JPH0981889A - Vehicle detector - Google Patents

Abstract

PROBLEM TO BE SOLVED: To recognize a difference in each monitor camera setting place, automatically selecting a picture processing program optimum to the setting place to execute picture recognition processing and to easily, accurately and surely set up a monitor camera. SOLUTION: The feature point of an advancing vehicle photographed by a monitor camera 10 is extracted by a feature extracting part 12. A binary picture signal indicating the feature point is inputted to a camera position recognizing part 13, the appearance inclination of feature points (vehicle side lamps and tail lamps) of plural vehicles passing on a road is analyzed on a time base to judge the photographing position of the current monitor camera 10. For instance, a state that the camera 10 photographs an advancing vehicle from the upper part of the road and the rear part of the vehicle is recognized, a program switching control part 14a reads out a picture processing program matched with the photographing position of the camera 10 from a memory 14b and the vehicle is detected based upon the newly selected picture processing program.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is used in a traffic information system or the like for providing road congestion information, a vehicle passing on a road is photographed by a surveillance camera (video camera), and the image data is used for vehicle tracking. The present invention relates to a vehicle detection device that selects an optimum image processing program.

[0002]

2. Description of the Related Art Conventionally, in a traffic information system or the like, traffic congestion information is provided, for example, by recognizing the number and types of vehicles passing on the road from image data taken by a surveillance camera on the road. In this case, the analog image pickup signal taken by the surveillance camera is taken in every frame, and the image data obtained by digitizing the analog image pickup signal is processed,
The characteristic points of the vehicle are detected. The vehicle is recognized by comparing these characteristic points with the parameter conditions stored in advance.

The characteristic point of this vehicle is, for example, a comparison between image data obtained by photographing a road where the vehicle does not exist in the daytime and image data obtained when the vehicle passes, and the difference is compared with a parameter condition stored in advance. Car recognition. In the case of nighttime, the feature point obtained by extracting the high-intensity part of the lighting tail lamp is compared with the parameter condition stored in advance to recognize the vehicle.

[0004]

However, in the above-described conventional vehicle detection device, it is necessary to match the image processing program (vehicle recognition algorithm) for each shooting state (installation state) of the surveillance camera. If the shooting state (installation state) and the image processing program do not match well, the recognition rate deteriorates or a stable recognition state cannot be obtained. In other words, it is necessary to create the image processing program according to the shooting location and strictly install (set) the surveillance camera in the shooting status (installation status) matching the image processing program. In this case, the work of initial installation of the surveillance camera is troublesome.

When the installation position (shooting direction) changes after the initial installation of the surveillance camera, for example, the road surface changes due to road construction or the installation position of the surveillance camera is changed.
If it changes due to vibration or the like, the image processing program used cannot perform accurate image processing. That is,
In the conventional vehicle detection device, it is necessary to constantly manage the shooting state of the surveillance camera, in other words, it is difficult to always detect the vehicle with high accuracy.

The present invention solves the above-mentioned drawbacks of the prior art, recognizes the difference in the installation location (imaging direction) of the surveillance camera (imaging means), and selects the optimum image processing program for this installation location. Can be automatically selected, and the initial installation work of the surveillance camera can be performed easily and reliably, and an appropriate image processing program can be used even when this installation state (shooting direction) changes after the surveillance camera is installed. It is an object of the present invention to provide a vehicle detection device capable of easily and surely detecting various types of vehicles on various roads by automatically selecting.

[0007]

In order to achieve the above object, a vehicle detection device according to a first aspect of the present invention extracts a characteristic point of a vehicle from a photographing means for photographing the road and image data photographed by the photographing means. Extraction means, feature analysis means for analyzing the characteristics of the vehicle from the appearance tendency of the feature points extracted by the extraction means in association with the movement of the vehicle, and shooting position indicating the shooting position of the shooting means from the characteristics of the vehicle analyzed by the feature analysis means. Based on the photographing position data sent from the photographing position recognition means, the storage means for storing a plurality of image processing programs, and the photographing position recognition means, the image processing program suitable for the photographing position is selected from the storage means. And a selection means.

