JPH1183437A - Distance measuring system and recording medium - Google Patents

Abstract

(57) Abstract: The present invention relates to a distance measuring system and a recording medium for measuring the position of an object on a road, and relates to an image of a camera when monitoring the position of the object on a road with a plurality of linked cameras. Detects the presence or absence of an object over a wide area in an area with low distance accuracy in the middle, and when the object is detected, accurately measures the position of the object in the area with high distance accuracy of the other camera that cooperated, and controls multiple cameras An object of the present invention is to measure a position of an object on a road surface in a wide range at high speed and with high accuracy in cooperation. SOLUTION: A plurality of cameras sequentially arranged along a road in such a manner that an area with a low distance accuracy in an image of a certain camera and an area with a high distance accuracy in an image of the next camera overlap each other, Detects the presence or absence of an object on the road in an area with low distance accuracy in the image of one of the cameras and measures the position of the object in the area with high distance accuracy in the image of the corresponding camera when detected And means for performing the operation.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a distance measuring system for measuring the position of an object on a road and a recording medium.

[0002] It is not preferable from the viewpoint of safety that an object is left in a passage such as a road or a corridor in a road or a company. A camera such as a TV camera is required to be installed along a passage such as a road or a corridor, to detect an abandoned object, and to notify a management station of its position. There is a demand for a system that can measure at higher speed with higher accuracy than conventional cameras.

[0003]

2. Description of the Related Art Conventionally, as a system for measuring the position of an object placed on a passage such as a road or a corridor, there is a system having the concept shown in FIG. 7, for example. With the system shown in FIG. 7, an object placed on a road surface or a corridor is imaged by a TV camera, and the position of the object on the road surface is measured and monitored by an image processing device. FIG. 8 shows the measurement method. The distance L from the point P of the object on the road surface to the road surface point O immediately below the camera (TV camera), the focal length f of the camera, the inclination θ of the camera, the height H of the camera, and the image of the object on the imaging surface of the camera If the position a (distance from the center of the image) is known, L can be calculated by (Equation 1) shown.

[0004]

The distance difference when the position a of the object on the image is shifted by one pixel using the above-mentioned (Equation 1) is shown in (Equation 1.1). As can be seen from (Equation 1.1), the resolution with respect to the distance is high and the accuracy is high in front of the camera (where the position a is small). However, as the position a increases, the distance difference increases and the measurement accuracy decreases. Conventionally, the processing of calculating the distance L has been performed by (Equation 1) over the entire surface of the image of the camera.
There has been a problem that accuracy is reduced.

[0005] The present invention solves these problems,
When monitoring the position of an object on the road surface with multiple linked cameras, another camera linked when an object is detected by detecting the presence or absence of an object in a wide range in an area with low distance accuracy in the image of one camera It is an object of the present invention to accurately measure the position of an object in an area with high distance accuracy, and to measure the position of the object on a road surface at high speed and with high accuracy in cooperation with a plurality of cameras.

[0006]

Means for solving the problem will be described with reference to FIG. In FIG. 1A,
S2 performs image processing on the area B of the camera i. In this case, the presence or absence of an object on the road is detected by image processing in an area B (for example, an upper half area) in the image 12 of the camera i shown in FIG.

In S3, it is determined whether or not there is an object in the area B. Y
In the case of ES, the process proceeds to S4. If no, repeat for the next camera. In step S4, image processing is performed on the area A of the camera (i + 1) (next camera). In this case, since it is determined that an object exists in the area B of the camera i in S2, image processing of the object is performed in the area A of the next camera corresponding to this area B, and the position of the object is accurately measured.

As described above, the image processing is divided into two stages: image processing for detecting an object in the area B of S2 and image processing for detecting the accurate position of the object in the area A when the object is detected. Is to be detected.

Next, an example of image processing will be described. A camera is sequentially arranged along a road in such a manner that an area with low distance accuracy (area A) in an image of a certain camera and an area with high distance accuracy (area B) in the image of the next camera overlap each other, The presence / absence of an object on the road is detected in an area (area B) having a low distance accuracy in an image of one of the plurality of cameras, and an area having a high distance accuracy in an image of the corresponding camera ( The position of the object is accurately measured in the area A).

