JPH05312818A - Speed measuring device and method through image processing - Google Patents

Abstract

PURPOSE:To correctly obtain the speed of a vehicle by correctly obtaining the transfer distance of the vehicle by allowing not only the vehicle head posi tion but the continuous division of a vertical file to contribute to the calculation of the transfer distance of the vehicle. CONSTITUTION:Two images including each vehicle head are taken in, keeping a time interval (Step n1, n4). A contour image consisting of contour image elements is prepared by extracting the edge parts having the sharp concentration variation, from the taken-in image. The contour image is scanned in the direction crossing with the direction corresponding to the advance direction of the vehicle. The number of contour elements existing on each scan line is counted, and a vertical profile is prepared. The relative calculation is applied for two vertical profiles which are obtained in correspondence with two images (Step n9). From the result of the relative calculation, the transfer distance of the vehicle in the taken-photograph time for two images is calculated (Step n10). The speed of the vehicle is calculated on the basis of the transfer distance and the taken-photograph time difference between two images (Step n11).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is used in a vehicle number automatic reading device or the like for automatically identifying a plate number by photographing a vehicle head portion of a vehicle running by a camera device, and the speed of the vehicle running by image processing. The present invention relates to an apparatus and method for measuring

[0002]

2. Description of the Related Art Conventionally, a vehicle number automatic reading device for optically detecting a vehicle license plate with a television camera installed on the road and for identifying the plate number by image processing has been used for controlling a vehicle violating a speed limit. Being developed. In order to control a vehicle that violates the speed limit, it is necessary to automatically read the vehicle number and also to measure the speed of the running vehicle. Therefore, in the past, a speed measuring device was provided separately from the vehicle number automatic reading device.
As seen in Japanese Patent Publication No. 76069, a technique for measuring the speed of a vehicle by image processing in a vehicle number automatic reading device has been developed. That is, since a speed measuring function is added to the vehicle number automatic reading device, it is not necessary to provide a special speed measuring device.

In the prior art disclosed in the above-mentioned publication, a predetermined area on the road is photographed by a camera device, and it is judged whether or not the front end of the vehicle is present in the photographed image. .. Further, when the vehicle head is present, the vehicle head position in the image is calculated. The detection of the vehicle head position is performed as follows. That is,
First, the edge portion of the image captured based on the digitized image signal is detected, and the contour of the image is extracted. An example of the contour image obtained by extracting this contour is shown in FIG.
Is shown in. The number of contour pixels forming the contour is scanned and counted along the horizontal direction x of the screen, and a vertical profile in the vertical direction y of the screen is created based on the count value for each scanning line. This vertical profile is shown in Figure 4.
It is shown in (b). The vertical direction y is a direction along the traveling direction of the vehicle.

Further, the vertical profile is scanned along the direction R opposite to the traveling direction of the vehicle, and the position yhead where the vertical profile exceeds the threshold value A is determined as the vehicle head position in the course of this scanning. The plate cutting process that cuts out the image of the license plate part for automatic identification of the car number is
This is performed based on the detected vehicle head position yhead. When the vehicle head position is detected in this manner, similar processing is performed on another image captured thereafter.

Therefore, if the vehicle head position obtained for the first image is taken as the first vehicle head position yhead1 and the vehicle head position obtained for the next image is taken as the second vehicle head position yhead2,
The difference between these two vehicle head positions (yhead2-yhead1) is 2
It is equal to the distance traveled by the vehicle head on the screen between the times when two images were taken. Therefore, the difference in vehicle head position (yhead2-yhead
By converting 1) into the actual travel distance of the vehicle, and further dividing this travel distance by the shooting time difference between the two images,
The speed of the vehicle can be determined.

[0006]

However, in the above-mentioned prior art, it is not always possible to accurately measure the speed of the vehicle. That is, the detected first vehicle head position yhead1 and second vehicle head position yhead2 are not necessarily positions corresponding to the same portion of the vehicle, and in this case, an error occurs in speed measurement.