According to another aspect of the vehicle detection device of the present invention, when the width of one feature point satisfies a preset width, the vehicle extraction means extracts the feature point as a feature point of the vehicle and also horizontally. When the interval between a pair of feature points existing in the direction satisfies a predetermined interval, this feature point is extracted as a vehicle, or the width of one feature point is a preset width, and,
The distance between the pair of feature points existing in the horizontal direction is a preset interval, and when the time difference between the one feature point and the pair of feature points existing in the horizontal direction is a predetermined time, these feature points are set. The feature points of one vehicle are extracted.

According to another aspect of the vehicle detection device of the present invention, when the photographing means is performing black-and-white photographing, the vehicle extracting means extracts a high brightness portion existing in the horizontal direction as a characteristic point corresponding to a lighting lamp of the vehicle. Or, the image taken only on the road is subtracted from the image when the vehicle is present, the difference data of the image conforms to the parameter condition stored in advance, and the horizontal and bottom Both side points are extracted as feature points, or when the image capturing means is performing color image capturing, the same color points existing on the both sides in the horizontal direction and on the lowermost sides are extracted as feature points.

According to a fourth aspect of the present invention, there is provided the vehicle detection device, wherein the characteristic analysis means causes the right characteristic point of the vehicle to pass a first time axis and the left characteristic point of the vehicle passes a second time axis. As a configuration for analyzing the feature of the vehicle from the appearance time of the feature point according to any one of the time when the pair of feature points on both sides of the rearmost part of the vehicle pass the third time axis or the combined time. There is.

According to a fifth aspect of the present invention, there is provided a vehicle detection device in which the vehicle detection device having the configuration according to the first to fourth aspects is used in at least a traffic information system which provides road congestion information.

The vehicle detection device having the above-mentioned structure according to the first aspect of the present invention extracts the characteristic points of the vehicle from the photographed image data and determines the appearance tendency of the characteristic points (the appearance tendency of the characteristic points due to the movement of the vehicle). Analyze the characteristics of the vehicle. The photographing position data which recognizes the photographing position of the photographing means is obtained from the characteristics of the vehicle. Further, based on the shooting position data, an image processing program suitable for the shooting position is selected.

As a result, the difference in the installation location (imaging direction) of the imaging means (monitoring camera) is recognized, and the image recognition program can be automatically selected and the image recognition processing optimum for this installation location can be performed. That is, the recognition rate improves without being affected by the difference in the installation location of the surveillance camera,
In addition, stable recognition characteristics can be obtained. That is, the work of initial installation of the surveillance camera becomes easy, accurate, and reliable. Further, even after the installation of the surveillance camera, even when the installation state (imaging direction) changes, an appropriate image processing program is automatically selected, and various vehicles on various roads can be detected easily and reliably. .

According to a second aspect of the vehicle detection device of the present invention, when one feature point has a predetermined width, and when a pair of horizontally existing feature points has a predetermined interval, these feature points are separated. Is extracted as a feature point of the vehicle. When the width is a predetermined width, the interval is a predetermined interval, and the time difference between one feature point and a pair of feature points is a predetermined time, this feature point is set as a feature point of one vehicle. Extracting. Therefore, detection of various vehicles on various roads can be performed easily, accurately and reliably.

In the vehicle detection device according to the third aspect, the vehicle extraction means extracts a high-intensity portion existing in the horizontal direction as a feature point corresponding to the lighting lamp of the vehicle when performing black-and-white photographing, or , The image taken only on the road is subtracted from the image when the vehicle is present, and the subtracted data conforms to the parameter conditions stored in advance, and the horizontal and bottom side points are extracted as feature points. are doing. Alternatively, in the case of color photography, the same color points on both sides in the horizontal direction and at the bottom are extracted as feature points. Therefore, the detection of various vehicles on various roads
Easy, accurate and reliable.

According to a fourth aspect of the present invention, there is provided a vehicle detection device in which the characteristic point on the right side of the vehicle passes through the first time axis, the characteristic point on the left side of the vehicle passes through the second time axis, and From the appearance tendency of the feature points (the appearance tendency of the feature points due to the movement of the vehicle) at any one of the times when the pair of feature points on both sides of the rearmost portion passes through the third time axis or the time obtained by the combination thereof,
The characteristics of the vehicle are analyzed. Therefore, detection of various vehicles on various roads is accurately and reliably performed.

