JPH08249588A - Vehicle sensor - Google Patents

Abstract

PURPOSE: To easily detect the body of a vehicle even at the time when it is dark such as at night by providing a high-S/N image suppressing noises by performing averaging processing to an image including that of a stopped vehicle. CONSTITUTION: An ITV camera unit 1 extracts and processes a processing object image in a prescribed cycle. An image input part 22 samples the output image signal of the camera unit 1, performs A/D conversion to that signal and stores it in a storage part 23. A stop judgement/integration part 24 judges the top end part of vehicle based on the stored digital image. Concerning the next processing image, the head of car is judged by similar processing as well. When the state of settling the first judged car head position and the car head position projected in the next processed image within the difference of ± one picture element is continued three times, it is judged that vehicle is stopped. When the vehicle is judged to be stopped, a fixed area inside a lane behind this car head is defined as an integration area. Next, averaging processing is performed to the part of overlapped integration areas concerning the plural images to be processed. Thus, the stopped vehicle can be sharply projected.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a vehicle detection device capable of detecting each vehicle existing in a certain area of a road and measuring the number of vehicles, the occupation rate and the length of congestion. is there.

[0002]

2. Description of the Related Art There are two types of vehicle detection devices, namely, a cross-section image processing type vehicle detection device and a space type image processing type vehicle detection device, for measuring traffic flow by image processing of an image of an ITV (Industrial Television) camera. Both of them have installed an ITV camera diagonally downward from a high place to detect vehicles existing on a certain range of roads, but the sectional image processing type vehicle detection device mainly measures the number of vehicles. 20m to do
While detecting a vehicle that passes a relatively narrow range of degree,
Since the spatial image processing type vehicle detection device measures the congestion length as well as the number of vehicles, it detects vehicles existing on a wide road of about 150 m.

The spatial image processing type vehicle detection device has an advantage over the sectional type image processing type vehicle detection device, which is already in a practical stage, in that it can grasp a wide range of traffic conditions in real time and finely. However, it will be installed at intersections and used widely. By the way, a spatial image processing type vehicle detection device generally measures the presence or absence of a vehicle and the number of vehicles by mainly detecting the contour of the vehicle (hereinafter referred to as “vehicle body”) in the daytime, but it is difficult to recognize the vehicle body at night. Therefore, the presence or absence of vehicles and the number of vehicles are mainly measured with the headlights as headlamps.

[0004]

Since the headlight is bright, it is easy to detect it. However, the size and length of the vehicle cannot be directly measured only by detecting the headlight, and it is estimated from the distance between the headlights. Statistical vehicle size,
You can only determine the length. In addition, in the dusk hours before and after sunset and before and after sunrise, some vehicles have headlights and some have no lights, but the bodies of the no lights cannot be detected, and the number of vehicles is missed. .

Further, since the parked vehicle at night is lightless, it cannot be detected. After all, even at night, the headlight detection alone is not enough to accurately detect the vehicle, and it is desirable that the vehicle body can be accurately detected. However, unlike the daytime, the illuminance is insufficient at night, so that the image signal of the vehicle body is weak and the noise is relatively strong, which makes it difficult to read the vehicle body accurately.

The present invention solves the above-mentioned technical problems in a vehicle detection device for determining a vehicle based on an image taken by a camera for taking an image of a road, and a vehicle existing in a certain area of the road even at night. An object of the present invention is to provide a vehicle detection device capable of accurately detecting each vehicle.

[0007]

[Means and Actions for Solving the Problems]

(1) The vehicle detection device according to claim 1 for achieving the above object, a stop detection unit that detects a stop of the vehicle based on a plurality of image information with different imaging times, and is detected as being stopped. In addition, the integration unit that limits the region including the vehicle and performs an averaging process on a plurality of images captured thereafter is provided.

According to the above arrangement, first, the stop of the vehicle can be detected on the basis of a plurality of pieces of image information having different image pickup times. This stop can be done by turning on the headlights, side lights (called parking lights), and fog lights to track the running vehicle, but even for unlit vehicles, the dark outline of the vehicle can be used. It can be done as a clue. This is because it is relatively easy to determine whether or not the vehicle is moving as compared to the detection of the vehicle body.

