JP2002058014A - Object detection system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To construct a system for automatically detecting failure generated in an image, namely an object for causing an obstruction. SOLUTION: The object detection system for detecting an object in a monitoring region by a plurality of cameras is provided with a perspective conversion part for converting the perspective of an image being shot by one of the cameras, a perspective/rotation conversion part for performing perspective and rotation conversions so that the image shot by the different camera is overlapped with an image obtained by the perspective conversion part, and an overlapping part for detecting an element as an object when there is a change in the element existing in the original image by overlapping the two images.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to an object detection system for processing an image obtained from a plurality of cameras to detect an object existing in the image.

[0002]

2. Description of the Related Art At present, cameras are installed at various places such as railroad crossings, power plants, substation power facilities, various plant facilities, building entrances, and the like, and images from the cameras are monitored even at remote places. Accordingly, it is common practice to monitor an abnormal state such as detection of an intruder or an obstacle.

[0003] In particular, at railroad crossings, an automatic obstacle detection device using a light beam or the like as a sensor may be installed, and an object as large as an automobile can be detected almost certainly.

Further, recently, web cameras and the like have been sold, and images to be monitored are also supplied to a display of a personal computer in a remote monitoring center via a server or a communication network. As a result, technologies for monitoring site images, storing and retrieving image data, and the like have been remarkably advanced.

[0005]

However, when an abnormality occurs in the monitored area, for example, when a human or a car is stuck in a railway crossing due to an accident, the image from the camera is monitored. Only the visual recognition of the operator is relying, and if the operator happens to miss it during another work, this abnormality is not found, and there is a risk that a secondary accident may occur.

Further, in an automatic obstacle detection device installed at a railway crossing or the like and using a light ray or the like as a sensor, it is difficult to detect a person, a bicycle, a wheelchair, and the like as an obstacle because the size thereof is small. Was.

An object of the present invention is to solve such a problem, and an object of the present invention is to construct a system capable of automatically detecting an abnormality that has occurred in an image, that is, an obstacle. .

[0008]

The present invention for achieving the above object is as follows. (1) In an object detection system that detects an object present in a monitoring area by a plurality of cameras, a perspective conversion unit that performs perspective conversion on an image captured by one camera, and the perspective conversion on an image captured by a different camera. A perspective / rotation conversion unit that performs perspective conversion and rotation conversion so as to overlap with the image obtained by the conversion unit; An object detection system, comprising: a superimposing unit that detects the object as an object. (2) The object detection system according to claim 1, wherein the superposition unit divides the monitoring area into a matrix area in advance and detects an object and detects a position of the object on the matrix area. (3) The object detection system according to Claim 1 or 2, wherein the number of cameras is two. (4) The object detection system according to claim 3, wherein the two cameras are directly opposed by 180 degrees so as to face the monitoring target. (5) The object detection system according to any one of claims 1 to 4, further comprising a communication unit that transmits the object detection information obtained by the superimposition unit to a communication network.

[0009]

Embodiments of the present invention will be described below in detail with reference to the drawings. FIG. 1 is an example of an object detection system embodying the present invention. The system of the present invention first captures an image of a monitoring area S such as a railroad crossing and a no-go zone by two web cameras, such as a first camera 1 and a second camera 2, to obtain image data.

The first camera 1 has an image maintenance unit 11 for maintaining and storing the image data as it is, a perspective / rotation conversion unit 12 for performing perspective conversion and rotation conversion of the image in the image maintenance unit 11, and a conversion of the converted image. An image cutout unit 13 for extracting only a necessary part is connected.

Similarly, the second camera 2 has an image inverting section 21 for vertically inverting the image data, a perspective / rotational converting section 22 for converting the inverted image into perspective and rotating, and a necessary portion of the converted image. The image cutout unit 13 for extracting only the image data is connected.

Further, an image superimposing section 3 for superimposing images from the image clipping sections 13 and 23 to detect an obstacle, and a communication section 4 for communicating information on the presence of an obstacle detected by the image superimposing section 3. Is provided.

Next, the operation of the system according to the present invention will be described with reference to FIG. In this example, a cylindrical object X is placed on a single book B, and this object X is detected as an obstacle object. On the other hand, the book B is being photographed.

First, an image photographed by the first camera 1 is once maintained as an image 1a in the image maintaining unit 11 like an image 1b,
The perspective conversion is performed as in the image 1c. In this case, since the first camera 1 is set at a position exactly 90 degrees with respect to the upper side of the cover of the book B, the rotation conversion is unnecessary.

In an actual site, it is extremely difficult to install a camera at a position 90 degrees with respect to the upper side of the monitoring target S, and the camera is usually slightly shifted. In such a case, the obtained image is once subjected to perspective transformation and then rotational transformation is performed to obtain an image 1c.

Subsequently, only the necessary portion of the image 1c is cut out by the image cutout section 13 to obtain the image 1d.
Get.

On the other hand, the second camera 2 is set at exactly 90 degrees with respect to the lower side of the book B for photographing.

The obtained image 2a is turned upside down by the image reversing section 21 to form an image 2b. Then, the image 2b is subjected to perspective conversion by the perspective / rotation conversion unit 22, and an image 2c is obtained.

