JP2000105137A - Method and apparatus for detection of avalanche of rocks and earth - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method and an apparatus in which the generation of an avalanche of rocks and earth is detected by using a camera, in a noncontact manner, so as to be resistant to a wind and to a change in sunshine and simply. SOLUTION: By an A/D converter 2, an image signal which is output by a camera 1 used to photograph a region to be monitored is fetched at set intervals so as to be converted into a digital signal. An image memory 3 stores the digital signal. A computing circuit 4 computes the difference in an image between data, in the image memory 3, stored before and data, in the image memory 3, stored newly, and difference data is stored in a difference image memory 5. A number-of-changed-pixels computing means 6 binarizes the difference data by using a determined threshold value, and it finds the area of a changed region as the number of changed pixels. A moving- average computing means 7 finds two kinds of a short-term moving average and a long-term moving average on the basis of time-series data on the obtained number of changed pixels. A judgment means 8 judges the generation of an avalanche of rocks and earth when the increase or the decrease ratio of the short-term moving average to the long-term moving average is at a set ratio or higher and continued in a set cycle number or higher.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for detecting occurrence of debris flow from a camera image of a debris flow. In particular, it processes video signals from ITV cameras,
The present invention relates to a method and an apparatus for performing generation of a debris flow, determination of the scale of the generated debris flow, and prediction of occurrence of a debris flow from a moving average of periods in which the amount of change in a designated monitoring area differs.

[0002]

2. Description of the Related Art It is important to detect or predict the occurrence of debris flow quickly and reliably in order to prevent a debris flow disaster in the construction of a sabo dam, a work for removing sediment, and the like. Furthermore, it is desired to be able to cope with secondary and tertiary debris flow disaster prevention in rescue work after a disaster occurs.

There is a wire sensor method for detecting debris flow, in which a wire sensor is installed at a debris flow passage point and the debris flow cuts the wire to detect the debris flow.
There is also a system using an acoustic sensor or a vibration sensor for detecting from sound or vibration when a debris flow occurs.

[0004]

However, conventional non-image sensors such as a wire sensor, an acoustic sensor, and a vibration sensor alone cannot confirm the occurrence of debris flow when the sensor is activated. On the other hand, in the image method, the occurrence of debris flow can be visually checked by transmitting an image when the sensor is operating.

[0005] The wire sensor method has a risk of malfunction due to animals or falling rocks. Once the wire is cut, the next debris flow cannot be detected unless the wire is re-stretched. Therefore, there are many restrictions such as topography.

Also, vibration sensors and acoustic sensors have malfunctions due to noise from roads and heavy equipment, and it is difficult to set a trigger level due to attenuation characteristics due to distance from a debris flow as a sound source and ground characteristics.

[0007] The 24-hour continuous monitoring by the camera alone imposes severe labor on the observer and requires labor costs. Further, in the image processing method, it is conceivable to detect a debris flow from a moving area and a moving speed by obtaining a plurality of moving vectors, but high-speed processing is required and it is expensive. As a simple method, there is also a background subtraction method in which a background image is compared with a reference image. However, the background image is easily affected by changes in sunshine, and updating of the background image is difficult.

The present invention has been made in view of such circumstances, and it is an object of the present invention to provide a simple method and apparatus which uses a camera so as to be non-contact, is resistant to wind and sunshine changes, and is simple.

[0009]

According to a first aspect of the present invention, there is provided a method for detecting a debris flow, wherein a video signal is taken from a camera for photographing an area to be monitored at regular intervals, and the video signal is converted into a digital signal. The difference between the images in the designated monitoring region of each of the stored video signals is calculated, and the occurrence of debris flow is predicted or detected from the time-series change. In the present invention, debris flow detection is also defined together with prediction or detection of debris flow occurrence.

Further, in the debris flow detection method of claim 2, in addition to the debris flow detection method of claim 1, the scale of the generated debris flow is divided into a plurality of levels and output.

According to a third aspect of the present invention, there is provided a debris flow detection device, a camera for photographing a monitored area, an A / D converter that takes in a video signal output from the camera at regular intervals and converts it into a digital signal, and an A / D converter. Memory for storing digital signals output from the imager, an arithmetic circuit for calculating an inter-image difference amount from image data of a monitoring designated area of each video signal stored in the image memory, and a debris flow based on a time-series change in the inter-image difference amount. Means for predicting or detecting the occurrence of the occurrence.

Further, the debris flow detecting device according to a fourth aspect of the present invention includes, in addition to the debris flow detecting device according to the third aspect, a visibility determining means for determining the field of view of the monitored area and making it impossible to perform processing when the visibility is poor.

Further, the debris flow detecting device according to a fifth aspect of the present invention includes, in addition to the debris flow detecting devices according to the third and fourth aspects, a communication means for transmitting an alarm signal and a detected image when the debris flow is detected.