At this time, each camera repeats detecting an object on the road in an area (area B) having a low distance accuracy in the image of each camera, and when the object is detected, the distance accuracy in the image of the corresponding camera is detected. In a high area (area A), the position of the object on the road is measured with high accuracy.

In addition, an object on the road is detected in a region (region B) with a low distance accuracy in the image of each camera, and an approximate position is obtained, and a region with a high distance accuracy (region A) in the image of the corresponding camera. In this method, the position of an object on the road is measured with high precision and at high speed by restricting the position to a predetermined range including the detected approximate position.

Therefore, when monitoring the position of an object left on the road surface with a plurality of linked cameras, the presence or absence of the object is detected over a wide area in an area with a low distance accuracy (area B) in the image of the camera. By accurately measuring the position of an object in a region (region A) with high distance accuracy of another camera that cooperated when the object was detected, a plurality of cameras can cooperate to detect an object left on the road surface. The position can be measured at high speed and with high accuracy over a wide range.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, an embodiment and operation of the present invention will be sequentially described in detail with reference to FIGS.

FIG. 1 is a diagram for explaining the operation of the present invention. FIG.
(A) shows a flowchart. In FIG. 1A, S1 is initially set to a variable i = 1. here,
As shown in FIG. 2 to be described later, n cameras are arranged in order along the road surface, and i is initially set to 1 (the first camera 1).

In step S2, image processing is performed on the area B of the camera i. This is performed by performing image processing to detect the presence or absence of an object placed on the road in an area B (for example, an upper half area) having a low distance accuracy and which can be detected in a wide range in the image 12 of the camera i shown in FIG. I do.

In S3, it is determined whether or not there is an object in the area B. Y
In the case of ES, the process proceeds to S4. In the case of NO, the process proceeds to S5. S4 is the area B of the camera (i + 1) (next camera)
Image processing. In this case, since it is determined that an object exists in the area B of the camera i in S2, image processing of the object is performed in the area A of the next camera corresponding to this area B, and the position of the object is accurately measured.

In step S5, i = i + 1. In S6, it is determined whether or not the processing is over. In the case of YES, the determination of whether or not there is any object on the road surface in the area B for all the cameras has been completed, and thus the processing ends (END). If NO, S
Return to step 2 and repeat.

As described above, a plurality of cameras are sequentially arranged on the road surface, and the presence or absence of an object is sequentially detected in the area B in the image from the first camera 1 to the last camera n. It repeats the accurate measurement of the position of the object in the area A in the image of the above, and detects the presence / absence of the object over a wide range in the area B with low distance accuracy, and only when the object is detected, the distance accuracy of the next camera is high. By accurately measuring the position of the object in the area A, it becomes possible to measure the position of the object left on the road surface at high speed and with high accuracy in cooperation with a plurality of cameras.

FIG. 1B shows an example of an image 12 of the camera 11. Here, the upper half in the image is an area B with low distance accuracy, and the lower half is an area A with high distance accuracy. The details will be sequentially described below.

FIG. 2 is an explanatory view (part 1) of the present invention. FIG. 2A shows an example of camera arrangement. Here, as cameras i (i = 1 to n), cameras A, B,
C: As shown in the drawing, the area B in the image of a certain camera and the area A in the image of the next camera overlap so as to satisfy Equation 2 described later in order along the road surface. An example is shown below. Here, A1, B1: a point at which the end of the area A of the cameras A, B intersects the road surface A2, B2: a point at which the optical axes of the cameras A, B intersect the road surface A3, B3: the points at the cameras A, B HA, HB: distance from camera A, B to road surface l: distance between cameras A, B, etc. LA ': distance between O-A1 LA: Distance between O-A2 ・ LAB: Distance between O-B1 ・ LB: Distance between O-B2 ・ θA, θB = angle between the optical axis of cameras A and B and the road surface ・ φA, φB: camera A, And the maximum angle of the B image.

FIG. 2B shows an example of camera arrangement conditions. Here, as shown in the upper equation, referring to the diagram of FIG. 2A, it is assumed that LA ′ <LAB <LA <LB, and LA ′, LAB, LA, and LB are substituted for these. (Equation 2) is obtained.