At this time, as described above, the second camera 2
Is set at exactly 90 degrees with respect to the lower side of the cover of book B, so that rotation conversion is unnecessary. If the installation of the second camera 2 is slightly deviated, the obtained image is converted into perspective and then rotationally converted to obtain an image 2c.

Subsequently, only a necessary portion of the image 2c is cut out by the image cutout unit 13 to obtain the image 2d.
Get.

Next, the two cut-out images 1d and 2d
Are superimposed by the image superimposing unit 3. Then, in the superimposed image 3z, it is found that the cylindrical object x has changed to almost twice the original size. By detecting this change, the image superimposing unit 3 recognizes that an obstacle (cylindrical object x) exists in the monitoring target S.

The communication section 4 notifies the presence of the object (obstacle) as an abnormal signal to a remote monitoring center via a dedicated line, a communication network or the like.

In this way, it is possible to detect an obstacle or the like that has entered a captured image using image data from a plurality of cameras.

In the example of FIG. 2 described above, the two cameras are installed exactly 180 degrees, and the upper and lower sides of the camera to be monitored are also 90 degrees. However, at an actual site, for example, at a railroad crossing, it is ideal if the cameras 1 and 2 can be directly opposed to the same height by 180 degrees as shown in FIGS. Such an installation is extremely difficult.

Therefore, the image 1b obtained in FIG.
After performing the perspective conversion on both of the images 2b and performing the post-processing cutout processing, the rotation conversion processing is performed so that one of the images is adjusted to the inclination of the other image so that the two images overlap. After that, a superposition process is performed to detect an obstacle object.

After the detection of the obstacle, the post-processing is optional, for example, transmission to a remote monitoring center or the like via a communication network, or sounding an alarm buzzer or turning on an alarm lamp in the area. Can be built.

Further, in the image superimposing section 3, the monitoring area S is defined by rows a, b, c, d,.
By dividing the area in advance in the form of a matrix such as a row, a D row, and so on and setting the divided area, the superimposing unit 3 can also recognize where the obstructing object is located.

[0028]

As is apparent from the above description,
ADVANTAGE OF THE INVENTION According to this invention, even if the obstruction object exists in the monitoring area, the presence of this obstruction object can be automatically detected by a plurality of cameras.

Further, by transmitting the detection of the object through the communication network, the remote monitoring center can communicate the presence of the obstacle in the monitoring area.

By dividing the monitoring area in a matrix on the system side, it is possible to automatically detect an obstacle and detect its position information, and transmit its presence to an arbitrary location. Can be.

[Brief description of the drawings]

FIG. 1 is a conceptual diagram illustrating the configuration of an object detection system according to the present invention.

FIG. 2 is a diagram showing an image for explaining the operation of the object detection system of the present invention.

FIG. 3 is a diagram illustrating an example in which the object detection system of the present invention is applied.

FIG. 4 is a diagram illustrating an example in which the object detection system of the present invention is applied.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 1st camera 2 2nd camera 3 Image superposition part 4 Communication part 11 Image maintenance part 12,22 Perspective / rotation conversion part 13,23 Image cutout part 21 Image inversion part

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Koji Ishibashi 2nd floor of Seibu Ikebukuro Station South Building, 1-18-18 Minami Ikebukuro, Toshima-ku, Tokyo In-house (72) Inventor Naoki Matsumoto Musashino, Tokyo 2-9-32 Ichinakacho Yokogawa Electric Co., Ltd. (72) Inventor Tetsu Noguchi 2-9-132 Nakamachi Musashino-shi, Tokyo Yokogawa Electric Co., Ltd. (72) Inventor Shuichi Takeuchi Nakamachi, Musashino-shi, Tokyo 2-9-32 Yokogawa Electric Corporation F term (reference) 5C022 AA01 AB61 AB62 AB65 AC01 AC69 AC75 CA00 5C054 AA01 CC06 CG02 CH01 DA06 EA05 FA09 FC12 FE12 FF06 HA18 5C087 AA02 AA03 BB02 BB74 DD02 DD03 DD08 DD21 DD22 EE05 EE06 FF22 GG02 GG11 GG12 GG19 GG32 GG55

Claims (5)

[Claims] 1. An object detection system for detecting an object present in a monitoring area by a plurality of cameras, wherein a perspective conversion unit that performs perspective conversion on an image captured by one camera and an image captured by a different camera. A perspective-rotation conversion unit that performs perspective conversion and rotation conversion so as to overlap with the image obtained by the perspective conversion unit, and if there is a change in the element existing in the original image by superimposing the two images, the element And an overlapping unit for detecting the object as an object. 2. The object detection system according to claim 1, wherein the superposition unit divides the monitoring area into a matrix area in advance and detects an object and detects a position of the object on the matrix area. 3. The object detection system according to claim 1, wherein the number of cameras is two. 4. The object detection system according to claim 3, wherein the two cameras are opposed to each other by 180 degrees so as to face the monitoring target. 5. The object detection system according to claim 1, further comprising a communication unit that transmits the object detection information obtained by the superimposition unit to a communication network.