JPH10208045A - Vehicle entering detector - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a vehicle detector which does not have so much error detection due to sunshine change, fast detects a vehicle and has high reliability. SOLUTION: This device is provided with an image pickup device 1 which picks up an image in the direction of an entrance, image memory 2 that stores an image that is outputted by the device 1, reference memory 3 which registers an image when no vehicles exist, a correlation arithmetic circuit 5 which divides the contents of the memory 3 and the contents of the memory 2 into a vertical rectangular area and a horizontal rectangular area and calculates a normalized correlation value, binary circuits 61 and 62 which binarize an output of an arithmetic circuit of the circuit 5 by a threshold that is preliminarily defined, and AND circuit 7 which performs an AND operation of an output of the circuits 61 and 62 and a characteristic extraction circuit 8 that inputs a signal from the circuit 7. The circuit 7 judges a difference point between the memory 2 and 3 as an AND part of a low area of each normalized correlation value of the vertical rectangular area and the horizontal rectangular area and detects it.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for detecting a vehicle at a parking lot or a tollgate on a toll road.

[0002]

2. Description of the Related Art The prior art is shown in FIGS. FIG.
1 is a structural diagram of a first example of a vehicle detection device according to the related art. FIG. 5 is an explanatory diagram of a process of a first example of the vehicle detection device according to the related art.

FIG. 6 is a structural diagram of a second example of a vehicle detection device according to the prior art. FIG. 7 is an explanatory diagram of a process of a second example of the vehicle detection device according to the related art. In parking lots and toll roads, it is detected that a vehicle has entered the gate, and the entrance gate is opened, or a toll issuance is issued, or an instruction to start the operation of a toll settlement machine is issued. It is necessary to output information to the device.

In addition, it is necessary to measure the size, such as the area, length, width, etc., of the vehicle entering the vehicle, determine the type of vehicle, and change the type of toll ticket to be issued and the fee to be settled. As a vehicle detection device for this purpose, a system using an ultrasonic sensor, a system using an electromagnetic wave, a photoelectric sensor, and the like have been conventionally used. Here, an image processing system will be described.

FIG. 4 is a configuration diagram of a first embodiment of a vehicle detection device according to the prior art. The image pickup device 1 includes a vehicle 10 entering the lane 12 and a background 11. Is imaged.

[0006] The captured image is digitized and stored in the image memory 2. In the reference memory 3, an image in which only the background 11 when the vehicle 10 does not exist is registered in advance.

[0007] The contents of the image memory 2 and the reference memory 3 are input to a difference calculation circuit 21 and the absolute value of the difference is calculated for each pixel of the two input images. The output of the difference operation circuit 21 is
An area of a pixel which is input to the binarization circuit 6 and whose absolute value of the difference is larger than a preset threshold is detected.

[0008] The output of the binarization circuit 6 is input to a feature extraction circuit 8. The feature extraction circuit 8 measures features output from the binarization circuit 6, such as the area of the vehicle region, the circumscribed rectangle, and the position of the center of gravity.

The output of the feature extracting circuit 8 is displayed on a display 9 or output to a higher-level device by a communication means (not shown) to control a charge machine. The operation of the present apparatus will be described in more detail with reference to FIG.

When an image of the vehicle is taken by the imaging device 1 from the front of the vehicle 10 as shown in FIG. 4, as shown in FIG.
An image including 0 and the background 11 is captured. This input image is stored in the image memory 2.

In the reference memory 3, an image of only the background 11 as shown in FIG. 5B is registered in advance. The difference calculation circuit 21 calculates the absolute value of the difference between the contents of the image memory 2 and the contents of the reference memory 3 for each pixel.

The background 11 on the image is the same image, and the difference is calculated to be 0. However,
When the vehicle 10 enters, the density of the pixel in that part changes, and when the difference is obtained, some value is output.

Here, the case where the paint color of the vehicle 10 is white,
In the case of black, since the sign of the difference value is reversed, the absolute value of the difference value indicates an output of a certain value or more only in the area of the vehicle 10.

The output of the difference calculation circuit 21 is compared with a preset threshold value in the binarization circuit 6, and a pixel having a threshold value or more outputs "H" and a pixel having a value less than the threshold value outputs "L". As a result, an image in which only the area of the vehicle 10 becomes “H” as shown in FIG. 5C is obtained.

The feature extraction circuit 8 obtains feature amounts of the vehicle 10 area as shown in FIG. The features are shown in FIG.
(C) upper and lower ends j _{min of} the vehicle 10 area in the height direction,
The circumscribed rectangle such as j _max and the left and right ends i _min and i _max in the horizontal direction, the area, that is, the number of pixels in the “H” region, and the like are used.

