JP5134432B2 - Vehicle discrimination device and vehicle discrimination method - Google Patents

Description

  The present invention relates to a vehicle discrimination device and a vehicle discrimination method for discriminating a specific classification vehicle from a license plate of the vehicle.

In parking lots and expressways, the usage charges generally differ depending on the type of vehicle (classification such as light, small, normal, and large). There is a need to discriminate vehicles belonging to (hereinafter referred to as “specific classification vehicles”).
Conventionally, as a means of discriminating a specific classification vehicle, an image of a vehicle type code (lower case) displayed on a license plate is imaged, and a specific classification vehicle (for example, a light vehicle) is discriminated by identifying the imaged vehicle type code There is. However, since some of the registered automobiles and light automobiles have the same vehicle type code payout numbers, it is difficult to accurately discriminate a light automobile from the vehicle type code alone.

  In addition, as a means for discriminating a specific classification vehicle, there is a type in which vehicle specifications such as the number of axes and tread (vehicle width) are measured by a sensor, and whether or not the vehicle is a light vehicle is determined based on these measurement results. However, in recent years, for example, imported vehicles exceeding the tread width specified by laws and regulations may be registered as light vehicles, so it is difficult to accurately determine whether the vehicle is a light vehicle based on vehicle specifications such as the tread width. .

Therefore, there has been proposed a technique in which a license plate is imaged by a color camera and whether or not the vehicle is a light vehicle is determined based on the background color of the license plate (for example, see Patent Document 1). That is, the background color of the license plate is determined to be yellow if it is a light vehicle for private use, white if it is not a light vehicle for private use, black if it is a light vehicle for business use, and green if it is not a light vehicle for business use. Therefore, it is possible to determine whether the vehicle is a light vehicle by the color of the license plates.
Japanese Patent No. 3926673

  However, when color determination using a color camera is performed as in the vehicle discrimination device described in Patent Document 1 described above, a white light source is required, and thus the driver of the vehicle is dazzled by white light. There is a problem of giving.

  The present invention has been made in view of such circumstances, and is a vehicle determination device and a vehicle that can allow a driver to pass comfortably and quickly and accurately determine a specific classification vehicle. An object is to provide a discrimination method.

In order to solve the above problems, the present invention provides the following means.
The vehicle discriminating apparatus according to the present invention captures an imaging region, which is a predetermined virtual plane region, and acquires monochrome image data including a license plate of a vehicle passing through the imaging region, and the imaging A light distribution center calculation unit for calculating a light distribution center in the imaging region based on luminance information of the license plate in the monochrome image data acquired by the unit and position information of the license plate in the monochrome image data; Correction for correcting the brightness information of the license plate in the monochrome image data to be determined based on the position information of the license plate in the target monochrome image data and the light distribution center calculated by the light distribution center calculation unit And the monochrome image data to be discriminated corrected by the correction unit. Based on the luminance information of the license plate, characterized in that it comprises a discriminator for discriminating the specific classification vehicle.

  In addition, the vehicle determination device according to the present invention includes a determination function calculation unit that calculates a determination function based on an explanatory variable correlated with the specific classification vehicle, and the determination unit is the determination target corrected by the correction unit. The specific classification vehicle is discriminated based on the discriminant function calculated by the discriminant function calculator from the luminance information of the license plate.

  In the vehicle determination device according to the present invention, the explanatory variable includes a luminance value of a background portion of the license plate.

  Moreover, the vehicle discrimination device according to the present invention is characterized in that the explanatory variable includes a luminance value of a character portion of the license plate.

In addition, the vehicle determination method according to the present invention captures an imaging region that is a predetermined virtual plane region, and acquires monochrome image data including a license plate of a vehicle that passes through the imaging region . A light distribution center calculating step for calculating a light distribution center in the imaging region based on luminance information of the license plate in the monochrome image data acquired by the imaging step and position information of the license plate in the monochrome image data; The brightness information of the license plate in the monochrome image data to be determined is corrected based on the position information of the license plate in the monochrome image data to be determined and the light distribution center calculated in the light distribution center calculation step. And a correction step to be corrected by the correction step. It was based on the luminance information of the license plate in the monochrome image data of the determination target, characterized by comprising a determination step of determining the specific classification vehicle.

  According to the present invention, it is possible for a driver to pass comfortably, and a specific classification vehicle can be determined quickly and with high accuracy.