When it is detected that the vehicle is stopped, the area including the vehicle is limited. Then, the averaging process is performed on a plurality of images captured thereafter. By this processing, it is possible to obtain an image with excellent S / N in which noise is suppressed, so that the vehicle body can be easily detected. (2) The vehicle detection device according to the second aspect stops the averaging process by the integrating means in the area when the vehicle detected to be stopped starts.

This is because when the vehicle starts to move, the image is blurred when the averaging process is performed, and it becomes difficult to detect the vehicle body.

[0011]

Embodiments will be described in detail below with reference to the accompanying drawings showing embodiments. Figure 2 is a conceptual diagram of the ITV camera installation. (A)
Shows a plan view and (b) shows an elevation view. The ITV camera unit 1 is provided on an upper portion of a pole installed at an intersection of a four-lane road on one side, and a control unit 2 is provided on a lower portion of the pole. The ITV camera is about 150m on a 4-lane road
Is monitoring an area of length.

FIG. 1 is a block diagram showing an electrical configuration of the control unit 2. The control unit 2 inputs an image input unit 22 for inputting an image signal acquired from the ITV camera unit 1 for imaging a road, and an input. The storage unit 23 that temporarily stores the image and the stop of the vehicle are determined based on the stored plurality of pieces of image information having different processing times, and the area including the vehicle that is determined to be stopped is limited, and thereafter, A stop determination / integration unit 24 that performs an averaging process on a plurality of images to be processed, and a vehicle detection unit 25 that detects a vehicle based on the image that has been subjected to the averaging process and the image that has not been subjected to the averaging process. And have.

The ITV camera unit 1 has 5 frames as shown in FIG.
Images to be processed (hereinafter simply referred to as "processed images") are extracted and processed once per frame. This period of 5 frames is called a processing cycle. The image input unit 22 samples the output image signal of the camera unit 1 and
The storage unit 23 stores the digital image data that has been A / D converted in the video RAM. The sampling is performed on points corresponding to the meshes of a grid arranged on the road like a grid, with the crossing direction of the road and the traveling direction of the vehicle as coordinate axes.

The stop determination / integration unit 24 determines the head of the vehicle (referred to as the vehicle head) based on the stored digital image. The procedure for determining the vehicle head is as follows. The digital image is two-dimensionally differentiated to detect a portion (edge) where the density of the image changes. Then, it is binarized based on a predetermined threshold value. The binarization makes the outline of the vehicle or headlight clear. The binarized signal is stored as a preprocessed image in a temporary memory (not shown), and the data stored in the temporary memory forms a profile (vertical projection view) along the vehicle traveling direction. This profile is scanned from below, and the position where the number of added pixels exceeds the threshold value is determined to be the vehicle head.

The same processing is performed on the next processed image,
Determine the car head. And the head position determined previously,
It is checked whether or not the difference between the vehicle head position shown in the next processed image and the vehicle head position is within ± 1 pixel. The reason for seeing the margin of ± 1 pixel is because the imaging error is taken into consideration. The vehicle head position previously determined and the vehicle head position shown in the next processed image are ± 1.
If the state within the pixel difference continues three times, it is determined that the vehicle is stopped. The reason for making three consecutive determinations is to prevent erroneous determinations.

When a stopped vehicle is determined, a certain area in the lane behind the vehicle head is set as an integration area. The vehicle position previously determined and the vehicle position shown in the next processed image are ±
If the state where it does not fit within the difference of 1 pixel continues 3 times,
It is determined that the vehicle has started to move, and the set integration area is released.

FIG. 4 illustrates the above-described vehicle stop determination and integration area setting procedure. If all three vehicles are moving first (a) and the leading vehicle stops, set a certain area A in the lane behind the vehicle (for example, 10 m).
(b). If the second vehicle stops, a certain area B behind the vehicle head is set, but since it overlaps with the area A previously set,
The overlapped area is the logical sum A + B (c). If the third vehicle also stops, a certain area C behind the vehicle head is set, but the area A + B overlaps with the area A + B previously set, so the area A
It becomes + B + C (d).

When a plurality of vehicles are stopped, as a general rule, the vehicle starts to move from the leading vehicle. Therefore, it is necessary to pay attention only to the head of the leading vehicle (that is, the tip of the area A) to judge whether or not the vehicles have started to move. Good. Therefore, the region to be the movement determination target of the vehicle may be the tip of the region A.
The processing load is reduced. Area A if the first vehicle starts moving
Cancel (e).

