)
In step S59, the process returns to step S59. On the other hand, if no moving object exists (NO), the process in step S22 (moving object detection process) ends.

"Template updating process" Next, in FIG.
Regarding the template updating process in step S23, FIG.
A description will be given with reference to 0. This template updating process is a process of updating or deleting the registered template T (j).

First, in step S71, the arithmetic control unit 40 determines the template with the smallest number (j) among the registered templates T (j) {that is,
The template designated by the pointer p (head) of the management structure Ta shown in FIG. 5} T (0) is selected and stored as the selected template (tmp) {tmp ← T.
(0)}. If the pointer p (head) is the terminal symbol (NULL), the template updating process of step S23 is terminated.

In the following step S72, the flag (flg) (FIG. 8) in the selection template (tmp) selected in the step S71 or step S76 described later.
It is set in step S35. ) Is TRUE
It is judged (confirmed) whether it is (valid) or FALSE (invalid).

In step S72, the flag (f
If it is determined that lg) is TRUE, the process proceeds to step S73.

In step S73, the arithmetic control unit 40
Performs the process of registering the image of the moving object as the original image (image) in the selection template (tmp). That is, the original image (im
age) registration process or existing template T
The original image (image) is updated for (j).

Specifically, as shown in FIG. 15, first,
The comparison target image (IMG) and background image (BK) obtained in steps S13 and S17 (see FIG. 6), respectively.
From G), the position (po) in the selection template (tmp)
s) and the size (size) (registered in step S60), the range W of the circumscribed rectangle of the moving object (falling stone C, vehicle M, etc.) is specified. Then, from this range W, an image w of the same area w of 16 × 16 square pixels
Cut out a and wb respectively.

In this case, the images wa, w in all the regions w within the range W are shifted while shifting the region w of 16 × 16 square pixels upward or downward and left or right in units of one pixel.
Try to cut out b. For example, the size of the range W is 9
In the case of 0 × 40 pixels, the number of areas w is 1875.
[= {90- (16-1)} × {40- (16-
1)}].

Next, an image w cut out for each area w
a and wb are compared with each other to obtain a normalized correlation coefficient, and all the obtained normalized correlation coefficients are compared with each other to obtain a region (image wa, wb) where the normalized correlation coefficient is the minimum.
The region w in which the inconsistency of is most remarkable is extracted. Then, the center coordinates of this area w are the representative point coordinates (point)
As the point field. If the height or width of the range W of the circumscribing rectangle is less than 16 pixels, the position of the center of gravity of the range W is used as the representative point coordinates (point).
And

In this case, the normalized correlation coefficient is the image wa,
It is obtained based on the distribution of wb shading. further,
The area w having the smallest normalized correlation coefficient is mainly an area (for example, an area w ′ in FIG. 15) extending over the boundary between the moving object and the background. Therefore, it is possible to avoid erroneously determining that images having similar brightness values (for example, an image of a roof portion of a vehicle and an image of a road surface) are the same image.

In step S73, the representative point coordinates (point) is not updated for the existing template {template in which the representative point coordinates (point) are already recorded} T (j).

The arithmetic control unit 40 determines the representative point coordinates (po
Following the setting of (int), the registration process of the original image (image) is performed. Specifically, an image of a region of 16 × 16 square pixels centered on the representative point coordinates (point) written in the point field is cut out from the comparison target image (IMG), and this is used as an original image (image). Write in the image field.

In step S72, the flag (f
If it is determined that lg) is FALSE, the process proceeds to step S74. In this step S74,
The arithmetic control unit 40 performs a process of deleting the selected template (tmp) (actually, the template T (j) corresponding to this). That is, in the example shown in FIG.
The template T (j) corresponding to the vehicle M in No. 4 is deleted.

In this step S74, as the pointer p (j-1) in the template T (j-1) immediately before the template T (j) to be deleted, the pointer p (j-1) after the template T (j) is deleted. Template T (j
+1) address is set. By this processing, the template T (j) is removed from the array structure.

Note that the template T to be deleted is
When (j) is the last template T (Nj-1), the terminal symbol (NULL) is set to the pointer p (Nj-2) in the immediately preceding template T (Nj-2). Then, as a pointer p (tail) of the management structure Ta shown in FIG. 5, this template T (Nj-
2) Set the address. When the template T (j) to be deleted is the first template T (0), the pointer p (head) of the management structure Ta
As a result, the address of the template T (1) immediately after that is set.

After the process of step S73 or S74 is performed, the process proceeds to step S75. This step S7
5, the arithmetic control unit 40 selects the selection template (t
It is determined whether the pointer p (j) of mp) is a terminal symbol (NULL) (YES) or not a terminal symbol (NULL) (NO), and if the determination result is YES, step The process of S23 (moving object detection process) is terminated, and when the determination result is NO, the process proceeds to the next step S76.

In step S76, the template having the next largest number (j) after the template T (j) set as the selected template (tmp) up to that point (ie, the template pointed to by the pointer p (j)}. Select T (j + 1) and set it as a new selection template (tmp) {tmp ← T (j +
1)}. At this time, the original selection template (tm
The storage process of p) {writing process to the linked template T (j)} is also performed. And this step S7
After the process of 6, the process returns to the process of step S72.