FIG. 2C shows an example of images of the cameras A and B. (C-1) of FIG. 2 shows an image of the camera A.
The presence or absence of an object is detected for the upper half area B in the image of the camera A. When it is detected that an object exists in the area B, the position of the object in the area A in the image of the camera B next to (c-2) in FIG.

FIG. 2C-2 shows an image of the camera B. The position of the object is accurately measured for the lower half region B in the image of the camera B. Next, the operation of detecting the position of the object in the area B of the camera A in FIG. 2 using FIG. 3 and accurately measuring the position of the object in the area A of the next camera B when the object is detected Will be described in detail.

FIG. 3 is an explanatory view (part 3) of the present invention. FIG. 3A shows an overall system configuration diagram. This satisfies the camera arrangement condition of FIG. 2B described above for camera i (i = 1 to n) (satisfies (Equation 2)).
In this arrangement, processing modules i (i = 1 to n) for processing images read from the cameras (i = 1 to n) are connected in order as shown in the figure. Thus, when a certain processing module i detects the presence of an object in the upper area B of the image, the next processing module (i + 1) accurately measures the position of the object in the lower area A of the image. Thus, the processing modules are sequentially connected.

FIG. 3B shows the first processing module 1.
Is shown (i = 1). Top processing module 1
Receives an image from the camera 1 as an input, extracts an object in an upper region B, outputs "no object" as a result,
Alternatively, the result of “object present” is notified to the next processing module 2. In addition, for the first processing module, since there is no notification of “object present” in the area B in the previous processing module, the object is directly extracted here, and if the object can be extracted, the result of “object position” is obtained. Output. If the object cannot be extracted, the result of the “object position” is not output.

By configuring the first processing module 1 as described above, when an object is detected in the area B in the upper part, the result of "object present" is notified to the next processing module 2, while the object is detected. If no object is detected, a result of "No object" is output. In the lower region A, the object is directly extracted. When the object can be extracted, the result of the “object position” is output. When the object cannot be extracted, the “object position” is output.
Do not output the result of

FIG. 3C shows a configuration diagram (i = 2 to n) of the processing module i. The processing module i receives an image from the camera i as shown in the figure, extracts an object in the upper region B, outputs "no object" as a result, or outputs "object present" to the next processing module (i + 1). Notification of the result. Also, the processing module i extracts an object and outputs the result of the "object position" only when the notification of "object presence" is received in the area B of the previous processing module (i-1).

By configuring the processing module i (I = 2 to n) as described above, when an object is detected in the upper region B, the result of "object present" is notified to the next processing module (i + 1). On the other hand, when an object cannot be detected, a result of "No object" is output. Also, in the lower area A, the object is extracted in the area A only when the result notification of "object present" is received from the previous processing module (i-1), and the result of the "object position" is output.

As described above, each processing module i detects the presence / absence of an object in the upper region B in the image and determines that “object present”.
, Each processing module i notifies the next processing module (i + 1) to detect and output the “object position” in the lower area A in the image. Two-stage image processing (the presence / absence of an object in the area B is detected by the processing module in the first stage, and the position of the object in the area A is measured by the next processing module in the second stage only when there is an object. It is possible to perform high-speed and high-precision position detection of the object by the two-stage image processing).

Here, considering the high speed, the processing amount of the region A and the region B is proportional to the area thereof. The area ratio between the area A and the area B is at least three times. Conventionally, since there is no overlapping of images, the number of cameras is の of that of the present invention.
It is. Therefore, assuming that the processing amount of the area B is P and the number of cameras of the present invention is n, conventionally, 4nP / 2 = 2n
P. In the present invention, nP + 3P when there is no object
Thus, the processing amount of about の of the conventional processing is sufficient. If there are objects, and if there are m cameras with objects in the area B, the processing amount of the present invention is nP + m3P + 3P. It is nP + m3P + 3P> 2nP where the processing amount is smaller than before, and m>
n / 3-1. For example, if there are six cameras, the amount of processing is reduced if leaving the object is one event. Assuming that the processing amount of the object is in an abnormal state, the frequency of occurrence of the event is not large, and there is a high possibility that the processing amount is reduced, and an effect can be expected.

FIG. 4 is an explanatory view (part 3) of the present invention. FIG. 4A shows a configuration diagram. 4A, the background difference calculation means 21 calculates a background difference based on the image of the area B (described later with reference to FIG. 4B).