Further, the above-mentioned method and apparatus are provided with means capable of storing images at regular intervals during a designated time after the occurrence of debris is detected immediately before the detection of occurrence of debris. In addition, an input cutout of another sensor such as a wire sensor or a vibration sensor may be provided, and the detection of debris may be performed in combination.

In such a debris flow detection method and debris flow detection device, since a camera is used, it is non-contact, resistant to wind and sunshine changes, and can be provided at a low cost by a simple method. Civil works in mountain streams are generally performed during the daytime, but lighting is used during nighttime work. A (near) infrared projector and a (near) infrared camera may be used to avoid influence on the neighborhood. At the time of debris flow detection, the image is transmitted to enable visual confirmation.

[0016]

FIG. 1 shows a processing flow of an embodiment of a debris flow detection method according to the present invention. First, initialization processing such as initialization of variables and setting of a monitoring mask area which is a monitoring designation area is performed (S1). An initial image is input from the ITV camera (S2). At the start of monitoring, the video signal of the initial image in S2 is A / D converted and loaded into the image memory (S3). In the first cycle, an inter-image difference is obtained between the initial input image and the following cycle in the subsequent cycle (S4).

When a plurality of image differences are obtained during one cycle, the difference images are added (S5). Here, it is unnecessary when only one monitoring image is captured during one cycle. Also, how many images are taken in one cycle can be parameterized and set. By capturing a plurality of images during one cycle and accumulating the difference amount, a normal water flow region or a debris flow region can be more easily captured.

It is checked whether the number of surveillance image fetches is the designated number (S6). Here, when the contrast is poor, the designated addend is increased. The difference image accumulated a specified number of times is binarized (S7). The difference amount is calculated as a binarized area (S
8).

Two types of moving averages are obtained from the difference obtained in each cycle (S9). Here, in calculating the two types of moving averages, a long-term moving average with a large number of times going back
A short-term moving average with a small number of times going back to the past. For example, a long-term moving average is calculated 20 times, and a short-term moving average is calculated 7 times retroactively.

FIG. 2 shows an example of a graph in which a long-term moving average and a short-term moving average are calculated from the number of changed pixels.
The long-term moving average can slowly follow the water flow, but the tracking of the debris flow that changes stepwise is delayed. The long-term moving average represents a steady-state water flow region. On the other hand, the short-term moving average can sufficiently cope with a sudden change.

An increase / decrease ratio of the difference is calculated (S10).
The increase / decrease ratio can be expressed by an increase / decrease ratio = short-term moving average / long-term moving average × 100 (%). Next, the occurrence of debris flow is determined from the increase / decrease ratio of the difference amount (S11).

(Debris flow generation judgment) The judgment method is that the increase / decrease ratio is R1 (detection judgment ratio threshold value) within the past N1 cycles.
If the above cycle is performed M1 times, it is determined that debris flow has occurred. For example, the determination threshold value is set such that N1 = 20 cycles, R1 = 120%, and M1 = 15 cycles.

(Determination of Debris Flow Size) When a debris flow is occurring, an alarm is output by classifying the scale of debris flow according to the increase / decrease ratio. An example is shown below. Increase / decrease ratio: 120 to 199% = alarm level 1 (or small): 200 to 399% = alarm level 2 (or medium): 400% or more = alarm level 3 (or large)

(Judgment of occurrence of debris flow notice)
If there are M2 cycles in which the increase / decrease ratio is less than or equal to R2 (the prediction determination ratio threshold) within two cycles, and there is visibility around the monitoring area, it is determined that the debris flow has occurred. For example, N2 = 20 cycles, R2 = 80%, M2 = 1
The determination threshold is set as in five cycles. The presence of the field of view is determined as the presence of the field of view if the edge component is obtained from the differential image in the field of view determination area provided around the debris flow monitoring area.

(Determination of Visibility Defect Measurement Unavailable) The determination method is that the cycle in which the increase / decrease ratio is less than R2 within the past N2 cycles is M
If there are two times and the visibility around the monitoring area is poor, it is determined that the visibility failure measurement cannot be performed. For example, the determination threshold is set such that N2 = 20 cycles, R2 = 80%, and M2 = 15 cycles. In the case of poor visibility, an edge component is obtained from a differential image in a visibility determination area provided around the debris flow monitoring area, and is compared with a threshold. As an example of the night view determination means, a high brightness point such as lighting is monitored, and when the brightness becomes lower than a certain level, it is determined that the view is poor.