More specifically, as shown in FIG. 9 (b), the first vehicle head position yhead1 is detected at the bumper 62 of the vehicle 61 as shown in FIG. 9 (a). In some cases, the second vehicle head position yhead2 may be detected below the license plate 63 of the vehicle 61. At this time, the difference L2 between the two vehicle head positions (= yhead
2-yhead1) is not equal to the accurate travel distance L0 of the vehicle head in the image. Therefore, the measured vehicle speed is

[0008]

[Equation 1]

An error of occurs. Therefore, an object of the present invention is to solve the above-mentioned technical problems and to provide a speed measuring device and a speed measuring method capable of accurately measuring the speed of a running vehicle by image processing.

[0010]

In order to achieve the above object, a speed measuring device by image processing according to claim 1 processes an image photographed by a photographing means for photographing a predetermined area on a road, A speed measuring device by image processing for measuring the speed of a running vehicle, wherein a contour extracting means for extracting, as a contour pixel, a pixel at an edge portion in which a density change rate of an image photographed by the photographing means is a predetermined value or more. And a contour image composed of the contour pixels is scanned in a direction intersecting a direction corresponding to the traveling direction of the vehicle, the contour pixels existing on each scanning line are respectively counted, and from the set of count values, A contour profile creating means for creating a contour profile, a storage means for storing the contour profile created by the contour profile creating means, and a storage means for storing the contour profile. Correlation calculation means for performing a correlation calculation between the contour profile corresponding to the image photographed by the photographing means after the photographing time of the image corresponding to the stored contour profile and the contour profile stored in the storage means. And a moving distance calculating means for calculating the moving distance of the vehicle between the photographing times of the images corresponding to the two contour profiles based on the result of the correlation calculation of the two contour profiles by the correlation calculating means, It is characterized by including a speed calculation means for calculating the speed of the vehicle on the basis of the photographing time difference of each image corresponding to one contour profile and the movement distance calculated by the movement distance calculation means.

Further, in the speed measuring device according to the second aspect of the present invention, the correlation calculation means compares one of the contour profiles to be subjected to the correlation calculation with a predetermined threshold value, and photographs the image with the photographing means. Vehicle position calculation means for calculating the vehicle head position in the captured image, means for extracting a partial profile which is a part of the one contour profile in the vicinity of the obtained vehicle head position, and the extracted partial profile and the other A matching unit that performs pattern matching with the contour profile and finds a matching portion that best matches in the other contour profile; and the movement distance calculation unit includes the matching unit based on a shift amount between the partial profile and the matching portion. It is characterized in that the moving distance of the vehicle is calculated.

According to a third aspect of the present invention, there is provided a speed measuring method based on image processing for measuring a speed of a running vehicle by processing an image photographed by a photographing means for photographing a predetermined area on a road. A method of extracting a pixel of an edge portion in which a density change rate of an image photographed by the photographing means is a predetermined value or more as a contour pixel, and the contour image formed by the contour pixel corresponds to a traveling direction of the vehicle. Scanning is performed in a direction intersecting with the scanning direction, the contour pixels existing on each scanning line are respectively counted, a contour profile composed of a set of these count values is created, and two images photographed at time intervals. Correlation calculation is performed between the first contour profile and the second contour profile created for the first contour profile and the second contour profile, and based on the result of this correlation calculation, The moving distance of the vehicle between the shooting times of the images corresponding to the contour profile is calculated, and the shooting time difference between the images corresponding to the first contour profile and the second contour profile and the calculated vehicle It is characterized in that the speed of the vehicle is obtained based on the moving distance.

Further, in the speed measuring method by image processing according to claim 4, in the correlation calculation, one of the contour profiles to be subjected to the correlation calculation is compared with a predetermined threshold value and the image is photographed by the photographing means. The vehicle head position in the image is calculated, the partial profile that is a part of the one contour profile in the vicinity of the obtained vehicle head position is extracted, and pattern matching is performed between the extracted partial profile and the other contour profile. Further, it is performed so as to find a matching portion that best matches the other contour profile, and the moving distance of the vehicle is calculated based on a shift amount between the partial profile and the matching portion. To do.

[0014]

According to the present invention, the contour pixel is extracted from the image photographed by the photographing means to create the contour image. This contour image is scanned in a direction intersecting with the direction corresponding to the traveling direction of the vehicle, and a contour profile is created by counting the contour pixels existing on each scanning line. This contour profile is stored in the storage means.