The labeling means 22 performs labeling based on the background difference calculated by the background difference calculation means 21, and a different label (number) is assigned to the binary image of 1 and 0 for each of eight adjacent neighbors. Is attached.
As a result of the labeling, the area corresponding to the same label corresponds to a left-behind object.

The area calculating means 23 includes the labeling means 22
The number of pixels having the same label is counted for each label after the labeling, and the value is calculated as an area. As a result, a label having an area equal to or larger than a certain value is determined as an object, and the next processing module is notified of the presence of the object (object present). If there is no labeled area, or if there is a labeled area but the area is less than a certain value, it is determined that the object does not exist (there is no object). Specifically, it outputs “1” when there is an object, and outputs “0” when there is no object.

FIG. 4B shows an example of a processing method by the background difference calculation means 21. 4 (b-1) shows an image at time t, FIG. 4 (b-2) shows a background image up to time (t-1), and FIG. 4 (b-3) shows an initial background image Is shown. Here, an image Pt (i, j) at time t, a background image B (i, j) until time (t-1), and an initial background image IB (i, j).

Step 1: Pt (i, j) -B (i,
j) is performed for each pixel, and if the result is:> 0, B (i, j) ← B (i, j) + α; if <0, B (i, j) ← B (i, j) If −α · = 0, B (i, j) ← B (i, j). Here, α is a fixed value. More specifically, when the image P0 (i, j) of the area B in the image from the camera i is at time 0 (when power is turned on or reset), (b)
-2) Copy the background image B (i, j) and (b-3) the initial background image IB (i, j) and store them in the system.
Then, the image Pt (i, j) at the time t (t> 0) is compared with the background image B (i, j) in pixel units (Pt (i, j)).
j) -B (i, j)). If it is larger than 0, the fixed value a is set to the background image B (i, j).
If the value is smaller than 0, the fixed value a is set to the background image B (i, j).
Update the image B (i, j) if they are equal.

Step 2: IB (i, j) -B (i,
The calculation of j) is performed for each pixel, and if the result is .gtoreq..beta., the result R (i, j) .ltoreq..ltoreq..beta. Here, β is a threshold value. Specifically, (b-
3) Subtract (b-2) the background image B (i, j) from the initial background image IB (i, j). If the result is larger than β, the result R (i, j) is smaller than 1 · β. Then, the result R (i, j) is set to 0. The result R (i, j) is output to the labeling means 22.

As described above, the result R (i,
The value of j) is output as 1 or 0 for each pixel. Known labeling is performed based on the result R (i, j) of the background difference, and an object having a predetermined area or more is detected as an object.

FIG. 5 is an explanatory view (part 4) of the present invention. FIG. 5A shows a configuration diagram. Here, the background difference calculating means 21, the labeling means 22, the area calculating means 23
Are the same as those in FIG. 4A described above, and a description thereof will be omitted.

In FIG. 5A, the position calculating means 24
Calculates the position of the object detected as being equal to or larger than the predetermined threshold value by the area calculating means 23 (described later using FIG. 5B).

FIG. 5B shows an example of position calculation. Hereinafter, the position calculation procedure will be described. Step 1: The image shown (labeling result) is scanned sequentially from the bottom to the top, and each label is first detected at a point P (pixel position).
Ask for.

Step 2: The distance a from the point P obtained in Step 1 to the center of the image is obtained. Step 3: The distance L is substituted into (Equation 1) of FIG.
Ask for. Thus, the position (distance L) of the object has been calculated.

FIG. 6 is an explanatory view (part 5) of the present invention. FIG. 6A shows a configuration diagram. Here, the background difference calculating means 21, the labeling means 22, the area calculating means 23
Are the same as those in FIG. 4A described above, and a description thereof will be omitted.

In FIG. 6A, the position (B) calculating means 24 obtains the existence range of the object detected as being equal to or larger than the predetermined threshold value by the area calculating means 23, and determines the position within the area A corresponding to this existence range. The object detection is executed at high speed within the range of (1) (described later using FIGS. 6B and 6C).

FIG. 6B shows an example of an object existence range. Here, as shown in the figure, the object on the area B of the camera i is set to a distance a from the center, and the distance a is set to the next camera (i +
Assuming a distance b from the center of the object on the area A in 1), b and Li + 1 can be obtained as shown in (Equation 3).

