JPH0452999A - Parking/stopping car detecting device - Google Patents

Abstract

PURPOSE:To detect a parking/stopping car without necessitating a lighting device even at night by image-picking up the passing route of a car such as a road, etc., at every definite time, and converting an outputted video signal into digital picture data, and after the completion of the cumulative summation of the picture data, extracting information related to the parking/stopping car. CONSTITUTION:The video signal of one scene portion outputted from a camera 9 is converted into the digital picture data by an A/D converter 11. An adder 13 sums cumulative picture data stored in a picture memory 14 and the picture data given from the A/D converter 11 for every picture element. The picture data of the prescribed number of the scenes are summed by the adder 13, and this summed result is stored as the cumulative picture data in the picture memory 14. Then, at night, the picture of the light spot of a head light or a tail light is cumulatively summed. Thus, only the parking/stopping car can be correctly detected as excluding the car traveling at low speed without being influenced by lightness, etc.

Description

DETAILED DESCRIPTION OF THE INVENTION TECHNICAL FIELD The present invention relates to an apparatus for detecting a vehicle parked on a vehicle passage such as a road by image processing.

Prior art and its problems Image data including parked and running vehicles is obtained by capturing an image of a vehicle passage such as a road from above or diagonally above, and by processing this image data, As a device to detect parked vehicles,
Conventionally, there are methods using frame difference and spatial difference methods. A parking/stop detection device using this image processing method performs frame difference on multiple input images taken with a time difference to extract a running vehicle. The system uses subtraction to emphasize lines and outlines of parked and parked vehicles, and then extracts parked and parked vehicles from the image.

However, such conventional devices require sufficient brightness to extract the lines and contours of a vehicle, and a large-scale illumination device is required to detect the lines and contours of a vehicle at night. Furthermore, since the traveling vehicle is extracted based on the frame difference, the traveling vehicle must change (move) to some extent between the two images. I wanted to,
When driving at low speeds in traffic jams, the amount of change due to frame differences is small, making it difficult to determine whether a vehicle is parked or running.

purpose of invention This invention is not affected by brightness etc.

To provide a device that correctly detects only parked vehicles while excluding vehicles traveling at low speed.

Structure 1 of the Invention Functions and Effects The parking and stopping detection device according to the present invention images a vehicle traffic path such as a road at regular intervals, and includes images of parked and running vehicles on the photographed traffic path. Imaging means for outputting a video signal every time an image is captured; means for converting the video signal outputted from the imaging means into digital image data; and a means for storing the added cumulative image data, and after the cumulative addition of the image data of a predetermined number of scenes is completed, from the cumulative image data stored in the storage means, the vehicle is parked or parked. The vehicle is characterized by comprising image processing means for extracting information regarding the vehicle.

According to the device of this invention, the captured images are cumulatively added. Therefore, during the day, by cumulatively adding many images, moving vehicles are smoothed out as a noise component against the stationary background and parked vehicles, making parked vehicles stand out. Become. Parked or parked vehicles can be detected by removing noise components from such cumulative images and then extracting contour lines, etc., which are characteristics of vehicles, using a spatial differential method or the like.

Also, at night, by adding up the light trails of headlights and tail lights, you can clearly identify the light trails, and if there are parked vehicles that obstruct traffic flow, Parked and parked vehicles can be indirectly detected by changes in light trails.

As described above, according to the present invention, the light trail of the vehicle's lights can be used at night, so there is no need for a lighting device at night. Furthermore, even if a sudden change in weather or the like occurs during the cumulative addition of one image data and the brightness of the screen changes, this can be treated as noise and is not easily affected. Furthermore, even if the moving vehicle is moving at a low speed due to traffic jams or the like, by cumulatively adding a sufficient number of images, the effect will be reduced. Since it is not necessary to continuously process images at high speed, it is possible to perform processing at another location by transmitting one image, and in this case, a low-capacity communication line or the like can be used.

DESCRIPTION OF THE EMBODIMENT FIG. 1 shows the arrangement of an embodiment for detecting vehicles parked on a road.

A television (video) camera 9 is installed on the top of a pillar erected on the side of the road. The camera 9 overlooks the road and images a predetermined range of the road.

Outputs a video signal representing the captured image. The camera 9 images the road, vehicles parked on the road, and vehicles running on the road. The video signal output from camera 9 is input to image processing device IO. Supports may be erected on both sides of the road, side branches may be placed on top of these supports, and the camera 9 may be installed on the side branches. Furthermore, the camera 9 can be attached to an overpass or other structure to capture an image of the road from above.

FIG. 2 shows the electrical configuration of one image processing device 10. As shown in FIG.

