JP2001067573A - Method for movement detection by image processing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method which can accurately detect important movement at a monitored site such as a steep slope through image processing while removing the influence of noise generated by a flying bird, a small moving animal, etc. SOLUTION: A sequential image signal obtained by picking up an image of the monitored site S by a video camera 1 is sent to an image processor 4 to find the mean of density relative differences showing differences in a measured area A and the mean of the density relative differences is compared with a limit value to decide that there is movement when the limit value is exceeded. This limit value is automatically found by multiplying a correction coefficient k (k>1) by the mean value obtained by dividing the sum of N mean values of density relative differences found successively by N. Further, After detecting the mean of the density relative differences exceeding the limit value, the mean of the density relative differences is compared with the limit value over predetermined times and it is decided that there is movement when the limit value is exceeded at larger than a specific frequency.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for processing an image obtained by an image pickup apparatus and detecting a motion in the image, and in particular, relates to landslides, rock-falls, rock-falls, falls, rock-slides, and the like. In a remote monitoring system that remotely monitors dangerous slopes such as avalanches, high waves surpassing breakwaters, and monitors intrusions into buildings and premises, abnormalities occur by processing images taken of sites to be monitored. The present invention relates to a method for detecting a dynamic state.

[0002]

2. Description of the Related Art In Japan, since about 70% of the country is mountainous, there are tens of thousands of dangerous places such as landslides, rockslides, rockfalls, falls, and rockslides. In most cases, such a method of monitoring a dangerous part is visually confirmed by a patrol that is inexpensive.

[0003] In addition, in order to ensure the safety of road traffic along the coastline, it is necessary to remotely monitor high waves passing over breakwaters, monitor traffic congestion conditions, and monitor intrusions into buildings and premises. Remote monitoring systems have been developed.
In such a monitoring system, an image signal obtained by imaging a site to be monitored by one or a plurality of video cameras is displayed on one or a plurality of monitors, and this display screen is monitored by an attendant. It is common to do. In addition, recording is performed by a video recorder at the same time as or instead of being displayed on a monitor. In addition, for monitoring intrusion into buildings and premises at night, an intrusion monitoring device using an infrared sensor is also used, so that when intrusion is detected, the lighting device is turned on and an image is taken with a video camera. A system has also been proposed.

[0004]

When monitoring dangerous slopes such as landslides, rockslides, rockfalls, falls, rockslides, etc. in normal patrol work, visual observations are mainly used. (For example, to confirm the scattering of pebbles). Also, if the patrol staff changes,
Judgment standards vary, and there is a problem in ensuring accuracy.

Further, in a system for monitoring while monitoring a screen captured by a video camera, it is necessary for a clerk to constantly monitor, and it is not easy to detect any abnormality from the displayed screen. It is very difficult to meet the needs of time monitoring and simultaneous monitoring of multiple locations. Therefore, this monitoring method may miss a significant change. In this case, an image processing system may be considered in which images taken sequentially are compared, and if there is a difference between them, it is determined that some movement has occurred and an abnormality is displayed. However, in this case, if the detection sensitivity is increased, even a slight change that can be regarded as noise is determined to be abnormal. Conversely, if the sensitivity is reduced, the change that should be detected is overlooked. There is.

An object of the present invention is to solve the above-mentioned problems and to provide a method capable of accurately detecting a movement in a site to be monitored by image processing.

[0007]

According to the present invention, there is provided a dynamic detection method based on image processing, wherein a site where motion is to be detected is imaged at a site relative to a density relative to a difference between two image signals obtained by imaging the site back and forth in time. Calculating a difference, a step of calculating an average of the concentration relative differences, and a step of comparing the average of the concentration relative differences with a limit value and determining that there is a motion when the difference exceeds the limit value. Features

In a preferred embodiment of the motion detection method by image processing according to the present invention, the average of the density relative differences is calculated by:
The sum of the density relative differences is obtained by dividing by the total number of pixels in the measurement range. Preferably, the limit value is automatically obtained based on an average of the density relative differences. For example, the correction coefficient k is obtained by dividing the limit value L by an average value obtained by dividing the total sum of the average M (i) of N density relative differences obtained in succession by N.
(K> 1).

Further, in the motion detection method based on image processing of the present invention, it may be determined that there is a motion when the average of the relative density differences exceeds the limit value even once. Many instantaneous movements exist outdoors due to various natural phenomena, and detecting these as abnormal movements is not preferable in many applications. Therefore, in a preferred embodiment according to the present invention, a trigger signal is generated when the average of the above-mentioned concentration relative difference exceeds a limit value, and when this trigger signal is generated, the trigger signal is generated for a predetermined number of times. Compare the average of the relative concentration differences with the limit value,
When the number of times exceeding the limit value is equal to or more than a predetermined number, it is determined that there is a motion.