(Embodiment)
Hereinafter, a vehicle determination device according to an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 shows a vehicle discrimination device as an embodiment of the present invention.
The vehicle discriminating apparatus 1 according to the present embodiment discriminates a personal light vehicle (number plate: yellow, letter: black). This vehicle discriminating device 1 is installed beside the road R on which the vehicle C travels.
Further, on both sides of the road R, vehicle detection sensors 2 that detect the presence or absence of the vehicle C are provided. The vehicle detection sensor 2 is composed of, for example, an optical sensor or a loop coil. The vehicle detection sensor 2 is connected to the vehicle determination device 1. The vehicle detection sensor 2 includes a column 2a installed on both sides of the road R, a sensor unit (not shown) attached to the column 2a, and a detection control unit (not shown) connected to the sensor unit. It has.
And if the vehicle C drive | works between the support | pillars 2a, for example, a sensor part will detect that the light for a detection between the support | pillars 2a was interrupted, and will output an ON signal to a detection control part. When the detection control unit reads the ON signal, the detection control unit outputs a trigger signal to the vehicle determination device 1.

As shown in FIG. 2, the vehicle discrimination device 1 includes a control unit (light distribution center calculation unit, correction unit, discrimination unit, discrimination function calculation unit) 10 that controls the overall operation, and a storage unit that stores various information. 11 and an imaging unit 12 that captures an imaging target. In addition, the vehicle determination device 1 includes an output unit 13 that outputs a calculation result of the control unit 10, an information input unit 14 that receives input of various information, and a signal input unit that serves as an interface between the vehicle detection sensor 2 and the control unit 10. 15.
The signal input unit 15 receives the trigger signal output from the vehicle detection sensor 2 and inputs the trigger signal to the control unit 10.
The imaging unit 12 is composed of a black and white camera and does not use white light. Further, the imaging unit 12 captures an image of a single plane region through which the vehicle C passes based on an instruction from the control unit 10. This area is defined as an imaging area S (shown in FIG. 1). The imaging area S is an area extending between the columns 2a and extending in a direction orthogonal to the traveling direction of the vehicle C. The imaging unit 12 images the license plate N of the vehicle C by imaging the imaging region S. Further, the imaging unit 12 captures the imaging region S to generate monochrome image data of the imaging region S, and outputs the monochrome image data to the control unit 10.

  The control unit 10 reads the monochrome image data output from the imaging unit 12, and acquires the position information of the license plate in the monochrome image data from the monochrome image data by image processing. The position information at this time assumes the center point of the license plate. The imaging unit 12 captures an image area S, and the image area S and the monochrome image data have the same positional relationship. Therefore, the position information in the monochrome image data becomes position information in the imaging region S. Note that the positional relationship between the imaging region S and the monochrome image data does not need to be completely coincident, and for example, the positional coordinates of the two in a predetermined region may correspond to each other.

Further, the control unit 10 extracts only the luminance information of the background excluding the character information from the license plate in the monochrome image data, and calculates the luminance average. Then, the average brightness and the position information of the license plate are stored in the storage unit 11. When the luminance information and position information for a predetermined number of samples (for example, 1,000 vehicles) are obtained, the control unit 10 distributes the information in the imaging region S from the luminance information and position information for the 1,000 samples. The optical center is calculated and stored in the storage unit 11.
Further, the control unit 10 acquires the position information of the license plate in the monochrome image data of the vehicle C to be actually determined, and corrects the luminance information of the license plate of the vehicle C to be determined from the position information and the light distribution center. .
And the control part 10 discriminate | determines a specific classification | category vehicle from the corrected brightness | luminance information based on a discriminant function.
Details of the processing of the control unit 10 will be described later.

The output unit 13 outputs the determination result of the control unit 10 to another aggregation analyzer.
The information input unit 14 receives input of explanatory variables (to be described later) for calculating the discriminant function, and outputs data of these explanatory variables to the control unit 10. The explanatory variable data is stored in the storage unit 11.

Next, operation | movement of the vehicle discrimination device 1 in this embodiment comprised in this way is demonstrated.
The operation of the vehicle discrimination device 1 includes three phases: a light distribution center calculation phase, a luminance information correction phase, and a discrimination phase based on a discrimination function.
Below, these three phases are demonstrated in order.