According to a fifth aspect of the present invention, there is provided a vehicle detection device which uses the vehicle detection device having the configuration according to any of the first to fourth aspects of the present invention for a traffic information system which provides at least road congestion information. Will be obtained.

[0018]

BEST MODE FOR CARRYING OUT THE INVENTION Next, an embodiment of a vehicle detection device of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing a configuration of an embodiment of a vehicle detection device of the present invention. In FIG. 1, this example is provided with a surveillance camera (video camera) 10 for photographing the road and an A / D converter 11 for converting an analog image pickup signal from the surveillance camera 10 into a digital image signal. Further, a feature extraction unit 12 that extracts the feature points of the vehicle from the image pickup, and a camera position recognition unit 13 that recognizes the shooting position (shooting direction) of the surveillance camera 10 from the feature points extracted by the feature extraction unit 12 are provided. There is.

Further, based on the shooting position (shooting direction) of the surveillance camera 10 recognized by the camera position recognizing unit 13, switching is performed to select an image processing program capable of optimal vehicle recognition processing at the installation position. A program switching control unit 14a and a memory 14b that stores a plurality of image processing programs in advance are provided. Further, based on the image processing program from the program switching control unit 14a, the vehicle recognition unit 15 for analyzing the characteristic points of the vehicle input from the characteristic extraction unit 12 and recognizing the presence of the vehicle.
And a D / A conversion unit 16 for converting the recognition data obtained by the vehicle recognition unit 15 into an analog signal. Also,
A printing unit 17 that prints and outputs the analog recognition data from the D / A conversion unit 16 and a display unit 18 that displays the analog recognition data from the D / A conversion unit 16 on the screen are provided.

The feature extraction unit 12 to the vehicle recognition unit 15
Up to this point, it is generally composed of a microcomputer, a digital signal processor (DSP) and the like.

FIG. 2 is a diagram showing a photographing state of the surveillance camera 10, and FIG. 3 is a diagram showing a photographed image of FIG. FIG.
Also, in FIG. 3, the monitoring camera 10 is installed so as to be photographed from above the road 22 and behind the traveling vehicle 23. In this case, the right lane shoulder 3
1, it has a center line 32 and a left lane shoulder 33.

FIG. 4 is a view showing a photographing state when the surveillance camera 10 is installed in a place different from the example of FIG. 2, and FIG.
It is a figure which shows the picked-up image of FIG. 4 and 5, the surveillance camera 10 in this case is installed so as to be photographed obliquely above the road 22 and from behind the traveling vehicle 23. In this case, the road 22 is imaged on the right lane shoulder 31 when photographed from diagonally above and behind the traveling vehicle 23.
It has a center line 32 and a left lane shoulder 33.
Hereinafter, description will be made with reference to these two shooting states.

Next, the operation of this embodiment will be described. Monitoring camera 10 installed as shown in FIG. 2 or FIG.
Takes a picture of the road 22. In this case, when the surveillance camera 10 takes an image from above the road 22 and from behind the traveling vehicle 23 as shown in FIG. 2, the right lane shoulder 31 and the right lane shoulder 31 as shown in FIG. The left lane shoulder 33 is distributed, and the road 22 near the shooting position is wide, and the road 22 far from the shooting position is narrow.

Further, as shown in FIG.
However, when the image is taken obliquely above the road 22 and from behind the traveling vehicle 23, the entire road 22 is projected obliquely as shown in FIG. 5, and the right lane shoulder 31 and The left lane shoulder 33 is sorted, and the width of the road 22 near the shooting position is wide, and
An image is taken in which the width of the road 22 whose shooting position is far is narrow.

