JP2001260885A - Obstacle detecting system and obstacle detecting method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To ensure detection of an obstacle without malfunction due to surrounding environment. SOLUTION: Two digital monitoring cameras 11a, 11b are used to trap polygonal images in a preset monitoring region, which are three-dimensionally processed by a three-dimensional image processing part 21. A difference between the image generated by the three-dimensional processing and a background image in the preset monitoring region stored in a background image storage part 23 is picked up for extracting a monitored object from the image with a background portion removed. This monitored object is analyzed by an analysis part 22 to judge whether the monitored object is an obstacle or not from the information of the three-dimensional image. If it is an obstacle, alarm information is output to an external alarm system through a data interface part 24. Such a detecting system utilizing the image prevents malfunction which may be caused in a conventional system utilizing an infrared ray and ensures detection of an obstacle.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to an obstacle detection system and an obstacle detection method for detecting an obstacle at a railroad crossing, for example, on a railway.

[0002]

2. Description of the Related Art Conventionally, as an obstacle detection system at a railroad crossing, a system utilizing infrared rays or light is known. This means that, for example, a transmitter and a receiver are installed on both sides of a railroad crossing at a height of about 1 m, and when the infrared or light between the transmitter and the receiver is blocked when the circuit breaker is closed, there is an obstacle. Is detected. The obstacle mentioned here is an obstacle that obstructs the passage of a train, for example, a vehicle or a person. When an obstacle is detected, warning information is output to an external warning system, and the train is stopped urgently to prevent a railroad crossing accident.

[0003]

As described above, an obstacle at a railroad crossing has conventionally been detected using infrared rays or light. However, in such a system, for example, infrared rays or light emitted from a transmitter may enter a receiver at another adjacent railroad crossing and malfunction. In addition, the light of a stopped vehicle before a railroad crossing at night, reflected light such as sunset, or reflected light from the road surface due to rain or snow may enter the receiver and malfunction. Further, if newspaper or the like is flying between the transmitter and the receiver due to wind or the like, it may be erroneously detected as an obstacle.

[0004] As described above, the method using infrared rays or light has a high possibility of malfunction due to the surrounding environment and the like, and there is a problem that an obstacle cannot be reliably detected. That is the current situation.

The present invention has been made in view of the above points, and provides an obstacle detection system and an obstacle detection method capable of reliably detecting an obstacle without malfunction due to the surrounding environment. With the goal.

[0006]

SUMMARY OF THE INVENTION An obstacle detection system according to the present invention comprises at least two surveillance cameras and an image processing for three-dimensionally processing an image in a predetermined surveillance area captured by the surveillance cameras from various angles. Means, an analyzing means for analyzing the three-dimensional image obtained by the image processing means and detecting the presence or absence of an obstacle, and a notifying means for notifying to the outside when an obstacle is detected by the analyzing means And is provided.

According to such a configuration, by detecting an obstacle by analyzing an image captured by the surveillance camera, a signal leaks to another place and malfunctions as in the conventional system using infrared rays or the like. Therefore, it is possible to correctly detect an obstacle only in the area to be monitored only when the obstacle is not illuminated by a light of a vehicle or the like. Also,
By creating a three-dimensional image using a plurality of surveillance cameras, an obstacle can be reliably detected from more information than an image captured planarly by one surveillance camera.

Further, the present invention has a background image storage means for storing a background image in the predetermined monitoring area, wherein the image processing means stores the image in the predetermined monitoring area in the background image storage means. Removing the extracted background image and performing a process of extracting a monitoring target, wherein the analysis unit determines whether or not the extracted monitoring target is an obstacle based on information of the three-dimensional image. I do.

According to such a configuration, only the monitoring target is correctly extracted from the image obtained by removing the background portion in the predetermined monitoring area from the image captured by the monitoring camera, and the monitoring target is an obstacle. Whether or not there is can be determined based on the information of the three-dimensional image.

The present invention also provides an image transfer means for transferring an image of each of the monitoring cameras, an image recording means for recording the image transferred from the image transfer means for a predetermined unit,
An image monitoring means for displaying an image recorded on the image recording means on a screen is further provided.

According to such a configuration, it is possible to use the image of the surveillance camera, for example, by holding the image on the external surveillance system side and displaying the image on the screen as necessary.

