JPH0883390A - Vehicle recognizing device - Google Patents

Abstract

PURPOSE: To effectively utilize prescribed vehicle recognizing performance without fail by improving position calculation accuracy at the number plate part of a vehicle by making the horizontal part of the vehicle, of which the image is picked up by an image pickup means, horizontal in the picked up image on the stage before the extraction of a vertical edge component. CONSTITUTION: The image of a running vehicle (c) is picked up from the obliquely upper side by a camera 1 installed at the side part of a passage 2 and that picked up image is inputted to an image processor. Therefore, since the horizontal part of the real vehicle (c) is made horizontal in the picked up image although the image picked up by the camera 1 is obliquely deformed as a whole, the horizontal part is matched with horizontal features in the case of picking up the image of the vehicle (c) from the front upper side of a running direction A. Besides, the distribution of plate numbers on the number plate is not vertically deviated but horizontally distributed. Thus, since the vertical edge component is extracted from such a picked up image and the number plate part is segmented based on that extracted vertical edge component, the position of the number plate can be accurately calculated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is installed in an automobile passage for obtaining traffic congestion information on a highway or information on the detection of a violating vehicle, and is used to optically detect a vehicle using an image pickup means such as a camera. The present invention relates to a vehicle recognition device that recognizes passage.

[0002]

2. Description of the Related Art In a conventional vehicle recognizing device of this type, as shown in FIGS. 13 and 14, a gate shape is formed at a predetermined position of a passage 20 for an automobile such as an expressway so as to cross the passage width. A column 22 is provided, and a camera 22 as an image capturing unit for capturing an image of a vehicle C (automobile) traveling in the passage 20 from above the front side in the traveling direction A is provided on an upper horizontal side portion 21a of the gate column 21. Install this camera 22
Thus, as shown in FIG. 15, an image of the front portion including the license plate N of the vehicle C is captured, the captured image is input to the image processing device 23, and in the image processing device 23, as shown in FIG. Since the edge component e in the vertical direction is extracted from the input captured image, and the components in the direction closer to vertical than the other parts of the vehicle C exist closer to each other based on the extracted vertical edge component e. It was configured to cut out a license plate portion having a characteristic that the level change of the contrast in the horizontal direction is very large and calculate the position of the license plate on the screen.

[0003]

By the way, when the camera 22 is installed above the front side in the traveling direction of the vehicle C as in the conventional general vehicle recognition device described above, the gate-shaped support column 21 is installed or When mounting the camera 22 on the gate-shaped support 21, large-scale construction such as temporarily restricting traffic in the aisle 20 or restricting traffic in some lanes is required, and the construction cost is not only expensive. However, there is a problem that it is likely to cause traffic congestion.

As a method of solving the problems in construction of such a vehicle recognition device, as shown in FIGS. 17 and 18, the camera 22 is vertically erected on a side portion of a passage 20 for an automobile. The vehicle C, which is installed above, is imaged from diagonally above with respect to the traveling direction A of the vehicle C, and the captured image is input to the image processing device 23. Similarly to the above, it is conceivable to cut out a license plate portion having a larger characteristic than other portions and calculate the position of the license plate.

However, as shown in FIGS. 17 and 18, the camera 22 is installed on the side portion of the passage 20 and the vehicle C is installed.
In the diagonally upper installation type that captures images from diagonally above,
As shown in FIG. 19, the entire captured image is tilted obliquely, and in the case of imaging from above the front, a horizontal edge component that is a characteristic of the horizontal direction, specifically, a part of the front grill G of the vehicle C is vertical. It is extracted as the edge component e. Further, the license plate N itself is also deformed obliquely, and the distribution of character numbers is shifted vertically. Therefore, as shown in FIG. 20, the vertical edge component e is extracted from the captured image,
On the basis of the extracted vertical edge component e, the position calculation accuracy of cutting out a license plate portion having a larger feature than other portions and calculating the position is lowered, and
Since the distribution of character numbers also shifts in the vertical direction, it is difficult to know their positions. Therefore, although the vehicle recognition device of the obliquely upper installation type has an advantage in construction, it has the above-mentioned problems in achieving a predetermined vehicle recognition performance.

The present invention has been made in view of the above-mentioned circumstances, and improves the position calculation accuracy of the license plate portion of the vehicle while ensuring the constructional advantage as an obliquely upward installation method for imaging the vehicle from obliquely above. It is an object of the present invention to provide a vehicle recognition device capable of reliably exhibiting a predetermined vehicle recognition performance.