The output of the feature extraction circuit 8 is displayed to an operator on a display 9 to determine a vehicle type, open and close a gate, and the like. In the first conventional example, the difference calculation circuit 21 is used for detecting the vehicle 10. However, in the case of a difference, if the density of the entire image changes due to a change in sunshine conditions, all the areas are changed areas, that is, It may be determined that the vehicle 10 is used.

In the second conventional example shown in FIG. 6, a correlation operation circuit 22 is used instead of the difference operation circuit 21. The other parts are the same as those of the first conventional example, and therefore description thereof is omitted, and only the operation of the correlation operation circuit 22 is described.
This will be described with reference to FIG.

The correlation calculation circuit 5 divides the image memory 2 and the reference memory 3 into several rectangular areas in exactly the same manner as shown in FIGS. 7 (a) and 7 (b). Image memory 2
Then, a normalized correlation value represented by the following equation is calculated for each corresponding rectangular area of the reference memory 3.

Normalized correlation value = Σ (fg−μν) / σπN Expression (1) where f: density data of each pixel in each rectangular area of the image memory 2 g: each pixel in each rectangular area of the reference memory 3 Density data μ: average value of density data in each rectangular area of the image memory 2 ν: average value of density data in each rectangular area of the reference memory 3 σ: standard of density data within each rectangular area of the image memory 2 Deviation π: Standard deviation of density data in each rectangular area of reference memory 3 N: Number of data in each rectangular area This is obtained by normalizing the density of each rectangular area by the standard deviation of density to obtain a difference. Even if there is a change in the density of the entire image due to changes in the sunshine conditions, etc., a region with a small change in the density distribution is unaffected by a large correlation value and a region with a large change in the density distribution is It shows a small correlation value.

As a result, as shown in FIG.
Only the rectangular area including the zero area indicates a small correlation value, and the binarization circuit 6 reversely binarizes the area below the threshold to “H” and the area above the threshold to “L”. Can be detected.

[0021]

However, the prior art has the following problems. (1) In a vehicle detection device using a difference calculation circuit according to the prior art, when the density of the entire image changes due to a change in sunshine or the like, the entire image is determined to be a vehicle area regardless of the presence or absence of a vehicle, and reliability is improved. Low. (2) In the vehicle detection device using the correlation calculation circuit according to the related art, in order to increase the resolving power of the finally obtained vehicle detection result, the size of the rectangular area for performing the normalized correlation calculation is reduced. In this case, since the number of pixels in the rectangular area is small, when the density is normalized, the correlation value in the background area may be reduced due to noise on the image, or the correlation operation may not be performed. Since there is no characteristic density change area to perform, the reliability of the correlation result is low. (3) Conversely, in order to improve the reliability of the correlation calculation, the rectangular area must be enlarged, the resolution of the finally obtained vehicle area deteriorates, and the feature amount used for vehicle type determination is reduced. There was a problem that accuracy was worsened. An object of the present invention is to provide a device that can solve these problems.

[0022]

[Means for Solving the Problems]

(First Means) A vehicle detection device according to the present invention is a device for detecting the entry of a vehicle into a gate at a tollgate on a parking lot or a toll road, or measuring the shape of the vehicle. An image pickup device 1 for picking up an image, (B) an image memory 2 for storing an image output from the image pickup device, and (C) a reference memory 3 for registering an image when no vehicle is present.
(D) a vertical correlation operation circuit 4 for dividing the content of the reference memory 3 and the content of the image memory 2 into a vertically long rectangular area 13a and obtaining a normalized correlation value of each area;
And the contents of the image memory 2 are divided into horizontally long rectangular areas 13b, and the outputs of the horizontal correlation operation circuit 5 for obtaining the normalized correlation value of each area and (F) the output of the operation circuit of the vertical correlation operation circuit 4 are obtained in advance. A first binarizing circuit 61 for binarizing with a set threshold, and a second binarizing circuit (G) for binarizing the output of the arithmetic circuit of the horizontal correlation arithmetic circuit 5 with a preset threshold. And (H) an AND circuit 7 for performing an AND operation on the outputs of the first and second binarization circuits 61 and 62
And (I) a feature extraction circuit 8 for inputting a signal from a logical product circuit 7. (J) The logical product circuit 7 compares a difference between the image memory 2 and the reference memory 3 with a vertically long rectangular area and a horizontally long rectangular area. It is characterized in that it is determined as a logical product of the regions with low normalized correlation values of the rectangular regions, and the detection is performed with high resolution and high reliability.

(Second Means) In the vehicle detecting apparatus according to the present invention, in the first means, the vertical correlation operation circuit 4, the horizontal operation correlation circuit 5, and the first means for binarizing the output of the operation circuit. The binarization circuit 61 and the second binarization circuit 62 are operated at the same time to detect at high speed.