An analog image pickup signal for each frame from the monitoring camera 10 is converted into a digital image signal by the A / D converter 11 and input to the feature extraction unit 12. The feature extraction unit 12 extracts the feature points of the vehicle. As a method of extracting the feature points of the vehicle, there is a method of extracting high-intensity feature points in a monochrome image. For example, the light source of the tail lamp that is turned on at night is extracted as the feature point. Further, there is a background difference detection method as a method of detecting feature points in the daytime.

In this background difference detection method, the outline of the vehicle is extracted by subtracting the image in which the captured image is changed from the captured image only on the road in the image captured frame by frame, that is, the image when the vehicle is present. However, the points on both sides (for example, the position of the tail lamp) are extracted as feature points. Further, the color of the tail lamp or the like in the color image may be identified and the characteristic point thereof may be detected.

It is necessary to determine that this feature point is the feature point of the vehicle. For example, when the width of one feature point satisfies the preset width, this feature point is extracted as the feature point of the vehicle. To do. Further, when the interval between a pair of feature points existing in the horizontal direction satisfies a predetermined interval, this feature point is extracted as a vehicle. Further, the width of one feature point is a preset width, and the interval of a pair of feature points existing in the horizontal direction is a preset interval, and the one feature point and a pair existing in the horizontal direction are When the time difference from the feature points of 1 is a predetermined time, these feature points are extracted as the feature points of one vehicle.

An example in which a high brightness point of a lit tail lamp in a monochrome image is detected as a characteristic point will be described below. FIG. 6 is a flowchart showing a processing procedure for analyzing the feature points and recognizing the shooting position of the surveillance camera 10.
FIG. 7 is a diagram showing characteristic points of the traveling vehicle 23 taken by the monitoring camera 10 in the arrangement state shown in FIG. Also, FIG.
[Fig. 5] is a diagram showing characteristic points of the traveling vehicle 23 photographed by the monitoring camera 10 in the arrangement state shown in Fig. 4.

In FIGS. 1 to 8, the binary image signal obtained by extracting the characteristic points of the vehicle is input to the camera position recognizing unit 13. The camera position recognition unit 13 analyzes the characteristic points (vehicle side lamps and tail lamps) of each of a number of vehicles 23 passing on the road 22 shown in FIGS. 2 to 5.

First, the surveillance camera 1 in the arrangement state shown in FIG.
When shooting at 0, as shown in FIG.
3 analyzes the application tendency of the left-side lamp (the tendency of appearance of the feature points due to the movement of the vehicle) from the feature points of the image data captured by the device 3. In this case, the passage of the lamp (characteristic point) 70a mounted on the left side of the vehicle body of the vehicle 23a passing through one lane (the left side in the drawing) of the road 22 on the time axis T11 is recognized. Further, the passage of the lamps (feature points) 71a and 71b mounted on the left side of the vehicle body of the vehicle 23b passing through the other lane (right side in the figure) of the road 22 on the time axis T14 is recognized (step (S 41).

After that, the application tendency of the right side lamp is similarly analyzed. In this case, the lamps (characteristic points) 72a, 72b, 72c mounted on the right side of the vehicle body of the vehicle 23a
Of the vehicle 23 on the time axis T12 is recognized.
Lamp (characteristic point) 74 mounted on the right side of the vehicle body of b
The passage of a on the time axis T15 is recognized (step 4
2).

Further, the difference between the lamps (characteristic points) on the left and right sides of the vehicle 23a is analyzed. That is, the ramp (characteristic point) 70a on the time axis T11 and the time axis T12 in FIG.
The lamps (feature points) 72a are recognized at intervals on the time axis T13. Similarly, the difference between the lamps (feature points) on the left and right sides of the vehicle 23b is analyzed. That is, the time axis T14 ramp (feature point) 71a and the time axis T15 ramp (feature point) 74a in FIG. 7 are recognized at intervals on the time axis T16 (step 43). Next, the tendency is analyzed from the vehicle lamps (characteristic points) for each lane. For example, when traffic is regulated for each lane, the passing vehicle is analyzed (step 44).

Such steps 41 to 44
Through the process of, the current shooting position of the surveillance camera 10 is determined. In this case, it is recognized that the surveillance camera 10 is shooting from above the road 22 and behind the traveling vehicle 23, as shown in FIG.