[0007]

In order to achieve the above object, a vehicle recognition device according to the invention of claim 1 is installed on a side portion of a passage, and includes an image including a license plate of a vehicle traveling in the passage. Image pickup means for picking up the image from above obliquely with respect to the traveling direction of the vehicle, vertical edge component extraction means for extracting edge components in the vertical direction from the image picked up by this image pickup means, and vertical edges extracted by this vertical edge component extraction means. A license plate portion cutting-out means for cutting out a license plate portion from the imaged image based on an edge component, and a horizontal portion of the vehicle imaged by the image pickup means is horizontal in the imaged image before the extraction of the vertical edge component. It is characterized in that

In the vehicle recognizing device having the above-mentioned structure, as described in claim 2, the whole image pickup means is rotated about its optical axis and installed in an inclined posture so that the horizontal portion of the vehicle is in the picked-up image. Even if it is made horizontal, the image pickup means is installed in a stable posture, and the image pickup section located inside the image pickup means is rotated about the optical axis so that the horizontal portion of the vehicle is kept horizontal. You may make it horizontal in a captured image.

Further, the vehicle recognition apparatus according to the invention of claim 4 is installed on a side portion of a passage, and an image including a license plate of a vehicle traveling in the passage is imaged obliquely from above in the traveling direction of the vehicle. Image capturing means, vertical edge component extracting means for extracting edge components in the vertical direction from the image captured by the image capturing means, and license plate portion from the captured image based on the vertical edge components extracted by the vertical edge component extracting means. A number plate portion cutting-out means for cutting out the image, an image converting means for rotating the imaged image by the image capturing means so that the horizontal portion of the vehicle is horizontal in the imaged image, and at least the number of the image converted by this image converting means. And a memory for storing the data of the plate portion.

Further, it is desirable that the vehicle recognizing device having the above-mentioned angular structure be provided with plate number recognizing means for recognizing the character number of the license plate portion cut out by the license plate portion cutting means.

[0011]

According to the first and fourth aspects of the invention, the traveling vehicle is imaged obliquely from above by the image pickup means installed on the side of the passage. There is no need for large-scale construction work such as restricting traffic of some lanes, and it is possible to construct easily and economically without causing traffic congestion.
Although the image captured by the image capturing means is deformed diagonally as a whole, the horizontal portion of the actual vehicle is horizontal in the captured image. Consistent with horizontal features. In addition, the distribution of plate numbers on the license plate will also be distributed horizontally without shifting in the vertical direction. By extracting the edge component in the vertical direction from such a captured image and cutting out the license plate portion having a larger feature than other portions based on the extracted vertical edge component, the position of the license plate can be made extremely high. It is possible to calculate with high accuracy.

Further, as in claim 2, there is adopted a means for rotating the entire image pickup means about its optical axis to install it in an inclined posture, or as in claim 3, inside the image pickup means. When the means for rotating the image pickup unit located around the optical axis is adopted, the image pickup unit is arranged so that the horizontal portion of the actual vehicle is horizontal while observing the output of the image pickup unit on the monitor during installation. All that is required is to rotate the entire or internal image pickup unit, the construction is very simple, and special image processing is not required, and the cost is low. Particularly, in the case of claim 3, it is not necessary to install the entire image pickup means in a tilted posture that is unstable and visually unappearing, and the horizontal portion of the actual vehicle is installed while being installed in a stable and well-formed posture. Can be horizontal in the captured image.

Further, when the plate number recognition means for recognizing the character number of the license plate portion cut out by the license plate portion cutting means is added, the vehicle can be specified more accurately.

[0014]

Embodiments of the present invention will be described below with reference to the drawings. 1 and 2 are a schematic plan view and a schematic front view showing the situation around the installation portion of a vehicle recognition device according to an embodiment of the present invention. In FIG. 1 and FIG. Is installed on a pole 3 standing upright on the side of an automobile passage 2 and images a vehicle C traveling in the passage 2 obliquely from above with respect to the traveling direction A of the vehicle C. The captured image will be described later. Input to the image processing device 4.

As shown in FIGS. 3 and 4, the camera 1 is tilted as a whole on the support plate 5 fixed to the side surface of the pole 3 in a protruding state by rotating about the optical axis a. It is installed and fixed in a posture. More specifically, as shown in FIG. 4, when the camera 1 is viewed from the optical axis a, the horizontal axis x of the camera is normally on the horizontal plane and the vertical axis of the camera is vertical as shown by the phantom line in FIG. The camera 1 is rotated around the optical axis a from the state where y is located on the vertical plane, and the camera horizontal axis x and the camera vertical axis y are installed in an inclined posture having a rotation angle θ with respect to the horizontal plane and the vertical plane, respectively. Fix it.