That is, in order to solve the problems of the prior art, the present invention provides an image pickup apparatus 1 for picking up an image in an entrance direction, an image memory 2 for storing an image output from the image pickup apparatus, A reference memory 3 for registering an image when not to be stored, a vertical correlation operation circuit 4 for dividing the content of the reference memory and the content of the image memory into a vertically long rectangular area and obtaining a normalized correlation value of each area, and a content of the reference memory And a horizontal correlation operation circuit 5 for dividing the contents of the image memory into a horizontally long rectangular area to obtain a normalized correlation value of each area, and two binary circuits for binarizing the output of each operation circuit with a preset threshold value. It has value converting circuits 61 and 62 and a logical product circuit 7 for performing a logical product operation of the outputs of the two binarizing circuits.

Therefore, the operation is as follows. In the above configuration, the vertical correlation operation circuit 4 and the horizontal correlation operation circuit 5 divide the contents of the image memory and the reference memory into a vertically long rectangular area and a horizontally long rectangular area, respectively, to obtain a normalized correlation value.

At this time, the size of the rectangular area is such that the short side is long enough to correspond to the resolution of the finally required vehicle area, and the long side is reliable for the normalized correlation operation of each rectangular area. To a size that can be used.

The outputs of the vertical correlation operation circuit 4 and the horizontal correlation operation circuit 5 are compared with threshold values set in advance by the corresponding binarization circuits. Is converted into an inverted value so that it becomes "H", and the logical product of both outputs is obtained by the logical product circuit 7 to obtain a resolution corresponding to the short side of the rectangular area and the short side and the long side. It is possible to perform detection with a sufficiently reliable normalized correlation value calculated with the product of the sides, that is, the number of pixels corresponding to the area of the rectangular region.

[0028]

BEST MODE FOR CARRYING OUT THE INVENTION

(First Embodiment) FIG. 1 shows a first embodiment of the present invention.
3 to FIG. FIG. 1 is a configuration diagram of a vehicle detection device according to a first embodiment of the present invention.

FIG. 2 is an explanatory diagram of a correlation calculation area for vehicle detection according to the first embodiment. FIG. 3 is an explanatory diagram of a logical product operation for determining a vehicle area from the result of the correlation operation.
The components having the same functions as those of the related art are denoted by the same reference numerals, and description thereof is omitted.

In FIG. 1, an image including a vehicle 10 and a background 11 captured by an imaging device 1 is digitized and stored in an image memory 2.

In the reference memory 3, FIG.
An image of only the background 11 without the vehicle 10 as shown in (d) is registered in advance. The vertical correlation operation circuit 4 compares the contents of the image memory 2 and the contents of the reference memory 3 with each other as shown in FIG.
As shown in (c), the image is divided into vertically long rectangular areas, and the respective normalized correlation values are calculated.

At this time, the short side of the rectangular area is set to a length corresponding to the resolution of the finally required vehicle area, and the long side is set to a size large enough to obtain the reliability of the normalized correlation calculation of each rectangular area. And

For example, if it is assumed that the reliability of the normalized correlation value can be ensured by using the area of 8 × 8 pixels in the normalized correlation operation according to the prior art, here, 2 × 32 pixels and 32 × 2 pixels The normalized correlation operation is performed in the region of.

As a result, since the data having the same number of pixels as in the conventional method is used, the reliability of the obtained normalized correlation value can be said to be substantially the same. In addition, since the position of each rectangular area does not change for each image and may always be the same, the position of each rectangular area is stored in a storage element, and data in that area is read.

The output of the vertical correlation operation circuit 4 is input to a first binarization circuit 61. The first correlation circuit 61 compares the input normalized correlation value with a preset threshold value,
"L" if it is equal to or more than the threshold, "H" if less than the threshold
Is output.

Similarly, the horizontal correlation operation circuit 5 compares the contents of the image memory 2 and the contents of the reference memory 3 with each other as shown in FIGS.
As shown in (1), after dividing into a horizontally long rectangular area and calculating the respective normalized correlation values, the second binarization circuit 62 performs inversion binarization.

The vertical correlation operation circuit 4 and the first binarization circuit 61
And the processing by the horizontal correlation operation circuit 5 and the second binarization circuit 62 can be simultaneously executed in parallel to increase the speed.

The outputs of the first and second binarization circuits 61 and 62 are as shown in FIGS. 2 and 3 (e) and (f). In the figure, hatching is applied only to the “H” region.