The present invention is also an obstacle detection system for detecting an obstacle at a railroad crossing, comprising at least two surveillance cameras, and a plurality of surveillance cameras within the crossing area captured by the surveillance cameras. Image processing means for three-dimensionally processing the image; analysis means for analyzing the three-dimensional image obtained by the image processing means to detect the presence or absence of an obstacle; and obstacle analysis by the analysis means before the train passes the railroad crossing. And a notifying means for notifying an external alarm system when the presence of the object is detected.

According to such a configuration, it is possible to detect an obstacle that has entered a railroad crossing from an image from the surveillance camera and notify an external alarm system. In this case, as described above, there is no malfunction as compared with the conventional method using infrared rays or the like, and more information can be obtained by creating a three-dimensional image, so that an obstacle in a railroad crossing can be reliably detected. it can.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a diagram showing an external configuration of an obstacle detection system according to an embodiment of the present invention.

The obstacle detection system according to the present embodiment detects an obstacle at a railroad crossing at a railroad crossing, and includes two digital surveillance cameras 11a and 11b as shown in FIG. ing. The digital surveillance cameras 11a and 11b are installed at predetermined positions on both sides of the railroad crossing with the track 10 interposed therebetween. Specifically, the traffic signals 12a, 12a at a high level of a railroad crossing, for example, about 4 m in height.
b, respectively, and is installed so as to monitor a level crossing below from that position. In this case, the installation angles of the digital monitoring cameras 11a and 11b are devised in order to capture the predetermined monitoring area 13 in the railroad crossing from various angles. The configuration of the digital surveillance cameras 11a and 11b is not particularly limited, but is assumed to be an all-weather type that can be used outdoors.

FIG. 2 shows a functional configuration of the obstacle detection system.

The present system comprises the above-described digital surveillance cameras 11a and 11b, a three-dimensional image processing unit 21, and an analysis unit 2.
2. Background image storage unit 23, data interface unit 24
It is composed of

The three-dimensional image processing section 21 receives as input the first image in the level crossing taken by the digital surveillance camera 11a and the second image in the level crossing taken by the digital surveillance camera 11b in the monitoring mode. First and second
For example, a process of generating a three-dimensional image by using the image is performed. The analysis unit 22 performs a process of detecting the presence or absence of an obstacle from the three-dimensional image generated by the three-dimensional image processing unit 21. The background image storage unit 23 stores, as a background image, an image of the inside of a railroad crossing taken without an obstacle before monitoring.

The data interface unit 24 is an interface with an external alarm system.
When an obstacle is detected, alarm information for alerting the user of the obstacle is output to an external alarm system.

Next, the processing operation of the present system will be described.

Two digital surveillance cameras 11a and 11b
Is installed on the traffic signals 12a and 12b as shown in FIG. 1, for example, and simultaneously monitors a predetermined monitoring area 13 in the railroad crossing. The reason why a plurality of monitoring cameras (here, two) are monitored simultaneously instead of a single monitoring camera is to generate a nearly three-dimensional image by image processing. Since one monitoring camera can capture an image only in a planar manner, there is a possibility that a shadow projected in a railroad crossing by a vehicle light, for example, is erroneously detected as an obstacle. On the other hand, if a plurality of surveillance cameras capture an object from multiple angles to create a three-dimensional image, the amount of information obtained from the three-dimensional image increases, and such erroneous detection is prevented. be able to.

The images captured by the digital surveillance cameras 11a and 11b in advance without any obstacles at the railroad crossings are processed by the three-dimensional image processing unit 21 and are stored in the background image storage unit 23 as background images. Remember. By taking the difference between the background image and the camera image captured when entering the monitoring mode, only the monitoring target can be correctly extracted. In this case, if the background image is stored in the background image storage unit 23 before monitoring every time, the background image according to the change in brightness, color, and shadow in the morning, day, night, fog, snow, rain, etc. is used. In addition, the accuracy of extraction of the monitoring target can be improved.

Here, when the alarm sounds and the circuit breaker closes a few meters before the train passes the railroad crossing, a monitoring mode is entered,
An obstacle detection process as shown in FIG. 3 is executed. In addition,
Passage of the train can be known from information from a sensor installed on a track, for example, a few meters before the railroad crossing, information from a system that centrally manages each line, and the like.

FIG. 3 is a flowchart showing the flow of the obstacle detection processing in the present system.