As described above, in the rockfall monitoring system 10 according to the embodiment of the present invention, the rockfall prediction area 20 is provided.
The presence or absence of rockfall can be monitored on the basis of the image of the rockfall prediction area 20 acquired by the CCD camera 22 installed at a position overlooking. That is, at a place near the rockfall prediction area 20 at the monitoring site, a CCD
Since only the camera 22 needs to be installed, the installation cost can be reduced and the installation work can be facilitated.

In this case, since the CCD camera 22 is not in contact with the expected rockfall region 20, it is not damaged by the rockfall. Therefore, it can be continuously used even after the occurrence of rockfall. That is, it is possible to realize a long-term and even semi-permanent device life.

When the rockfall monitoring device 12 (i) detects the occurrence of rockfall, the information is used for the rockfall prediction area 2
The image information of 0 is supplied to the display processing device 16 installed in the surveillance office. Therefore, the surveillance staff stationed at the surveillance office can quickly perform disaster response work.

Furthermore, the observer can accurately determine the occurrence of rockfall by checking the image information of the predicted rockfall area 20 displayed on the screen of the display processing device 16. Therefore, erroneous detection of a parked vehicle or the like on the road surface 19 as the falling stone C can be avoided.

Further, since it is possible to grasp the specific situation of the monitoring site remotely and promptly, it is possible to reduce the number of patrols, and it is also possible to eliminate the need for patrols.

[0111]

According to the present invention, it is possible to obtain a rock fall monitoring system that can be installed easily and at low cost, has excellent durability, and can remotely monitor a monitoring site.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a rockfall monitoring system according to an embodiment of the present invention.

FIG. 2 is a diagram showing an example of a monitoring site where a rockfall monitoring device constituting the rockfall monitoring system of FIG. 1 is installed.

FIG. 3 is a block diagram showing an image transmission processing device constituting the rockfall monitoring system of FIG. 1.

FIG. 4 is a diagram showing a template in which information of a moving object is recorded.

5 is a diagram showing an array structure of the template of FIG.

6 is a flowchart showing a process mainly performed by the arithmetic and control unit for monitoring a rock fall in the image transmission processing device of FIG. 3;

FIG. 7 is a flowchart showing a rock fall detection process in the flowchart of FIG.

FIG. 8 is a flowchart showing a rock fall determination process in the flowchart of FIG.

9 is a flowchart showing a moving object detection process in the flowchart of FIG.

FIG. 10 is a flowchart showing template update processing in the flowchart of FIG.

FIG. 11 is a diagram showing an example of an image actually acquired by a CCD camera.

FIG. 12 is a diagram showing an example of an image actually acquired by a CCD camera.

FIG. 13 is a diagram showing an example of an image actually acquired by a CCD camera.

14 is a diagram showing an example of a difference image obtained as a difference between the image of FIG. 11 and the image of FIG.

FIG. 15 is an explanatory diagram showing processing for obtaining representative point coordinates of a moving object.

[Explanation of symbols]

10 ... Rockfall monitoring system 12 (i) ... Rockfall monitoring device 14 ... Communication network 16 ... Display processing device 20 ... Expected rockfall region 22 ... CCD camera 30 ... Image transmission processing device 32 ... Image input unit 34 ... Frame memory 36 ... Memory 40 Calculation control unit 42 Image compression unit 44 I / O control unit

─────────────────────────────────────────────────── --- Continuation of the front page (56) References JP-A-3-211700 (JP, A) JP-A-6-231252 (JP, A) JP-A-8-241480 (JP, A) JP-A-8- 77487 (JP, A) JP 56-157862 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) G01D 21/00 G01B 11/00 G06T 7/00 H04N 7/18

Claims (5)

(57) [Claims] 1. An image acquisition means for acquiring an image of an area where rockfall is to be monitored, an image analysis means for analyzing an image acquired by the image acquisition means to extract a moving object, and an image analysis means for the image analysis means. extracted moving object have a determination means for determining whether a falling stone, the image analysis means, two obtained at a time by the image acquisition unit
A moving object from an image by extracting, from the extracted moving object region, the small composed of a predetermined number of pixels
Cut out the image of the area and span the entire area of the moving object
The correlation with the previously acquired image of the two images
And a small area having the smallest correlation is specified, and the determination means determines the specified area in the new images acquired at a further time interval.
Image of the small area at the same position as the
Correlate with the image of the specified small area of the moving body,
If the correlation is above a certain level, the moving object is a falling stone.
A rockfall monitoring system characterized by determining that there is. 2. A rockfall monitoring system of claim 1, wherein said image analysis means, before SL obtains the difference image from the two images, further the moving object by the difference image is binarized, labeling A rockfall monitoring system characterized by extraction. 3. The rockfall monitoring system according to claim 1 or 2, wherein the determination means determines whether the moving object is a falling stone based on whether the moving object has stopped for a predetermined time or more. A rockfall monitoring system characterized by: 4. The rockfall monitoring system according to any one of claims 1 to 3 , further comprising display processing means for displaying the determination result of the determination means, the image acquired by the image acquisition means, and the like. A rockfall monitoring system characterized by. 5. The rockfall monitoring system according to claim 4 , wherein the image acquisition means, the image analysis means, and the determination means are installed at a plurality of monitoring sites and are respectively connected to the display processing means via a communication network. A rockfall monitoring system characterized by being connected.