Next, FIG. 3 shows a block diagram of one embodiment of the debris flow detecting device of the present invention. I photographed the monitored area
An image signal from the TV camera 1 is converted into a digital signal by the A / D converter 2. The digital signal is stored in the image memory 3. The arithmetic circuit 4 calculates an image difference between the previously stored data in the image memory 3 and the newly stored data in the image memory 3, and stores the difference data in the difference image memory 5.

The above operation is repeated for the number of times designated by the parameter during one cycle operation, and the difference value is stored in the difference image memory 5.
Accumulates. The change pixel number calculation means 6 binarizes the difference value stored in the difference image memory 5 with a predetermined threshold value and obtains the area of the change region as the change pixel number. The moving average calculation means 7 calculates two types of short-term and long-term moving averages from the obtained time-series data of the number of changed pixels.

The judging means 8 judges that debris flow has occurred when the increase / decrease ratio of the short-term moving average to the long-term moving average continues over a certain ratio and over a certain number of cycles. The increase / decrease ratio and the like are the values described in the debris flow detection method, but are stored in the increase / decrease ratio storage unit 9 in advance. The judging means 8 judges the debris flow scale at the same time based on the degree of the increase / decrease ratio, and outputs a detection alarm to the alarm display means 10.

If the increase / decrease ratio of the short-term moving average to the long-term moving average continues at a certain ratio or less and continues for a certain cycle or more, and when the visibility is further secured, it is determined that a debris flow may occur (notice), and the alarm display means 10 And output an advance warning. When the increase / decrease ratio of the short-term moving average to the long-term moving average continues below a certain ratio and over a certain cycle,
Further, when there is no view and the view determination means 12 determines that the view is poor, the view determination warning is output to the warning display means 10. The view determination unit 12 generates a differential image from the view determination area of the video signal stored in the image memory 3. Further, the visibility determining means 12 compares the value of the edge component of the differential image with a predetermined threshold value, and determines that the visibility is poor when the value of the edge component of the differential image is equal to or less than the threshold value.

In a system in which a moving image cannot always be transmitted, when the detection alarm or the advance alarm is output by the communication means 11, the image information is transmitted together with the setting.

Further, a part of the image memory 3 is constituted by a hard disk (HDD), and images can be stored at regular intervals from a time immediately before the detection of the occurrence of debris to a specified time after the occurrence is detected. The saved video signal can be used for subsequent debris flow analysis and can be used for debris flow research. In addition, it has an input cutout of another sensor such as a wire sensor or vibration sensor,
Debris detection can also be performed in combination. Thereby, it is used as a backup device at the time of camera failure or poor visibility, and the reliability of the system can be improved.

[0032]

As described above, the debris flow detection method and debris flow detection device of the present invention have the following effects. A simple algorithm can reliably detect debris flow without being affected by changes in sunlight. Since the relative ratio is used instead of the absolute value of the change amount, the setting is easy. The scale of the generated debris flow can be divided into levels. Visual detection of the detected debris flow is possible. When visibility is poor, low visibility warning is possible. The monitoring work load of monitoring workers can be reduced.

[Brief description of the drawings]

FIG. 1 is a processing flow showing an embodiment of a debris flow detection method according to the present invention;

FIG. 2 is a graph showing a moving average of the number of changed pixels.

FIG. 3 is a block diagram showing an embodiment of a debris flow detection device according to the present invention.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... ITV camera, 2 ... A / D converter, 3 ... Image memory, 4 ... Operation circuit, 5 ... Difference image memory, 6 ... Changed pixel number calculation means, 7 ... Moving average calculation means, 8 ... Determination means, 9
... memory for increase / decrease ratio, 10 ... alarm display means, 11 ... communication means, 12 ... visibility determination means.

Claims (5)

[Claims] An image signal is captured from a camera for photographing an area to be monitored at regular intervals, the image signal is converted into a digital signal and stored, and a difference between images of the stored image signal in a monitoring designated area is calculated. A debris flow detection method for predicting or detecting the occurrence of debris flow from the time series change. 2. The debris flow detection method according to claim 1, wherein the scale of the generated debris flow is divided into a plurality of levels and output. 3. A camera (1) for photographing an area to be monitored, an A / D converter (2) for taking in a video signal output from the camera at regular intervals and changing it into a digital signal, and the A / D converter comprises: Image memory for storing digital signals to be output (3)
An arithmetic circuit (4) for calculating an inter-image difference amount from the image data of the monitoring designated area of each video signal stored in the image memory; and predicting or detecting occurrence of debris flow from a time-series change in the inter-image difference amount (5, 6, 7, 8, 9). 4. The debris flow detecting device according to claim 3, further comprising a visibility determining means (12) for determining the visibility of the monitored area, wherein processing is disabled when visibility is poor. 5. The debris flow detection device according to claim 3, further comprising a communication unit (11) for transmitting an alarm signal and a detected image when the debris flow is detected.