When entering the monitoring mode, the image of the inside of the railroad crossing captured by the digital monitoring cameras 11a and 11b is taken into the three-dimensional image processing section 21 (step S11).
The capture of the camera image is performed at predetermined intervals, for example, after entering the monitoring mode. That is,
In the present embodiment, the following obstacle detection is performed using snapshots obtained at predetermined intervals.

First, the three-dimensional image processing section 21 combines the two images obtained from the digital surveillance cameras 11a and 11b to generate a three-dimensional image in the railroad crossing (step S12). The three-dimensional image is generated by a well-known method from the relationship with the camera position, and the present invention is not particularly limited to the generation method.

Next, the three-dimensional image processing unit 21 reads the background image from the background image storage unit 23, and obtains the difference between the image in the railroad crossing obtained in the monitoring mode and the background image,
Remove stationary objects such as rails and sleepers (Step S1
3) A process of extracting only the monitoring target from the camera image is performed (step S14).

The analysis unit 22 analyzes the monitoring target obtained by the three-dimensional image processing unit 21 based on information from the three-dimensional image, and analyzes whether or not this is an obstacle (step S15). . The obstacle is an obstacle to the passage of a train, for example, a vehicle or a person, and can be determined by three-dimensionally analyzing the shape and the position of the monitoring target.

If the analysis unit 22 determines that the object is an obstacle (Yes in step S16), the fact is notified to the data interface unit 24, and an alarm is issued to an external alarm system through the data interface unit 24. Information is output (step S17). Thereby, the alarm system is configured to, for example, sound an emergency alarm or turn a traffic light installed on the track red to avoid collision with the train.

As described above, by detecting the obstacle by analyzing the image of the railroad crossing captured by the monitoring camera, the signal from the transmitter leaks to another adjacent receiver and malfunctions as in the conventional system. It is possible to detect and output alarm information only when there is an obstacle in the level crossing to be monitored, without causing a malfunction such as a vehicle light entering the receiver.

Further, by generating a three-dimensional image using two surveillance cameras and analyzing the three-dimensional image, it is possible to obtain more information than an image captured two-dimensionally by one surveillance camera. Can be. This makes it possible to reliably detect only obstacles to the passage of trains as obstacles, for example, without falsely detecting shadows projected by vehicle lights or newspapers flying in the wind as obstacles. Can be.

In the above embodiment, the surveillance cameras are
However, if a plurality of surveillance cameras are installed to capture images from various directions, the accuracy of obstacle detection can be further improved.

Further, as an application example, an image of a railroad crossing is transferred to a station building under the jurisdiction and an image for one day is recorded at the transfer destination, so that the cause of the accident can be analyzed from the recorded image when an accident occurs. You can also do that. FIG. 4 shows a configuration example of such a system.

FIG. 4 is a block diagram showing an application example of the obstacle detection system of the present invention.

On the level crossing side, two digital surveillance cameras 3
The image inside the railroad crossing captured by 1a, 31b is transferred to the station building under the jurisdiction through the data modem 32. On the station building side,
The image in the railroad crossing is received via the data modem 33. Then, the received image of the railroad crossing is converted into a three-dimensional image processing unit 3.
After the three-dimensional processing in step 4, the image is recorded in an image recording unit 35 composed of, for example, a magnetic disk device or an optical disk device. The unit of recording is, for example, one day from the first train to the last train. The image in the railroad crossing recorded in the image recording unit 35 may be a moving image or a still image.

Here, for example, when an accident occurs,
By reading an image at a corresponding time from the image recording unit 36 and displaying the image on the screen by the image monitor unit 36, the cause of the accident can be analyzed.

It is also possible to display an image of a railroad crossing on the image monitor section 36 in real time, and to monitor the state of the railroad crossing while the station staff watches the monitor screen. In this case, since a person watches the monitor screen, the image need not necessarily be a three-dimensional image.

Further, the present system can be applied not only to railroad crossings but also to other fields. For example, FIG.
As shown in the figure, a pedestrian crossing 40 of a road on which a vehicle such as an automobile runs is to be monitored, and digital surveillance cameras 41a and 41
It is also possible to monitor by attaching b.