Furthermore, in a preferred embodiment of the motion detecting method according to the present invention, a plurality of measurement areas are set in a site where motion is to be detected, and an average of relative density differences is determined for each of the plurality of measurement areas. Each of the measurement areas is compared with the limit value, and when the average of the relative density differences in any one of these measurement areas exceeds the limit value, it is determined that there is a motion, or the density is determined in a plurality of predetermined measurement areas. When the average of the relative differences exceeds the limit value, it is determined that there is a movement. In such a method, a plurality of measurement areas can be set at optimal positions according to the situation of the site, so that the accuracy of motion detection can be further improved.

[0011]

FIG. 1 shows the entire configuration of a debris flow monitoring system to which a motion detection method based on image processing according to the present invention is applied. The image of the site S where the debris flow is to be monitored is captured by the video camera 1 having the solid-state imaging device, and the obtained image signal is transmitted to the monitoring office 2. In this example,
The image signal is transmitted via a cable, but may be transmitted wirelessly. The monitoring office 2 is provided with a monitoring monitor 3 that projects an image signal transmitted from the monitoring site S. The attendant can monitor while watching the image projected on the monitor 3. Therefore, it is preferable that the video camera 1 is a color video camera and the monitoring monitor 3 is a color monitor.

An image signal sent from the video camera 1 is also supplied to an image processing device 4. This image processing device 4
Includes a signal processing unit 5 for performing processing such as Y / C separation and time-base correction on a color image signal, an arithmetic processing unit 6 for performing motion detection according to the present invention, and an image signal and a detection result. A storage unit 7 for storing, and an image processing display unit 8
Is provided. Further, a motion detection signal output when it is determined that there is a motion in the image processing device 4 is supplied to a controller 9, which supplies an alarm signal to alarm devices 10 to 12 installed at a predetermined place. And give an alarm.

In the debris flow monitoring system as described above, a staff member monitors the monitoring site S displayed on the monitoring monitor 3.
While monitoring the occurrence of debris flow while looking at the image of, the image processing device 4 performs image processing to automatically detect movement, and when an abnormal movement is detected, the controller 9.
Supply an abnormal signal to the monitoring site S
False warnings often occur when birds fly, small animals move, or trees sway due to the wind. If the sensitivity of motion detection is reduced to eliminate such a problem, the possibility that important motions are overlooked increases, and in any case, the reliability of the system is impaired. The motion detection method according to the present invention solves such a problem.

In the present embodiment, instead of detecting the motion in the entire screen imaged by the video camera 1, the image signal in the measurement area (indicated by A in FIG. 1) set therein is processed. This is to detect movement. The operator operates the keyboard of the arithmetic processing unit 6 while looking at the screen of the image processing display unit 8 and sets an arbitrary number of the measurement areas A at an arbitrary position in the screen with an arbitrary size. Can be. In this case, the shape of the measurement area A may be fixed or arbitrarily changeable.

FIGS. 2 to 5 are diagrams for explaining the basic operation of the motion detection method based on image processing according to the present invention. FIG. 2 shows an image in one measurement area A set in an image of the monitoring site S sent from the video camera 1. This shows two sequential images when the object B in the measurement area is moving from right to left.
Now, taking into account the scanning line CC crossing the object B, let Y be the measurement range that is the entire length of this scanning line. That is, it is assumed that the horizontal size of the measurement area A is Y pixels.

FIG. 3 shows the pixels of the above two images on the horizontal axis and the density of each pixel on the vertical axis.
In this example, the image signal is processed as a digital signal,
The density of one pixel is represented by 8 bits, that is, 256 gradations.

FIG. 4 shows the relative density difference between these two images obtained for each pixel. Since the object B is moving, the relative density differences appear at two places. This density relative difference is the absolute value of the density difference between the two images. In the present invention, after calculating the relative density difference of each pixel of the two images in this way, the average is calculated. That is, the sum of the relative density differences is obtained, and a value obtained by dividing the sum by the total number of pixels Y within the measurement range is obtained as the average of the relative density differences.

In the present invention, the average of the relative density differences obtained as described above from the relative density differences obtained by comparing the two images is compared with a limit value, and the average of the relative density differences is equal to or greater than the limit value. When there is, it is determined that there is movement.
This limit value can be determined in consideration of, for example, the properties of the individual monitoring sites and the desired detection sensitivity. However, as described later, it is preferable to automatically calculate the limit value based on the average of the relative concentration differences. is there.