1 scene (] frame or 1 scene output from camera 9
The video signal for each field is converted into digital image data by an A/D converter 11, and is supplied to an adder 13 via a bit conversion circuit 12 if necessary. The adder 13 adds the cumulative image data stored in the image memory 14 and the A/
The image data provided from the D converter J1 is added for each pixel. This added image data is stored in image memory 1.
4 is stored. The image data of the first scene is stored in the image memory 14 without being added. In this way, image data for a predetermined number of scenes are added by the adder 11, and the addition result is stored in the image memory 14 as cumulative image data. The cumulative number of scenes is the gradation (number of bits) of one image data and the capacity of image memory (number of data bits)
It can be determined depending on the parked vehicle extraction process. For example, if the image memory is 18 bits and the gradation of the image data is 8 bits, then 8 bits, or 2
Cumulative addition of 56 scenes is possible. If more cumulative addition is necessary for extracting parked vehicles due to extreme traffic jams, the conversion circuit 12 converts 8-bit image data into 7-bit image data, making it possible to cumulatively add 512 scenes.

After the required number of scenes of image data are cumulatively added to the image memory 14, the CPU 15 uses the cumulatively added image data to extract information regarding parked and parked vehicles according to a predetermined image processing algorithm.

An example of an image processing algorithm for extracting parked vehicles that can be suitably used during the daytime is shown in FIG.

One scene consists of 512 x 512 pixels vertically and horizontally, and image data in which each pixel is represented by 8 bits (258 gradations) is cumulatively added for 256 scenes. 1/15 per scene
If it takes seconds, it will take approximately 17 seconds to accumulate 256 scenes of cumulative image data (512 x 512 x 16 bits).
is obtained. An image of a running (moving) vehicle is stored as if it were touched by a camera, and if accumulated over a long period of time, it becomes a noise component. On the other hand, the images of the background and the parked vehicle are strengthened by addition and appear clearly.

In order to remove noise components such as accumulated data of images of moving vehicles, image data of 16 bits for each pixel is converted to 8 bits.
Shift and discard the lower 8 bits.

Only the top 8 bits are left. Image data is 512 x 512
X 8 bits.

In order to remove the background, calculations are performed using a measurement area mask. As a result, only images of parked vehicles on the road remain. Furthermore, a spatial differential calculation or the like is performed on the image data obtained in this manner to extract feature quantities such as the outline of a car, and parked or parked vehicles are detected based on this feature quantity.

FIG. 4 is for explaining an image processing algorithm for extracting parked and parked vehicles that can be suitably used at night.

The outline of this nighttime algorithm is as follows.

The head and tail can be used as clues to check the vehicle at night.
There are lights and tail lights. By cumulatively adding up images of these light spots, the progress status of the traveling vehicle is stored as a light trail.

By cumulatively adding up over a long period of time.

It is possible to understand the traffic flow situation as a group. In other words, when there are no parked or parked vehicles that obstruct traffic flow, a light trail along each lane is obtained, and the brightness of the image increases in that area. On the other hand, if there is an obstacle that obstructs the traffic flow (such as a parked vehicle), the traffic flow changes in a curved manner to avoid the obstacle, and the brightness of the image near the obstacle changes. By detecting this, parked vehicles can be indirectly detected.

Referring to FIG. 4, 16-bit cumulative image data is obtained by cumulatively adding 256 scenes of 8-bit image data. By converting the image data of each scene into fewer bits, more scenes can be added. For example, by extracting tail lights (or head lights) from the image data of each scene, each scene can be expressed with 2 bits. In this case, in order to obtain 16-bit cumulative addition image data, 855
It is possible to add 36 scenes. Assuming that it takes 1/15 seconds to measure one scene, it takes about 17 seconds to add 256 scenes and obtain 16-bit cumulative image data from 8-bit image data, and it takes about 17 seconds to add 6553B scenes and obtain 2-bit image data from 2-bit image data. It takes approximately 73 minutes to obtain 16-bit cumulative image data.

By appropriately selecting the bit conversion, the number of scenes to be added can be optimally set in consideration of the total measurement time.

Next, right shift processing of a predetermined number of bits is performed on the cumulative image data according to the number of cumulatively added scenes or time. to obtain 8-bit image data. This image data clearly shows the light trails caused by the vehicle's tail lights or head lights.

Next, the background is removed using a measurement area mask. Finally, set up small areas on the road along the lanes (indicated by solid lines) and check the brightness within each small area. The light trail curves in areas where there are parked vehicles, so the brightness of the small area (indicated by A) set in that area is relatively low, and there are no parking spaces in this small area A that impede traffic flow. It can be seen that there is a parked vehicle.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing an arrangement when the present invention is applied to detecting parking and stopping on a road. FIG. 2 is a block diagram showing the electrical configuration of the image processing device. FIG. 3 is a diagram for explaining an algorithm suitable for detecting parking during the daytime, and FIG. 4 is a diagram for explaining an algorithm suitable for detecting parking during nighttime. 9...TV (video) camera. 10... Image processing device 11... A/D converter. 13... Adder 14... Image memory I5... CPU. that's all

Claims (1)

[Scope of Claims] A vehicle traffic route such as a road is imaged at regular intervals, and a video signal representing an image including images of parked and running vehicles on the imaged traffic route is output for each image capture. a means for converting the video signal output from the imaging means into digital image data; a means for adding the converted digital image data between corresponding pixels for each scene imaging; and adding the added cumulative image data. After the storage means and the cumulative addition of image data of a predetermined number of scenes are completed,
A parked/stopped detection device comprising: image processing means for extracting information regarding parked vehicles from cumulative image data stored in the storage means.