Further, instead of mounting the entire camera 1 on the support plate 5 in an inclined posture and fixing it, as shown in FIGS. 5 and 6, a CCD which is an image pickup section located inside the camera 1.
1A may be rotated by a predetermined rotation angle θ via a rotary plate 1B capable of rotating around the optical axis a. In this case, the entire camera 1 is unstable with respect to the support plate 5,
In addition, it is not necessary to install it in an appearance that is unsightly in appearance, and it is possible to install it on the support plate 5 in a stable and good appearance. Further, the rotation adjusting means of the CCD 1A such as the rotating plate 1B is not provided, and the internal C is rotated by a predetermined rotation angle θ in advance.
You may rotate and fix CD1A.

Here, the rotation angle θ of the camera 1 or CCD 1A is input to the horizontal portion of the vehicle C imaged by the camera 1, for example, the front grill G or the license plate N before the extraction of the vertical edge component, that is, the input. At the image stage, as shown in FIG. 7, the angle is set such that the horizontal portion of the vehicle C such as the front grill G and the license plate N becomes horizontal in the captured image. The rotation angle θ does not need to be calculated, and the camera 1
When the camera 1 is installed and fixed on the support plate 5, if the camera 1 is rotated and fixed so that the horizontal portion of the actual vehicle C becomes horizontal on the image while observing the output image of the camera 1 on the monitor. It's good and very easy. Also, FIG.
The imaginary line (the chain double-dashed line) indicates the field of view of the camera 1 when the rotation angle θ is zero.

FIG. 10 is a schematic block diagram showing the structure of the image processing apparatus 4 for processing the image picked up by the camera 1. In FIG. 10, reference numeral 6 denotes an A / D conversion circuit for inputting an image input by the camera 1. A / D convert. An image memory 7 sequentially takes in multi-valued image data that is A / D converted by the A / D conversion circuit 6 and sequentially output. here,
As the camera 1, a television camera having a mechanical or electronic shutter function, for example, a CCD still image pickup camera may be used to capture a still image of the vehicle and capture the output of the still image. .

Reference numeral 8 denotes a space differentiating circuit that horizontally differentiates the image of the vehicle C including the license plate N, which is multi-valued image data stored in the image memory 7, in the horizontal direction. Reference numeral 9 is a vertical edge component extraction circuit, which extracts an edge component e in the vertical direction from the image obtained by differentiating the spatial differentiation circuit 8 as shown in FIG. Reference numeral 10 is a binarization circuit which converts multi-valued image data output from the image memory 7 into binary image data. Reference numeral 11 denotes a license plate partial cutout circuit, which scans an integration mask on the vertical edge component e extracted by the vertical edge component extraction circuit 9 to integrate the vertical edge component e as shown in FIG. The license plate portion that is being cut out is cut out, and the position of the cut out license plate portion on the screen is calculated based on the scanning detection of the location of the maximum integrated value of the vertical edge component e.

A plate number recognition circuit 12 cuts out a character number from the license plate partial cutout circuit 11 from the image data binarized by the binarization circuit 10 and identifies the cut out character number.
A vehicle information output circuit 14 outputs vehicle information based on the output from the plate number recognition circuit 12 and the output from the license plate position calculation circuit 13.
Here, as the vehicle information, when operating as a vehicle recognition operation, based on the recognition result such as the number and code output from the plate number recognition circuit 12,
For example, when the vehicle information such as a four-digit serial number, land code, vehicle type code, and application code is output, and when it operates as a vehicle detection operation, the plate number recognition circuit 1 described above is used.
The output processing from 2 is removed, and the presence or absence of the cutout position of the character number or its position information is output as the output of the license plate partial cutout circuit 11, and the presence or absence of the cutout position or the presence or absence of the vehicle based on the position information. Is output.

Next, a series of vehicle recognition processing by the vehicle recognition device configured as described above will be briefly described with reference to the flowchart of FIG. An image of the vehicle C traveling in the passage 2 is taken from diagonally above with respect to the traveling direction A of the vehicle C by the camera 1 installed on the side of the passage 2,
The captured image is input to the image processing device 4 (step S
30). Here, regarding the image input to the image processing device 4, the camera 1 or the CCD 1A inside the image processing device 4 is shown in FIGS.
Since it has a rotation angle θ as shown in Fig. 7, it is deformed as a whole as shown in Fig. 7, but is in a horizontal portion of the vehicle C, such as the front grill G or the license plate N.
Is horizontal in the image.