As described above, according to the present invention, two images are compared, and when a relative density difference is detected between corresponding pixels of these images, it is not determined that there is a motion immediately, but a measurement is performed. The average of the relative density differences between pixels in the area A is determined, and it is determined that there is a motion when the average is equal to or more than a predetermined limit value. If a bird simply crosses through it, it will not be detected as motion, but if the ground moves over a wide area,
Even if the amount of movement is slight, the average of the density relative differences becomes a large value, so that the movement is properly detected.

Further, in the present invention, it is preferable to automatically set the above-mentioned limit value based on the relative density difference, and an example thereof will be described below. The sum ΣM (i) of N relative density differences between two successive images of N + 1 images obtained by imaging the monitoring site S is obtained, and this sum is divided by N to obtain an average value. And the average value is multiplied by a correction coefficient k (k> 1) to obtain the limit value L. That is, the limit value L can be calculated according to the following equation (1).

(Equation 1)

The correction coefficient k in the above equation (1) is a value larger than 1, but can be set empirically in each measurement area. In general, if the measurement area exists in an environment where there is a relatively large amount of minute movements that can be regarded as noise, the value of k is set large, and if the noise is small, the value of k is set small. be able to.

Further, in the present embodiment, it is not determined that there is a motion immediately when the average of the above-mentioned density relative difference is equal to or more than the limit value, but instead when the average of the density relative difference first becomes equal to or more than the limit value. A trigger signal is generated, and after the trigger signal is generated, image comparison is performed a predetermined number of times. When it is determined that there is a motion more than a predetermined number of times, there is a significant movement. Is determined. For example, when the above-described trigger signal is generated, the image comparison is performed five times, and if the average of the relative density differences is equal to or more than the limit value in three or more of the five times, it is determined that there is a motion. is there. By employing such a determination method, accurate motion detection can be performed even in a measurement area where noise due to momentary movement is relatively large.

For more accurate detection, it is preferable to determine the movement based on a combination of a plurality of measurement areas. For example, as shown in FIG.
Are set in the monitoring site. In the detection condition in this case, the average of the relative density differences is compared with the respective limit values for each of the measurement areas, and the average value of the relative density differences in one or a plurality of predetermined measurement areas is determined. When the threshold value is exceeded, it can be determined that there is movement. For example, in some cases, it is determined that there is a motion when the average value of the relative density differences exceeds the limit value in any one of the entire measurement areas. Alternatively, in the measurement areas A1 and A2, when it is simultaneously detected that the average value of the density relative difference exceeds the limit value, or in any one of the measurement areas A3 to A6, the average value of the density relative difference is obtained. When it is detected that exceeds the limit value, it can be determined that there is movement. In this case, the position, the size, the shape, the condition of the motion determination and the like of the measurement area in the site can be optimally set according to the situation of the site, so that the accuracy of the motion detection can be further improved.

FIG. 6 is a flowchart showing the sequential steps of the motion detection processing in the above-described embodiment. After the start, in step S1, various parameters are initialized. That is, input of the position and size of the measurement area A, setting of the correction coefficient k, setting of the number of image comparisons N for automatically setting the limit value, setting of the number of image comparisons after the generation of the trigger signal, and setting of the shutter speed in the video camera Perform setting, automatic / manual setting of alarm device, setting of measurement time, setting of detection condition.

Next, in step S2, the limit value L is calculated based on the above-mentioned equation (1) and is automatically set. In step S3, it is checked whether or not the set measurement time has elapsed. If it is still within the measurement time, the average of the relative density differences is calculated in step S4. Next, in step S5, it is checked whether or not the average of the calculated relative density differences is equal to or greater than the limit value automatically set in step S2 described above. In S6, a trigger signal is generated.

Next, in step S7, after the trigger signal is generated, five image comparisons are performed, and it is checked whether or not the average of the relative density differences is three times or more among the three times or more. If it is not determined that the average of the relative density differences is equal to or more than the limit value in the image comparison of three or more times, the process returns to step S3, and it is determined that the average of the relative density differences is equal to or more than the limit value in the image comparison of three or more times. If so,
In step S8, it is determined whether the alarm device is automatic or manual. Here, when it is determined that the measurement is automatic, the alarm device is activated in step S9, and further, in step S10, it is checked whether or not the measurement is to be terminated. If not, the process returns to step S3.

On the other hand, if it is determined in step S8 that the alarm device is manually operated, a detected image is displayed (displayed on the image processing display unit 8 in FIG. 1) in step S11, and an operator in step S12 displays the detected image. After confirming the detection screen, it is determined in step S13 whether or not to activate the alarm device.
At 9 the alarm is activated.