Next, a procedure for obtaining the distance L will be described with reference to FIGS. 6 (c) and 6 (d). Step 1: Scan the region B of the camera i in FIG. 6D upward from the center to obtain a point P at which an object is detected first (for the illustrated image (labeling result), , And a point P (pixel position) at which each label is detected first is determined).

Step 2: The distance a from the point P obtained in step 1 to the center of the image is obtained. Step 3: The range of the distance (a-1) and the distance (a + 1) of the pixels before and after the distance a obtained in step 2 is determined as the existence range of the object (point P at which the object is detected first).

Step 4: With respect to the distances (a-1) to (a + 1) determined in step 3, the distance (b-1) from the center in the area A of the next camera (i + 1) is substituted into (Equation 3). ) To (b + 1) as the search range.

Step 5: Scan upward from a distance (b + 1) from the search range obtained in Step 4,
First, the position of the point where the object is detected is obtained. The distance L between the obtained point and the center is substituted into the above-described (Equation 1) to calculate the distance L.

As described above, the object is detected in the region B and its existence range is detected, and the position of the object is scanned by scanning the corresponding existence range of the region A in the image of the next processing module according to (Equation 3). By calculating accurately,
High-speed processing can be performed by narrowing the scanning range at the time of object position detection.

[0050]

As described above, according to the present invention,
When monitoring the position of an object on the road surface with a plurality of linked cameras, the presence or absence of the object is detected over a wide area in an area with low distance accuracy (area B) in the image of the camera, and the cooperation is performed when the object is detected. Since the camera adopts a configuration that accurately measures the position of the object in the area (area A) where the distance accuracy of other cameras is high, the position of the object left on the road surface can be widely Can be measured at high speed and with high accuracy.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating the operation of the present invention.

FIG. 2 is an explanatory view (No. 1) of the present invention.

FIG. 3 is an explanatory view (No. 2) of the present invention.

FIG. 4 is an explanatory view (No. 3) of the present invention.

FIG. 5 is an explanatory view (No. 4) of the present invention.

FIG. 6 is an explanatory view (No. 5) of the present invention.

FIG. 7 is an explanatory diagram (part 1) of a conventional technique.

FIG. 8 is an explanatory diagram (part 2) of a conventional technique.

[Explanation of symbols]

 11: Camera 12: Image 13: Processing Module 21: Background Difference Calculation Means 22: Labeling Means 23: Area Calculation Means 24: Position Calculation Means Area A: Area with Low Distance Accuracy Area B: Area with High Distance Accuracy

Claims (4)

[Claims] In a distance measuring system for measuring a position of an object on a road, a region having a low distance accuracy in an image of a certain camera and a region having a high distance accuracy in an image of a next camera are associated with each other so as to overlap with each other. A plurality of cameras sequentially arranged along a road, and the presence or absence of an object on the road in an area with low distance accuracy in an image of one of the plurality of cameras, and an image of a corresponding camera when detected. Means for measuring the position of the object in a region with high distance accuracy in the middle. 2. Each of the cameras repeats detecting an object on a road in an area with a low distance accuracy in an image of each of the cameras,
2. The distance measuring system according to claim 1, wherein the position of the object on the road is measured with high accuracy in a region with high distance accuracy in the image of the camera corresponding to the detected position. 3. A method for detecting an object on a road in an area with a low distance accuracy in an image of each of the cameras and obtaining an approximate position, and obtaining an approximate position in the area with a high distance accuracy in an image of a corresponding camera. 2. The distance measurement system according to claim 1, wherein the position of the object on the road is measured with high accuracy and high speed while being limited to a predetermined range including the position. 4. A computer is operated and sequentially arranged along a road in such a manner that an area having a low distance accuracy in an image of a certain camera and an area having a high distance accuracy in an image of the next camera are overlapped with each other. Means for capturing images from these cameras respectively, and detecting the presence or absence of an object placed on the road in an area with low distance accuracy in the image of one of the plurality of cameras, and detecting the corresponding camera when detected. A recording medium storing a program for functioning as a means for measuring the position of the object in an area with high distance accuracy in an image.