The obstacles here are the traffic signals 41a, 41
When b is red, the person is going to cross the pedestrian crossing 40. When an obstacle is detected from the camera image, for example, until the obstacle disappears (that is, when the person
If the traffic signals 41a and 41b are kept in a red light state until the vehicle stops running, a collision accident can be prevented. However, due to traffic regulations, it is impossible to forcibly stop the vehicle, and since many vehicles run ignoring the red light, people who cross the pedestrian crossing 40 by, for example, sounding an alarm, etc. A system that calls attention to is realistic.

In addition, taking advantage of the advantages of the camera images, the digital surveillance cameras 41a and 41b may be used to capture the number of the running vehicle when the traffic lights 41a and 41b are at a red light, and transfer the numbers to an external surveillance system. It is possible.

In short, the present invention is not limited to the above embodiment, but can be variously modified and applied to various fields without departing from the gist thereof.

[0043]

As described above in detail, according to the present invention, an image captured by a surveillance camera is analyzed to detect an obstacle, thereby eliminating a malfunction such as a conventional method using infrared rays or the like. Only when there is an obstacle in the area to be monitored, it can be correctly detected. Also,
By creating a three-dimensional image using a plurality of surveillance cameras, an obstacle can be reliably detected from more information than an image captured planarly by one surveillance camera. Thus, when a railroad crossing is targeted, an obstacle present in the railroad crossing when the circuit breaker is closed can be reliably detected to prevent a railroad crossing accident or the like.

[Brief description of the drawings]

FIG. 1 is a diagram showing an external configuration of an obstacle detection system according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a functional configuration of the obstacle detection system.

FIG. 3 is a flowchart showing a flow of an obstacle detection process in the obstacle detection system.

FIG. 4 is a block diagram showing an application example of the obstacle detection system of the present invention.

FIG. 5 is a diagram showing a conceptual configuration when the obstacle detection system of the present invention is applied to a crosswalk on a road.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 ... Track 11a, 11b ... Digital surveillance camera 12a, 12b ... Traffic light 13 ... Surveillance area 21 ... 3D image processing part 22 ... Analysis part 23 ... Background image storage part 24 ... Data interface part 31a, 31b ... Digital surveillance camera 32 ... Data modem 33 Data modem 34 Three-dimensional image processing unit 35 Image recording unit 36 Image monitoring unit 40 Crosswalk 41a, 41b Traffic light 42a, 42b Digital surveillance camera

Claims (5)

[Claims] At least two surveillance cameras, image processing means for three-dimensionally processing an image in a predetermined monitoring area captured by the surveillance cameras from various angles, and a three-dimensional image obtained by the image processing means An obstacle detection system comprising: analysis means for analyzing the presence of an obstacle; and notifying means for notifying to the outside when an obstacle is detected by the analysis means. . 2. The image processing apparatus according to claim 1, further comprising: a background image storage unit configured to store a background image in the predetermined monitoring area, wherein the image processing unit converts the image in the predetermined monitoring area into a background image stored in the background image storage unit. 2. A process for extracting a monitoring target object by removing the monitoring target object, wherein the analysis means determines whether or not the extracted monitoring target object is an obstacle based on information of a three-dimensional image. Obstacle detection system as described. 3. An image transfer means for transferring an image of each surveillance camera, an image recording means for recording a predetermined unit of the image transferred from the image transfer means, and an image recorded on the image recording means. 2. The obstacle detection system according to claim 1, further comprising: an image monitoring means for displaying the image on a screen. 4. An obstacle detection system for detecting an obstacle at a railroad crossing, comprising: at least two monitoring cameras; and three-dimensional images of the crossing area captured by the monitoring cameras from various angles. Image processing means for processing; analyzing means for analyzing a three-dimensional image obtained by the image processing means to detect the presence or absence of an obstacle; and existence of an obstacle by the analysis means before the train passes the railroad crossing. An obstacle detection system, comprising: a notifying unit that notifies an external alarm system when the detection is performed. 5. An image in a predetermined monitoring area is captured from a plurality of angles using at least two monitoring cameras, these images are three-dimensionally processed, and the predetermined monitoring is performed from an image generated by the three-dimensional processing. A background portion in the region is removed, a monitoring target is extracted from the image after the background portion is removed, and it is determined whether or not the monitoring target is an obstacle based on information of the three-dimensional image. An obstacle detection method comprising: outputting alarm information when it is determined that an obstacle has been detected.