Further, the surveillance camera 1 arranged as shown in FIG.
Even in the shooting state of 0, image processing is performed as in the shooting state shown in FIG. Also in this case, the camera position recognition unit 13 takes in the image data shown in FIG. 8 and recognizes the shooting position of the surveillance camera 10 by the processing procedure shown in FIG. That is, in FIG. 8, the left and right side lamps (feature points) 75a, 75b, 75c, 76a of the vehicle 23 are recognized on the time axis T21, 22.
Also, the left and right lamps (characteristic points) 75 on the time axis T23
Recognize the difference between a and 76a.

Also in this case, it is recognized (cleared) from the processing results so far that the monitoring camera 10 is installed so as to be photographed obliquely above the road 22 and behind the traveling vehicle 23.

Based on the current shooting position data of the surveillance camera 10 found in this way, the program switching control section 14a instructs the memory 14a to make a current surveillance from a plurality of image processing programs stored in advance. Camera 10
Read out the image processing program suitable for the shooting position.

Based on the image processing program from the program switching control unit 14a, the vehicle recognition unit 15 analyzes the characteristic points of the vehicle input from the characteristic extraction unit 12 and recognizes the vehicle. The result of this vehicle recognition is printed and output by the printing unit 17, or converted into analog recognition data by the D / A conversion unit 16 and displayed on the screen of the display unit 18.

As described above, in this embodiment, the characteristic point of the vehicle photographed by the surveillance camera 10 is identified, the photographing position (photographing direction) is recognized, and the optimum image processing program at this photographing position is automatically determined. Image selection and the like, and processing such as image recognition is performed. Accordingly, when the surveillance camera 10 is initially installed, it is not necessary to create and use an image processing program suitable for each photographing position (direction) for each photographing position (direction).

Further, the photographing position (direction) of the surveillance camera 10 is adjusted so as to exactly match the created image processing program.
There is no need to set. That is, the surveillance camera 10
The initial installation can be performed so as to be generally suitable for a plurality of image processing programs, the initial installation work can be facilitated, and vehicle detection thereafter can be performed accurately and surely.

After the initial installation of the surveillance camera 10,
If the installation position (imaging direction) changes, for example, the road surface changes due to road construction, or the installation position of the monitoring camera 10 changes due to vibration, etc. The image processing program is switched to a new image processing program suitable for the mounting position of 10. This eliminates the need to constantly manage the shooting state of the surveillance camera 10. In other words, the vehicle can be detected with high accuracy at all times.

[0041]

As is apparent from the above description, the vehicle detection device according to claim 1 analyzes the features of the vehicle from the appearance tendency of the feature points of the vehicle extracted from the photographed image data, and determines the characteristics of the vehicle. The image capturing position of the image capturing means is recognized from the characteristics of the image capturing means, and the image processing program suitable for the image capturing position is selected based on the image capturing position data. Is recognized, and the image recognition processing can be performed by automatically selecting the optimum image processing program for this installation location.

As a result, the work of initial installation of the surveillance camera becomes easy, accurate and reliable. In addition, even if the installation state (shooting direction) changes after the surveillance camera is installed, an appropriate image processing program is automatically selected, making it possible to easily and reliably detect various vehicles on various roads. Has the effect of becoming.

According to a second aspect of the vehicle detection device, when the width of one feature point or the interval between a pair of feature points existing in the horizontal direction is a predetermined width or interval, these feature points are detected. Since it is extracted as a vehicle feature point, and when the time difference between one feature point and a pair of feature points existing in the horizontal direction is a predetermined time, this feature point is extracted as a vehicle feature point. There is an effect that various kinds of vehicles can be easily, accurately and surely detected on various roads.

In the vehicle detection device according to the third aspect of the present invention, when performing black-and-white photographing, the high-intensity portion existing in the horizontal direction corresponding to the lighting lamp is extracted as a characteristic point of the vehicle, or the vehicle is subjected to the background subtraction method. It is extracted as feature points at both sides of the image in the horizontal direction and at the bottom, or when color photography is performed, the same color points at both sides in the horizontal direction and at the bottom of the image are extracted as feature points. This has the effect that various types of vehicles can be detected easily, accurately and reliably on the road.