When another image is photographed after the photographing time of the image corresponding to the contour profile stored in the storage means and a contour profile corresponding to the other image is obtained, the contour profile and the storage means are stored. Correlation calculation with the stored contour profile is performed. And
Based on the result of this correlation calculation, the moving distance of the vehicle between the photographing times of the two images is calculated. Then, the speed of the vehicle is calculated on the basis of the moving distance and the difference between the photographing times of the two images.

As described above, in the present invention, the moving distance of the vehicle is obtained by performing the correlation calculation on each contour profile corresponding to the two images photographed at the time intervals. Therefore, not only the head position as in the past,
The entire contour profile will contribute to the determination of the distance traveled by the vehicle. As a result, the moving distance of the vehicle can be accurately obtained.

The structure according to claim 2 or 4 is a structure for reducing the calculation amount of the correlation calculation. That is, first, one of the two contour profiles to be subjected to the correlation calculation is compared with a predetermined threshold value,
A significant portion near the vehicle head position is extracted as a partial profile. Then, pattern matching between this partial profile and the other contour profile is performed. Then, the matching portion that best matches the other contour profile is obtained, and the moving distance of the vehicle is obtained based on the shift amount between the matching portion and the partial profile.

In this way, since the correlation calculation is not performed for all the portions of the two contour profiles, but the calculation is performed for the partial profiles, the amount of calculation can be reduced. Moreover, since a significant portion near the vehicle head position is extracted as a partial profile for pattern matching, the accuracy of matching calculation can be increased.

[0019]

Embodiments of the present invention will be described in detail below with reference to the accompanying drawings. FIG. 2 is a plan view showing an installation example of a vehicle number automatic reading device to which the speed measuring device and method by image processing of one embodiment of the present invention is applied, and FIG.
Is a front view thereof. The vehicle number automatic reading device includes a vehicle number automatic reading device body 1, lighting devices 2a and 2b, a camera device 3, and the like, and the camera device 3 includes the lighting devices 2a and 2b.
It is attached to the upper part of the arm of the installation pole 4 beside the road together with b. The vehicle number automatic reader main body 1 is attached to a side portion of the installation pole 4. A traffic signal, pedestrian bridge, etc. may be used instead of the installation pole.

The camera device 3 is attached with a predetermined dip so that the field of view of the camera device 3 is set in a predetermined region R on the road (shown by hatching in FIG. 2). FIG. 1 is a block diagram showing the internal circuit configuration of the vehicle number automatic reading device main body 1. The image signal from the camera device 3 is converted into digital data by the analog / digital (A / D) conversion circuit 11. This digital data is given to the edge detection circuit 12. Edge detection circuit 12
Detects a portion (edge portion) of the image where the density changes,
The data corresponding to the density change rate is output. The output signal of the edge detection circuit 12 is given to the binarization circuit 13 and binarized based on a predetermined threshold value. The binarized signal is stored in the image memory 14 as a preprocessed image. By the above edge detection processing and binarization processing, pixels for which the density change rate of the image is equal to or greater than a predetermined value, that is, contour pixels forming the contour of the image in the photographing region are extracted. Therefore, the image memory 14 stores a contour image composed of contour pixels. The edge detection circuit 12 and the binarization circuit 13 constitute a contour extraction means.

The data stored in the image memory 14 is image processing processor 15 which is a contour profile creating means.
Is supplied to. The image processor 15 creates a vertical profile (vertical projection diagram) described below for each contour image of one frame. The created vertical profile is stored in the profile memory 16 which is a storage means. The data stored in the profile memory 16 is the correlation calculation means,
It is input to the general-purpose processor 17 functioning as a moving distance calculating means and a speed calculating means, and here, vehicle head position detection processing, plate cutting processing, character cutting processing, and character recognition processing are performed.

More specifically, the edge detection circuit 12
Is to extract only the contour of the vehicle by applying a two-dimensional differentiation to the image in which the vehicle or the like is reflected. By binarizing the output of the edge detection circuit 12 by the binarization circuit 13, the contour of the vehicle becomes clearer, and a contour image as shown in FIG. 4A is obtained. Based on this contour image,
The image processor 15 will create the above vertical profile. This vertical profile is shown in Figure 4 (b). The vertical axis y in FIG. 4 (b) is in the traveling direction of the vehicle.