(Light distribution center calculation phase)
The light distribution center calculation phase is performed in advance before the luminance information correction phase. The calculation of the light distribution center is performed after acquiring data for a predetermined number of samples such as 1,000.
FIG. 3 is a flowchart showing processing of the control unit 10 of the vehicle discrimination device 1 in the light distribution center calculation phase.
First, when the front end portion of the vehicle C enters between the columns 2 a and blocks the detection light, the vehicle detection sensor 2 outputs a trigger signal to the vehicle determination device 1. The control unit 10 of the vehicle determination device 1 reads the output of the signal input unit 15 and determines whether or not a trigger signal is output (step S1). When determining that the trigger signal is not output from the signal input unit 15, the control unit 10 repeats the process of step S1. On the other hand, when the control unit 10 determines that the trigger signal has been output, the control unit 10 images the imaging region S via the imaging unit 12 (step S2). Thereby, a monochrome image including the license plate of the vehicle C is obtained.

When the monochrome image output from the imaging unit 12 is read, the control unit 10 calculates position information that is the center position of a sample license plate (hereinafter referred to as “sample number plate”) from the luminance of the monochrome image. And stored in the storage unit 11 (step S3). This position information is a position in the imaging region S. Further, the control unit 10 calculates the average luminance information of only the background by excluding the character luminance from the luminance information of the sample license plate, and stores it in the storage unit 11 in association with the position information (step S4).
Then, the control unit 10 determines whether or not processing for a predetermined number of samples (for example, 1,000 times) has been performed using a counter or the like (step S5), and when it is determined that processing has not been performed, the control unit 10 returns to step S1. Repeat the process. On the other hand, when the control unit 10 determines that the processing for the predetermined number of samples has been performed (exceeds the number of samples), the light distribution in the imaging region S is obtained from the position information and the average luminance information for the number of samples from the storage unit 11. The center is calculated (step S6).

Now, the light distribution center calculation method in this embodiment pays attention to a subtle difference in white and black tones between a white plate and a yellow plate.
The white background of the license plate is the white plate, the yellow plate is the yellow plate, the green plate is the green plate, and the black plate is the blackboard. Further, the white plate and the yellow plate are white, and the green plate and the blackboard are black.
Since the brightness of the background of the license plate is greatly different between the white system and the black system, the white system and the black system can be easily distinguished by setting a predetermined threshold value. That is, if the control unit 10 determines that the background luminance of the license plate to be determined is equal to or less than a predetermined threshold, the control plate 10 immediately determines that the license plate is a black car and is not a private light vehicle. be able to. In the present embodiment, when the control unit 10 determines that the background brightness of the license plate is equal to or higher than a predetermined threshold (when it is a white system), the control unit 10 determines the yellow plate with high accuracy. .

When the white plate and the yellow plate in the white system are imaged under exactly the same conditions (iris value, CCD, exposure time, gamma curve, illumination light distribution, etc.), there are differences in the background portion and character portion luminance. That is, the white board is brighter than the yellow board.
However, since the brightness in the picture is greatly affected by the number of shots on the license plate, the yellow plate often appears brighter than the white plate depending on the orientation of the plate and the weather. In particular, industrial license plate recognition devices generally have spot-like illumination light distribution due to cost constraints. Accordingly, since the illumination light distribution in the image is not uniform, a change in luminance due to the position of the license plate in the imaging region cannot be ignored. For this reason, it is impossible to accurately discriminate between a white plate and a yellow plate by simple plate luminance comparison.

FIG. 4 shows a change in luminance in the picture in 8 bits by photographing a normal white board in an open space where there is no illumination reflection from other than the floor, and changing the position of the white board within the photographing area. Is.
As shown in FIG. 4, in the case of general illumination, the luminance of the object in the picture becomes darker as it deviates from the center of the illumination light distribution (portion from 220 bits), and the magnitude of the luminance change is the white plate at the same position. It is relatively larger than the luminance difference obtained when shooting a yellow plate and a yellow plate. For this reason, there is a possibility that the white plate is mistakenly identified as a yellow plate or the yellow plate is mistakenly identified as a white plate depending on the position of the light distribution center or the license plate in the imaging region.
In the vehicle discriminating apparatus 1 in the present embodiment, the influence of the illumination light distribution is removed before the determination of the yellow plate and the white plate.
In order to remove the influence of the illumination light distribution, it is necessary to quantitatively know the relationship between the illumination light distribution and the in-picture luminance. The illumination light distribution is determined at the time of illumination design and can be obtained approximately in advance. However, since the geometric conditions of the imaging system differ depending on the installation location of the vehicle discriminating apparatus 1 or a subtle difference occurs due to variations in installation adjustment, in this embodiment, correction is performed online with the actual machine. .