Next, the input image is A / D converted by the A / D conversion circuit 6 and then taken into the image memory 7 as a multivalued still image, and the image is interlocked with the shutter operation of the camera 1. Memory 7 to spatial differentiation circuit 8
The image of the vehicle C including the license plate N is spatially differentiated in the horizontal direction. This spatial differentiation circuit 8
The image obtained by being differentiated by is a vertical edge component extraction circuit 9
, And the vertical edge component e is extracted as shown in FIG. 8 (step S31). At the same time, the multivalued image data output from the image memory 7 is converted into binary image data in the binarization circuit 10.

Subsequently, the vertical edge component e extracted by the vertical edge component extraction circuit 9 is input to the license plate partial cutout circuit 11, where the integration mask is scanned for the extracted vertical edge component e. As a result, as shown in FIG. 9, the license plate portion L in which the vertical edge component e is accumulated is cut out (step S32). In this way, the position on the screen of the license plate portion cut out by the license plate portion cutting circuit 11 is calculated based on the scanning detection of the place of the maximum integrated value of the vertical edge component e (step S33). . And
The presence or absence of the license plate portion is determined (step S3
4) If there is no license plate part, step S3
Returning to 0, the shutter operation of the camera 1 is performed and the vehicle C is imaged from diagonally above again.

On the other hand, when there is a license plate, a character number is cut out from the license plate partial cut-out circuit 11 based on the binary image data binarized by the binarization circuit 10 and the cut-out character. The plate number recognition circuit 12 recognizes the number (steps S35, S36), and vehicle information is output from the vehicle information output circuit 14 based on the recognized character number or the output from the license plate partial cutout circuit 11. (Step S37).

In the above embodiment, the camera 1 itself or the inside of the camera 1 is provided as means for making the horizontal portion of the vehicle horizontal in the captured image before the extraction of the vertical edge component. The CCD 1A has been described by adopting the means for rotating the CCD 1A in the posture having the rotation angle θ with respect to the horizontal plane and the vertical plane about the optical axis, but the camera 1 is as shown by the phantom lines in FIGS. 3 and 4. While the horizontal axis x of the camera is placed on the horizontal plane and the vertical axis y of the camera is placed on the vertical plane, the captured image by the camera 1 becomes an image as shown in FIG. An image conversion circuit may be provided to rotate the image sensor by an amount corresponding to the rotation angle θ. Specifically, in the image processing circuit 4 shown in FIG. 10, the image conversion circuit 15 is provided between the image memory 7 and the horizontal differentiating circuit 8 as indicated by a virtual line in the figure.
And a memory 1 for storing all of the converted image data or at least data of the license plate portion
6 is interposed.

A series of vehicle recognition processing by the vehicle recognition device including the image processing device 4 having the image conversion circuit 15 for rotating the image picked up by the camera 1 is basically as shown in FIG. As shown in FIG. 12, the point to be set is between the image input step (step S30) and the vertical edge component extraction step (step S31).
Only the step S38 of rotating the input image by a predetermined rotation angle θ by the image conversion circuit 15 is added, and the other processing steps are exactly the same as in the case of FIG.

Further, in the above embodiment, the character number is cut out from the license plate partial cutout circuit 11, and the cut out character number is recognized by the plate number recognition circuit 12, thereby enhancing the recognition performance of the passing vehicle. However, it is also possible to detect the vehicle only by calculating the license plate position without providing such a plate number recognition circuit 12.

[0028]

As described above, claim 1 and claim 4
According to the invention, by adopting the method of imaging the traveling vehicle obliquely from above by the imaging means installed on the side of the aisle, the traffic of the vehicle is temporarily restricted or the traffic of some lanes is restricted. It eliminates the need for large-scale construction work such as
It can be constructed reasonably and economically without causing traffic congestion. Nevertheless, although the image captured by the image capturing means is deformed diagonally as a whole,
Since the horizontal portion of the actual vehicle can be made horizontal in the captured image, it can be made to match the horizontal feature when the vehicle is captured from above the front in the traveling direction. Also, the distribution of the character numbers on the license plate can be distributed horizontally without shifting in the vertical direction. Therefore, the edge component in the vertical direction is extracted from the captured image, and based on the extracted vertical edge component, the license plate portion having a greater feature than other portions is cut out,
When calculating the position, there is an effect that the position calculation accuracy can be kept high and the predetermined vehicle recognition performance can be surely and satisfactorily exhibited.