From this point of time, the area which is the object of movement determination of the vehicle is moved to the vicinity of the head of the second vehicle (the tip of the area B). If the second vehicle starts moving, release area B (f)
. From this point, the area that is the target of vehicle movement determination is
Move to the vicinity of the head of the second vehicle (the tip of area C) and
When the second vehicle starts moving, the area C is released (g). next,
The averaging process is performed on the portions of the plurality of images to be processed in which the integration regions overlap.

This process is a process in which the values of the same pixel in a plurality of images are added and divided by the number of pixels to average them. Note that the pixels other than the integration region are not subjected to the averaging process because the image is blurred when the averaging process is performed. The averaging may be performed between two consecutive processing cycles, or may be performed between three or more consecutive processing cycles. However, when the vehicle starts moving and the integration area is released, the integration area thereafter is not subject to averaging.

As described above, even if the vehicle body cannot be clearly recognized in the normal image, the vehicle body of the stopped vehicle can be clearly displayed by performing the integration. The signal of the image subjected to the averaging processing in the integration area is the vehicle detection unit 2
5 is input. In the vehicle detection unit 25, as is conventionally known, the image is edge-emphasized by performing a differential operation on the image, the result is binarized, and the image obtained by binarization is compared with the vehicle body. A mask of a similar size is applied, and vehicle body candidates are extracted based on the number of edge pixels present in the mask. Then, the stopped vehicle is discriminated by judging the characteristic amount such as the vehicle width, the vehicle length, the headlight interval, etc. based on the extracted profile of the vehicle body candidate (Japanese Patent Laid-Open No. Hei 5).
(See JP-A-307695). Further, the signal of the image outside the integration area, which is not subjected to the averaging process, is directly input from the storage unit 23 to the vehicle detection unit 25, and the moving vehicle is determined using the same method.

In this way, the number of vehicles existing in the road range and the length of congestion can be detected. These results are transmitted to the host computer through a communication line (not shown). The present invention is not limited to the above embodiment. In the above-described embodiment, 1/30 seconds is set as one frame, the processing cycle is set for every 5 frames, and the processed image is the object of integration. However, the integration target may be each image captured for each frame. In this case, as shown in FIG. 5, while determining the stop of the vehicle and setting the integration area on the basis of the processed image, the images of the following five frames are integrated in the integration area. In this case, the processed image and the integrated image are obtained in the same cycle, so that the vehicle detection process can be speeded up, but since it is necessary to add the images in real time, dedicated hardware is required.

[0023]

As described above, according to the vehicle detection apparatus of the present invention, an image including a stopped vehicle is averaged to obtain an image with excellent S / N in which noise is suppressed. Therefore, it is possible to easily detect the vehicle body even in a dark time such as at night. In particular, according to the vehicle detection device of the second aspect, when the vehicle detected to be stationary starts, the averaging process by the integration means of the area is stopped, so that the image is prevented from being blurred. You can

[Brief description of drawings]

FIG. 1 is a block diagram showing an electrical configuration of a vehicle detection device.

2A and 2B are conceptual views of installing an ITV camera on a road, where FIG. 2A is a plan view and FIG. 2B is an elevation view.

[Fig. 3] Fig. 3 is a diagram illustrating a configuration example of an imaging frame of an ITV camera unit.

FIG. 4 is an illustrative view of a vehicle stop determination and integration region setting procedure.

FIG. 5 is a diagram showing another example of the configuration of an imaging frame of an ITV camera unit.

[Explanation of symbols]

 1 ITV camera unit 2 control unit 22 image input unit 23 storage unit 24 stop determination and integration unit 25 vehicle detection unit

Claims (2)

[Claims] 1. A vehicle detection device for determining a vehicle based on an image captured by a camera for capturing a road, and a stop determination means for determining stop of the vehicle based on a plurality of image information having different image capturing times. Limited to the area that includes the vehicle determined to be
A vehicle detection device comprising: an integration unit that performs an averaging process on a plurality of images captured thereafter. 2. The vehicle detection device according to claim 1, wherein when the vehicle determined to be stopped starts, the averaging process by the integrating means in the area is stopped.