This vertical profile is obtained by scanning the contour image shown in FIG. 4 (a) along the horizontal direction x, which is a direction intersecting the traveling direction of the vehicle, and counting the number of contour pixels for each scanning line. , As a set of the count values. That is, the vertical profile is, so to speak, a histogram in the vertical direction y of the contour image. Next, the processing performed by the general-purpose processor 17 will be described with reference to the flowchart of FIG. First, the original image is captured (step n1), and the vehicle head position is detected based on the vertical profile obtained by the image processor 15 (step n2). This vehicle head detection processing is processing for scanning the vertical profile in a direction opposite to the vertical direction y and determining a position where the number of contour pixels exceeds the threshold value A as a vehicle head position y head (see FIG. 4). The vehicle head position detected in step n2 is defined as a first vehicle head position y1.

When the vehicle head position is detected (step n
3) Then, the original image is captured again (step n4).
The difference between the time when the original image is taken in step n4 and the time when the original image is taken in step n1 corresponds to the scanning time for one frame of the image. This is because, in this embodiment, the vehicle head detection processing is completed within the scanning time for one frame, but if the vehicle head detection processing is within the scanning time for n (n ≧ 2) frames. When completed, the difference between the time when the original image is captured in step n1 and the time when the original image is captured in step n4 corresponds to the scanning time for n frames of the image.

After capturing the original image in step n4, the vehicle head detection process is performed again (step n5) to detect the vehicle head position. The vehicle head position detected in step n4 is defined as a second vehicle head position y2. If the vehicle head cannot be detected (step n6), the process returns to the beginning, and if the vehicle head can be detected, the moving distance of the vehicle head on the screen is calculated as a temporary moving distance (step n7).

That is, the tentative movement distance LA is calculated from the vehicle head positions y1 and y2 already obtained. This tentative movement distance LA
Is obtained by LA = y2-y1 ... (2). If LA <0, the vehicle is traveling in the opposite direction, so the calculation is terminated.

If LA ≧ 0, LA is compared with the threshold Lo (step n8). If LA <Lo, the moving distance is too small and the speed calculation error becomes large. Therefore, without performing speed calculation, the process returns to step n4 to capture the next original image.
If LA ≧ Lo, it is determined that the moving distance is sufficiently large, and the process proceeds to step n9. At step n9, the vehicle head position y1,
The correlation calculation described later is performed on each vertical profile in which y2 is detected. Then, in step n10,
Based on the result of the correlation calculation, the moving distance L of the vehicle head on the screen is calculated, and in step n11, the speed of the running vehicle is calculated based on the calculated moving distance L.

A vehicle number reading process is performed in parallel with the process for obtaining such speed (step n12). In this vehicle number reading process, as shown in FIG. 6, the range Δ including the range of front α and the range of rear β of the vehicle head position yhead is limited to the plate cutout range on the entire screen 21, and the plate cutout range is further set. In Δ, the plate is cut out, the character is cut out to extract the character, and finally the character is recognized to identify the vehicle number.

Next, the correlation calculation in step n9 will be described. Generally, when a one-variable function f (x) and g (x) of x are given, the correlation between these two functions is given by the following equation (3).

[0030]

[Equation 2]

Is given by Assuming that the vertical profile is represented by a function p (y) representing each count value p by the position coordinate y in the vertical direction y on the screen, a function representing two vertical profiles to be subjected to correlation calculation is represented by p.
Assuming ₁ (y) and p ₂ (y), the correlation between the two vertical profiles is given by the following equation (4). However, for y outside the screen, p ₁ (y) = 0, p ₂ (y) =
Assume 0.

[0032]

[Equation 3]

Assuming that (p ₁ · p ₂ ) (x) has a maximum value at x = x ₁ , when p ₁ (y) and p ₂ (y) are shifted by x ₁ , p ₁ and p ₂ will be the most similar. By the way, since p ₁ (y) and p ₂ (y) are obtained for the same vehicle, the absolute value of x ₁ | x
₁ | is nothing but the travel distance of the vehicle. That is, L = | x ₁ | ... (5).