The AND circuit 7 calculates the logical product of the binarized images output from the first and second binarization circuits 61 and 62. As a result of the logical product, as shown in FIG. 3 (g), the region indicated by the thick line is the “H” region of the binarized image output from the first and second binarization circuits 61 and 62. Yes, only the hatched area is output.

As is apparent from FIG. 3, the final result of vehicle detection is detected at a resolution corresponding to the square formed by the short sides of the vertically long rectangular area and the horizontally long rectangular area. For example,
If the calculation of the normalized correlation value is performed using the area of 8 × 8 pixels by the method according to the related art, the resolution of the input image can be obtained only by １ ／ times in both the vertical and horizontal directions.
By performing a normalized correlation operation in the area of × 32 pixels and 32 × 2 pixels, the input image is divided by two pixels, which are the short sides of the rectangular area in both the vertical and horizontal directions, and the resolution of the vehicle is reduced by half. A detection result will be obtained. When the vehicle detection result of FIG. 3G is obtained, the feature amount on the image is extracted by the feature extraction circuit 8 and displayed on the display 9 as in the conventional case.

[0041]

Since the present invention is configured as described above, it has the following effects. (1) Since the image processing for detecting the vehicle performs processing by the normalized correlation calculation, erroneous detection due to a change in sunlight is small. (2) Since the number of pixels for calculating the normalized correlation value can be secured by a sufficiently high reliability, the obtained normalized correlation value has high reliability. (3) Since the resolution of the vehicle region finally obtained is the resolution corresponding to the short side of the rectangular region, the resolution of the feature amount on the image measured using the detection result of the present invention also increases, and the vehicle type It is possible to improve the reliability in the case where it is used for a judgment or the like. (4) The vertical correlation operation and the horizontal correlation operation can be performed in parallel, and the speed can be increased.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a vehicle detection device according to a first embodiment of the present invention.

FIG. 2 is an explanatory diagram of a correlation calculation area for vehicle detection according to the first embodiment of the present invention.

FIG. 3 is an explanatory diagram of an AND operation for determining a vehicle area from a correlation operation result.

FIG. 4 is a structural diagram of a first example of a vehicle detection device according to the related art.

FIG. 5 is an explanatory diagram of a processing procedure of a first example of a vehicle detection device according to the related art.

FIG. 6 is a structural diagram of a second example of a vehicle detection device according to the related art.

FIG. 7 is an explanatory diagram of a process of a second example of the vehicle detection device according to the related art.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Image pick-up device 2 ... Image memory 3 ... Reference memory 4 ... Vertical correlation operation circuit 5 ... Horizontal correlation operation circuit 6 ... Binarization circuit 7 ... Logical product circuit 8 ... Feature extraction circuit 9 ... Display 10 ... Vehicle 11 ... Background DESCRIPTION OF SYMBOLS 12 ... Lane 13a ... Vertical rectangular area 13b ... Horizontal rectangular area 21 ... Difference operation circuit 22 ... Correlation operation circuit 61 ... 1st binarization circuit 62 ... 2nd binarization circuit

Claims (2)

[Claims] An apparatus for detecting an entry of a vehicle into a gate at a tollgate on a parking lot or a toll road, or measuring a shape of the vehicle, (A) an imaging apparatus for capturing an image in an entrance direction. (B) an image memory (2) for storing an image output from the imaging device, (C) a reference memory (3) for registering an image when no vehicle is present, and (D) a reference memory (3). A vertical correlation operation circuit (4) that divides the contents and the contents of the image memory (2) into a vertically long rectangular area (13a) and obtains a normalized correlation value of each area, and (E) the contents of a reference memory (3) And the contents of the image memory (2) are divided into horizontally long rectangular areas (13b), and a horizontal correlation operation circuit (5) for obtaining a normalized correlation value of each area, and (F) a vertical correlation operation circuit (4) First binary for binarizing the output of the arithmetic circuit with a preset threshold Circuit (61), and (G) a second binarizing circuit for binarizing the preset threshold the output of the arithmetic circuit of the horizontal correlation computing circuit (5) (62), (H)
The first binarization circuit (61) and the second binarization circuit (6
An AND circuit (7) for performing an AND operation on the output of 2);
(I) a feature extraction circuit (8) for inputting a signal from an AND circuit (7); and (J) the AND circuit (7) is different from an image memory (2) and a reference memory (3). A vehicle detection device, wherein a point is determined and detected as a logical product of areas having a low normalized correlation value in each of a vertically-long rectangular area and a horizontally-long rectangular area. 2. A vertical correlation operation circuit (4) and a horizontal operation correlation circuit (5), and a first binarization circuit (61) and a second binarization circuit for binarizing an output of the operation circuit. 2. The vehicle detecting device according to claim 1, wherein (62) is detected at a high speed by calculating at the same time.