The present invention is not limited to the above-described embodiment, but can be variously modified and modified. For example, in the above-described embodiment, the relative density difference was obtained by comparing the images of successive frames of the image signal sent from the video camera. Relative differences can also be determined. In such a case, instead of using a video camera having a solid-state imaging device such as a CCD as an imaging device, an electronic still camera can be used and operated intermittently at a predetermined cycle. Further, processing using an infrared camera may be considered.

In the above-described embodiment, the limit value is calculated automatically. However, an appropriate value can be set according to each measurement area. Similarly, the number of comparisons of images after the trigger signal is generated can be set to an optimal value according to the situation of each measurement area.

Further, in the above-described embodiment, the motion detection method using image processing according to the present invention is applied to a monitoring system for debris flow on a steep slope, but the method for detecting a dangerous slope such as rock fall, rock fall, fall, rock slide, and avalanche. The present invention can be applied to various remote monitoring systems such as monitoring, monitoring of high waves over a breakwater, monitoring of intrusion into buildings and premises, and monitoring of traffic congestion of road traffic.

[0031]

As described above, according to the present invention, a landslide
In remote monitoring systems for dangerous slopes such as rock fall, rock fall, falls, and rock slides, significant movements can be accurately performed without being affected by noise such as birds coming, small animals moving, and trees fluctuating due to wind. Detection can be performed with a sufficiently high sensitivity for practical use, and thus the reliability of the monitoring system can be improved. As a result, safety can be ensured more reliably.

Further, in the embodiment in which the limit value is automatically calculated based on the average of the relative density differences, the influence of the overall change of the image, for example, the change due to time or the change due to the weather, etc. Can be reduced. further,
By setting the number of images after the trigger signal is generated and the limit number of times at that time to optimal values according to the situation of each measurement area, the accuracy and sensitivity of motion detection can be optimized, and therefore, individual measurement It is very flexible for the area.

[Brief description of the drawings]

FIG. 1 is a diagram showing an entire configuration of a debris flow monitoring system to which a motion detection method based on image processing of the present invention is applied.

FIG. 2 is a diagram showing an operation of a motion detection method based on image processing of the present invention.

FIG. 3 is a diagram for explaining the operation thereof.

FIG. 4 is a diagram for explaining the operation thereof.

FIG. 5 is a diagram showing a situation where a plurality of measurement areas are set at a site to be monitored.

FIG. 6 is a flowchart showing the operation of the monitoring system shown in FIG.

[Explanation of symbols]

S monitoring site, A, A1-A6 measurement area, 1
Video cameras, 2 surveillance offices, 3 surveillance monitors,
4 image processing device, 5 signal processing unit, 6 operation processing unit, 7 storage unit, 8 image processing display unit, 9 controller, 10-12 alarm device

Continued on front page F term (reference) 5C087 AA09 AA19 BB10 BB18 CC51 DD02 DD05 DD20 DD49 EE14 FF01 FF04 GG02 GG08 GG23 GG29 GG30 GG31 5L096 AA06 BA02 CA04 CA24 DA03 FA32 GA08 GA51 HA02

Claims (5)

[Claims] 1. A step of obtaining a density relative difference representing a difference between two image signals obtained by imaging a site whose motion is to be detected before and after by an imaging device, and a step of obtaining an average of the density relative difference. And a step of comparing the average of the relative density differences with a limit value, and determining that there is a motion when the difference exceeds the limit value. 2. The motion detection method according to claim 1, wherein the average of the relative density differences is obtained by dividing the sum of the relative density differences by the total number of pixels in the measurement range. 3. A correction coefficient k (k> 1) obtained by dividing the limit value L by an average value obtained by dividing the sum of the average M (i) of N density relative differences obtained in succession by N. 3. The motion detection method according to claim 2, wherein the motion detection value is obtained by multiplying the motion detection value. 4. After detecting that the average of the relative density differences has exceeded a limit value, comparing the average of the relative density differences with a limit value over a predetermined number of times, and determining the number of times exceeding the limit value as a predetermined value. The method according to any one of claims 1 to 3, wherein it is determined that there is a motion when the number of times exceeds the number of times. 5. A plurality of measurement areas are set in a site where the movement is to be detected, an average of the relative density differences is calculated for each of the plurality of measurement areas, and the average is compared with a limit value. The image processing according to any one of claims 1 to 4, wherein it is determined that there is a motion when the average of the relative density differences exceeds a limit value in one or a plurality of predetermined measurement areas. Motion detection method.