The correlation calculation in step n9 follows
Given by equation (4) above (p₁・ P₂) (X)
Processing in step n10 of FIG.
Is (p ₁・ P₂) (X) takes the maximum value x = x₁
It is nothing but an operation to obtain. Such correlation calculation
And the moving distance L as the moving distance of the entire vertical profile.
Will be required. JP-A-3-27606 mentioned above
In the prior art disclosed in Japanese Patent Publication No. 9, the above-mentioned temporary moving distance is
Obtain the travel distance of the vehicle based on the separation LA (= y2-y1)
I am sorry. That is, in this case
It means that and were matched in a zero dimension. to this
On the other hand, in this embodiment, the correlation between the vertical profiles is
Is required, so one-dimensional matching is performed.
Be done. Therefore, in comparison with the above-mentioned prior art, the present implementation
According to the example, the distance traveled by the vehicle can be calculated extremely accurately.
Growls That is, vehicles detected in different parts of the vehicle
Inaccurate as a result of distance calculation based on head position
The speed of the vehicle is calculated based on the distance traveled
You can prevent problems.

In order to obtain the moving distance of a specific object in a two-dimensional image, originally, the correlation calculation should be performed as a two-dimensional spread. However, in the automatic vehicle number reader, one moving object (that is, vehicle) is displayed on the entire screen.
Considering that the vehicle traveling direction is limited to one direction, it is sufficient to obtain a one-dimensional correlation with respect to the traveling direction of the vehicle as a correlation calculation for obtaining the traveling distance of the vehicle. .. That is, the moving distance of the vehicle can be accurately obtained with a small amount of calculation.

Further, since the vertical profile calculated for the vehicle head detection process is used and the moving distance is calculated by performing the correlation calculation of this vertical profile, it is not necessary to obtain a new function. There is also.
Next, the calculation of the speed of the vehicle performed in step n11 of FIG. 5 will be described with reference to FIG. The height of the camera device 3 is H, the dip angle is θ, the road surface is CO, and the average height of the vehicle head is ΔH, and the vertical field of view of the screen imaged by the camera device 3 is Y.
And

The positions on the road surface corresponding to both ends B and A of the vertical field of view of the screen are C and A, respectively, and the entry positions of the vehicle head are E and D, respectively. Let G be the entry position of the vehicle head corresponding to the first vertical profile that precedes in time, and
Let F be the approach position of the vehicle head corresponding to the second vertical profile that temporally succeeds one of the two vertical profiles. At this time, the moving distance L of the vehicle in the image obtained as a result of the correlation calculation is G,
The straight line connecting the point F and the camera device 3 is the distance between points J and K where the line segment AB intersects.

Further, r is the distance between the points A and K on the screen. Then, the following expressions (6) and (7) are established.

[0039]

[Equation 4]

Similarly, the following equations (8) and (9) hold.

[0041]

[Equation 5]

From this, GF = DE-DF (10) is obtained. The speed of the vehicle can be obtained by dividing GF by "the difference between the time when the original image of step n1 is taken and the time when the original image of step n4 is taken". The obtained speed data is output to the outside of the vehicle number automatic reading device and used for speed determination of the vehicle.

As described above, according to this embodiment, the moving distance of the vehicle can be accurately obtained by calculating the correlation between the vertical profiles corresponding to the two images taken at the time intervals. That is, since the moving distance of the vehicle is obtained based on the moving distances of the two vertical profiles as a whole, such moving distance can be obtained extremely accurately. Therefore, the speed of the vehicle can be calculated with high accuracy by calculating the speed of the vehicle based on the accurately calculated moving distance.

FIG. 8 illustrates a second embodiment of the present invention.
It is a figure. In this example, two vertical profiles
Is characterized by its correlation calculation, and other points are the same as in the first embodiment above.
Since this is the same as the example, only this correlation calculation will be explained.
It In the description of this embodiment, the above-mentioned FIG.
refer. Figure 8 shows the two vertical planes that are the targets of the correlation calculation.
Profile files P1 and P2 are shown in (a) and (b) respectively.
There is. First, the second image captured in step n4
From the second vertical profile P2 in Fig. 8 (b),
Section including head position y2 [y ₂-W₂, Y₂+ W₁] Section
Minute profile is reference pattern P_REFIs extracted as
It w is the number of scan lines to be included in the section,
If w = w₁+ W₂+1 ... (11)

When the reference pattern P _REF is extracted, a matching process with each part of the first vertical profile P1 of FIG. 8A corresponding to the first image captured in step n1 is performed next. Be seen. In this matching process, first, S (y) represented by the following expression (12) is calculated. That is, the reference pattern P _REF is adjusted to each part of the vertical profile corresponding to the image first captured, and the sum of the differences for each scanning line is obtained.