Further, as in claim 2, there is adopted a means for rotating the entire image pickup means around its optical axis to install it in an inclined posture, or as in claim 3, inside the image pickup means. When the means for rotating the image pickup unit located around the optical axis is adopted, the image pickup unit is arranged so that the horizontal portion of the actual vehicle is horizontal while observing the output of the image pickup unit on the monitor during installation. Only by rotating the whole or the inside image pickup unit, the construction can be further simplified, and no special image processing is required, and the cost can be reduced. In particular, in the case of claim 3, it is not necessary to install the entire image pickup means in a tilted posture that is unstable and is visually unappearing.
The horizontal part of the actual vehicle can be horizontal in the captured image.

Further, when a plate number recognition means for recognizing the character number of the license plate portion cut out by the license plate portion cutting means is added, the position of the license plate and the recognition of the character number on the license plate are Both can identify the vehicle more accurately.

[Brief description of drawings]

FIG. 1 is a schematic plan view showing a situation around an installation portion of a vehicle recognition device according to an embodiment of the present invention.

FIG. 2 is a schematic front view of FIG.

FIG. 3 is an enlarged perspective view of a main part for explaining the installation state of the camera.

FIG. 4 is a front view of the camera of FIG. 3 when viewed from the optical axis direction thereof.

FIG. 5 is a schematic enlarged side view of essential parts for explaining a state in which a CCD inside a camera is rotated.

6 is a front view of the camera of FIG. 5 as viewed from the optical axis direction thereof.

FIG. 7 is a front view showing an image picked up by a camera.

FIG. 8 is a front view showing an extracted image of a vertical edge component.

FIG. 9 is a front view showing a cutout image of a license plate portion.

FIG. 10 is a schematic block diagram showing the configuration of an image processing apparatus.

FIG. 11 is a flowchart illustrating a series of vehicle identification processing.

FIG. 12 is a flowchart illustrating a series of vehicle identification processing of a vehicle recognition device according to another embodiment of the present invention.

FIG. 13 is a schematic plan view showing a situation around the installation portion of the conventional vehicle recognition device.

FIG. 14 is a schematic front view of FIG.

FIG. 15 is a front view showing an image picked up by a camera in the vehicle recognition device of FIG.

16 is a front view showing an image in which a vertical edge component is extracted from the captured image of FIG.

FIG. 17 is a schematic plan view showing a situation around the installation portion of the vehicle recognition device, which is conceived in order to solve the problem in construction by the vehicle recognition device of FIG.

FIG. 18 is a schematic front view of FIG. 17.

19 is a front view showing an image picked up by a camera in the vehicle recognition device of FIG.

20 is a front view showing an image in which a vertical edge component is extracted from the captured image of FIG.

[Explanation of symbols]

 1 Camera (Imaging Means) 1A CCD (Imaging Unit) 2 Passage 9 Vertical Edge Component Extraction Circuit 11 License Plate Partial Cutout Circuit 12 Plate Number Recognition Circuit 15 Image Conversion Circuit 16 Memory C Vehicle N License Plate e Vertical Edge Component

Claims (5)

[Claims] 1. An image pickup means installed on a side portion of a passage for picking up an image including a license plate of a vehicle traveling in the passage from obliquely above the traveling direction of the vehicle, and an image picked up by the image pickup means. A vertical edge component extracting means for extracting an edge component in the vertical direction; and a license plate portion cutting means for cutting out a license plate portion from the captured image based on the vertical edge component extracted by the vertical edge component extracting means, A vehicle recognition device characterized in that a horizontal portion of a vehicle imaged by the image pickup means is made horizontal in a captured image before a vertical edge component is extracted. 2. The vehicle recognition device according to claim 1, wherein the entire image pickup means is rotated about its optical axis and installed in an inclined posture so that a horizontal portion of the vehicle is horizontal in the picked-up image. 3. The image pickup means is installed in a stable posture, and a horizontal part of the vehicle is made horizontal in a picked-up image by rotating an image pickup part located inside the picked-up part around an optical axis. 1. Vehicle recognition device. 4. An image pickup means installed on a side portion of a passage for picking up an image including a license plate of a vehicle traveling in the passage from obliquely above with respect to a traveling direction of the vehicle, and an image picked up by the image pickup means. Vertical edge component extracting means for extracting an edge component in the vertical direction; license plate portion cutting means for cutting out a license plate portion from the captured image based on the vertical edge component extracted by the vertical edge component extracting means; And a memory for storing data of at least the license plate portion of the image converted by the image conversion means, and the image conversion means for rotating the image captured by the means so that the horizontal portion of the vehicle becomes horizontal in the captured image. A vehicle recognition device characterized by the above. 5. The vehicle recognition device according to claim 1, further comprising plate number recognition means for recognizing a character number of the license plate portion cut out by the license plate portion cutting means.