According to another aspect of the vehicle detection device of the present invention, the time when the right characteristic point of the vehicle passes the first time axis, the time when the left characteristic point of the vehicle passes the second time axis, and the vehicle Since the characteristics of the vehicle are analyzed from the appearance tendency of the feature points at any one of the times when the pair of feature points on the both sides of the rearmost part pass through the third time axis or the combined time, the characteristics of the vehicle on various roads are analyzed. This has an effect that various types of vehicles can be detected accurately and surely.

According to a fifth aspect of the present invention, there is provided a vehicle detection apparatus which uses the vehicle detection apparatus having the configuration according to the first to fourth aspects for a traffic information system which provides at least road traffic congestion information. Information can be obtained.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an embodiment of a vehicle detection device of the present invention.

FIG. 2 is a diagram showing a shooting state of the surveillance camera shown in FIG.

FIG. 3 is a diagram showing an image captured by the surveillance camera shown in FIG.

FIG. 4 is a diagram showing a shooting state according to another installation state of the surveillance camera in the embodiment.

5 is a diagram showing an image captured by the surveillance camera of FIG.

FIG. 6 is a flowchart showing a processing procedure for recognizing a shooting position of a surveillance camera by analyzing feature points in the embodiment.

FIG. 7 is a diagram showing characteristic points of the traveling vehicle taken by the surveillance camera in the arrangement state shown in FIG. 2 in the embodiment.

8 is a diagram showing characteristic points of the traveling vehicle taken by the surveillance camera in the arrangement state shown in FIG. 4 in the embodiment.

[Explanation of symbols]

 10 surveillance camera 12 feature extraction unit 13 camera position recognition unit 14a program switching control unit 14b memory 15 vehicle recognition unit

Claims (5)

[Claims] 1. A photographing means for photographing a road, an extracting means for extracting characteristic points of a vehicle from image data photographed by the photographing means, and a tendency of appearance of the characteristic points extracted by the extracting means as the vehicle moves. A characteristic analysis means for analyzing the characteristic of the vehicle, a photographing position recognition means for transmitting photographing position data indicating a photographing position of the photographing means from the characteristic of the vehicle analyzed by the characteristic analysis means, and a plurality of image processing programs are stored. A vehicle detection apparatus comprising: a storage unit; and a selection unit that selects an image processing program suitable for the shooting position from the storage unit based on the shooting position data from the shooting position recognition unit. 2. The vehicle extracting means extracts a feature point as a feature point of a vehicle when the width of one feature point satisfies a preset width, and a pair of feature points existing in the horizontal direction. If the space between the feature points satisfies a predetermined space, this feature point is extracted as a vehicle, or the width of one feature point is a preset width, and the space between a pair of feature points existing in the horizontal direction is previously set. With the set interval, when the time difference between the one feature point and a pair of feature points existing in the horizontal direction is a predetermined time, these feature points are extracted as feature points of one vehicle. The vehicle detection device according to claim 1. 3. When the photographing means is performing monochrome photographing, the vehicle extracting means extracts a high-intensity portion existing in the horizontal direction as a characteristic point corresponding to the lighting lamp of the vehicle, or photographs only the road. The image is subtracted from the image when the vehicle is present, the difference data of the image is suitable for the parameter conditions stored in advance, and the horizontal and bottom side points are extracted as feature points, Alternatively, the vehicle detection device according to claim 1, wherein when the image capturing means is performing color image capturing, the same color points existing in the horizontal direction and located on both sides of the lowermost portion are extracted as feature points. 4. The feature analysis means is configured such that the right characteristic point of the vehicle passes a first time axis, the left characteristic point of the vehicle passes a second time axis, and the last of the vehicle. The feature of the vehicle is analyzed from the appearance tendency of the feature point at any one of the times when the pair of feature points on both sides of the part pass through the third time axis or the combined time. Vehicle detection device. 5. A vehicle detection device, characterized in that the vehicle detection device having the configuration according to any one of claims 1 to 4 is used in at least a traffic information system that provides road congestion information.