[0046]

[Equation 6]

Next, y = y at which S (y) takes the minimum value S _min.
min is required. However, since the traveling direction of the vehicle is the positive direction of y and p ₁ (y) is valid only when y ≧ 0, the range of 0 ≦ y ≦ y−w ₂ (13) _Will determine y _min .

This y _min has the following meaning. That is, the reference pattern V _REF, from the position of the second vertical Puroairu _{P2, L m = (y 2} -w 2) -y min ···· (14) by shifted, the reference pattern V _REF
Best matches the first vertical profile P1 of FIG. 8 (a).

That is, the reference pattern V _REF created by extracting the partial profile including the vehicle head position y2 is the second pattern.
In the first image, since it was shifted by the distance L _m ,
This distance L _m is nothing but the travel distance L of the vehicle. That is, L = L _m (15) In this way, by matching the partial profile (reference pattern V _REF ) extracted from the second vertical profile P2 with the first vertical profile P1, The travel distance of the vehicle can be obtained. Such a correlation operation is
Compared to the case where the correlation calculation is performed on all the two vertical profiles as in the case of the first embodiment, the product-sum calculation can be significantly reduced. Therefore, the amount of calculation can be reduced, and the calculation time for obtaining the moving distance L can be shortened.

Moreover, in the matching processing as in the present embodiment, as can be seen from the above expression (12), since no multiplication operation is performed, no special product-sum operation circuit or the like is required, and the device configuration is Can be simplified. Furthermore, in the present embodiment, the matching process is performed using a significant portion including the vehicle head position in the second vertical profile P2, so that the moving distance can be obtained sufficiently accurately.

The present invention is not limited to the above embodiment. For example, in the above-described second embodiment, the partial profile for performing the matching process is extracted from the second vertical profile P2, but such a partial profile may be extracted from the first vertical profile. Further, in the above embodiment, the correlation calculation is performed by the general-purpose processor 17, but the correlation calculation may be performed by the image processing processor 15.

Further, in the above embodiment, the example applied to the vehicle number automatic reading device has been described, but the speed measuring device by the image processing of the present invention is not limited to the vehicle number automatic reading device.
The present invention can be applied to any device that performs image processing, and can also be used as a speed measurement device alone.
Other various modifications may be made without departing from the spirit of the invention.

[0053]

As described above, according to the present invention, the moving distance of the vehicle can be obtained by performing the correlation calculation on each contour profile corresponding to two images taken at time intervals. Therefore, not only the vehicle head position as in the conventional case but the entire contour profile contributes to the determination of the moving distance of the vehicle. Therefore, the moving distance of the vehicle can be accurately obtained, and the speed measurement by image processing can be performed with high accuracy.

Further, if the correlation calculation is performed by pattern matching between a partial profile which is a significant portion near the vehicle head position of one of the contour profiles to be subjected to the correlation calculation and the other contour profile,
The calculation amount can be significantly reduced as compared with the case where the correlation calculation is performed on all the parts of the two contour profiles.

Moreover, a significant portion near the vehicle head position is
Since it is extracted as a partial profile for pattern matching, the accuracy of the matching calculation can be increased. Further, since the pattern matching can be realized without including the multiplication operation, a special product-sum operation circuit or the like is not required, and the speed measurement accuracy can be improved with a simple configuration.

[Brief description of drawings]

FIG. 1 is a block diagram showing an electrical configuration of a vehicle number automatic reading device to which a speed measuring device and a speed measuring method by image processing according to an embodiment of the present invention are applied.

FIG. 2 is a plan view showing an installation example of a vehicle number automatic reading device.

FIG. 3 is a front view showing an installation example of a vehicle number automatic reading device.

FIG. 4 is a diagram for explaining a vertical profile extraction process and a vehicle head detection process, where (a) shows a contour image obtained by binarizing after edge detection, and (b) shows a vertical profile.

FIG. 5 is a flowchart for explaining speed measurement processing.

FIG. 6 is a diagram for explaining a plate frame cutout process based on a vehicle head position.

FIG. 7 is a diagram for explaining a method for converting a moving distance of a vehicle within a screen into an actual moving distance of a vehicle.

FIG. 8 is a diagram for explaining a correlation calculation in the second embodiment of the present invention, in which (a) and (b) show vertical profiles corresponding to two images captured at time intervals. Show.

FIG. 9 is a diagram for explaining a problem of the prior art.

[Explanation of symbols]

 1 Vehicle Number Automatic Reading Device 3 Camera Device 12 Edge Detection Circuit 13 Binarization Circuit 15 Image Processor 16 Profile Memory 17 General Purpose Processor

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁵ Identification code Internal reference number FI technical display location G08G 1/052 7001-3H

Claims (4)

[Claims] 1. A speed measuring device by image processing for measuring a speed of a running vehicle by processing an image taken by a photographing means for photographing a predetermined area on a road, the photographing means photographing the speed of the vehicle. Contour extraction means for extracting, as a contour pixel, a pixel of an edge portion whose density change rate of the image is a predetermined value or more, and a contour image constituted by the contour pixel intersects a direction corresponding to the traveling direction of the vehicle. Direction contour scanning means, counts each contour pixel existing on each scanning line, and creates a contour profile composed of a set of these count values, and a contour profile created by this contour profile creating means is stored. And a storage unit for performing the image capturing, and the image capturing unit captured the image after the image capturing time of the image corresponding to the contour profile stored in the storage unit. The correlation calculation means for performing a correlation calculation between the contour profile corresponding to the image and the contour profile stored in the storage means, and the above-mentioned 2 based on the result of the correlation calculation of the two contour profiles in the correlation calculation means. Moving distance calculating means for calculating the moving distance of the vehicle between the photographing times of the images corresponding to the two contour profiles, the photographing time difference between the images corresponding to the two contour profiles, and the moving distance calculating means. And a speed calculation means for calculating the speed of the vehicle based on the moving distance. 2. The vehicle head position calculation for calculating the vehicle head position in the image captured by the image capturing means by comparing any one of the contour profiles to be subjected to the correlation calculation with a predetermined threshold value. Means, means for extracting a partial profile that is a part of the one contour profile in the vicinity of the obtained vehicle head position, pattern matching between the extracted partial profile and the other contour profile, and the other contour And a matching unit that finds a matching portion that best matches the profile, wherein the moving distance calculating unit calculates the moving distance of the vehicle based on the shift amount between the partial profile and the matching portion. An image processing speed measuring device. 3. A speed measuring method by image processing for measuring a speed of a running vehicle by processing an image taken by a photographing means for photographing a predetermined area on a road, which is photographed by the photographing means. The pixels of the edge portion whose density change rate of the image is a predetermined value or more are extracted as contour pixels, and the contour image formed by the contour pixels is scanned in a direction intersecting the direction corresponding to the traveling direction of the vehicle. , The contour pixels existing on each scanning line are respectively counted, and a contour profile composed of a set of these count values is created, and a first contour profile created for two images captured at time intervals. Calculation is performed between the first contour profile and the second contour profile, and each image corresponding to the first contour profile and the second contour profile is photographed based on the result of the correlation calculation. The moving distance of the vehicle between the times is calculated, and the speed of the vehicle is calculated based on the difference between the photographing times of the images corresponding to the first contour profile and the second contour profile and the calculated moving distance of the vehicle. A method for measuring speed by image processing, characterized in that 4. The correlation calculation is obtained by comparing any one of the contour profiles to be subjected to the correlation calculation with a predetermined threshold value to calculate the vehicle head position in the image photographed by the photographing means. A partial profile that is a part of the one contour profile in the vicinity of the vehicle head position is extracted, pattern matching between the extracted partial profile and the other contour profile is performed, and matching that best matches the other contour profile is performed. A speed measuring method by image processing, wherein the moving distance of the vehicle is calculated based on a shift amount between the partial profile and the matching portion.