5 and 6 are graphs showing the relationship between the position information of the sample license plate and the average luminance information of only the background. The horizontal axis in FIG. 5 corresponds to the horizontal position of the imaging region S, and the horizontal axis in FIG. 6 corresponds to the vertical position of the imaging region S. 5 and 6, one dot indicates the center position of one sample number plate in the imaging region S in the horizontal direction and the vertical direction.
The control unit 10 stores 1,000 samples in the storage unit 11, and based on the position information of the 1,000 license plate samples stored in the storage unit 11 and the average luminance information of only the background. The light distribution center is calculated as follows.

First, the control unit 10 calculates the provisional value of the light distribution center (x, y) to be obtained by calculating the illumination light distribution center (x ₀ , y ₀ ) in the imaging region S at the design time inputted in advance by the operator. Set as. Then, the control unit 10 calculates a quadratic approximation function (P1 = f (P2): P1 is the average luminance of the position information and the average luminance information by regression calculation from the position information and the average luminance information stored in the storage unit 11. Information, P2 is position information).
Note that the curves in FIGS. 5 and 6 indicate the quadratic approximation function.
Further, the control unit 10 calculates the position having the highest average luminance, that is, the light distribution center (x, y) by calculating f ′ (P2) = 0 using the quadratic approximation function, and stores it. Store in unit 11.

In addition, it turns out that the light distribution center in FIG.5 and FIG.6 exists around (640.256).
The above quadratic approximation function shows a case where the illumination distribution is unimodal. However, when there are a plurality of light distribution centers, bimodal and multimodality are obtained. May be used. Strictly speaking, there is a method for obtaining a peak by differentiating an approximate phase of the “xyz (luminance) three-dimensional graph”, but here, the above-described method is adopted in consideration of the calculation amount.

(Luminance information correction phase)
Next, the luminance information correction phase will be described.
The luminance information correction phase is performed in advance before the discrimination phase by the discrimination function.
Further, the luminance information correction phase is based on the light distribution center obtained in the light distribution center calculation phase and the position of the discrimination target license plate (hereinafter referred to as “discrimination target license plate”) imaged by the imaging unit 12. The change in luminance of the discrimination target license plate is estimated and corrected quantitatively.

FIG. 7 is a flowchart showing the processing of the control unit 10 of the vehicle discrimination device 1 in the luminance information correction phase.
First, in the same manner as described above, the control unit 10 acquires a monochrome image of the determination target license plate of the vehicle C to be determined by the imaging unit 12.
Then, the controller 10 calculates the position information of the discrimination target license plate in the imaging region S in the same manner as in step S3 (step S20), and the average brightness of only the background of the discrimination target license plate in the same manner as in step S4. Information is calculated (step S21).

Then, the control unit 10 reads the light distribution center stored in the storage unit 11 in the light distribution center calculation phase, and only the background of the determination target license plate is based on the light distribution center and the position information of the determination target license plate. Are corrected (step S22).
Here, the correction of the average luminance information will be described.
FIG. 8 is an explanatory diagram modeling the relationship between illumination light distribution and luminance attenuation.
A symbol M in FIG. 8 indicates an isoluminance line centered on the light distribution center. The reason why the isoluminance line M is not symmetric is because it is assumed that the imaging area S is imaged from an oblique direction. In oblique imaging, the illumination density on the opposite side of the left and right sides (on the right side in normal installation) is low. Also, the upper and lower sides are not symmetrical because there is reflection from the road surface. Therefore, it is modeled not with a spot illumination attenuation model (x-y2 axis, vertical and horizontal symmetry) but with four independent vertical and horizontal axes from the light distribution center.

Assume that the position of the discrimination target license plate is P with respect to the light distribution center (x, y) in the imaging region S.
The in-picture luminance p at this position P can be obtained by the following equation.
p (p ₀ , r, l, d, u) = k ₁ r + k ₂ l + k ₃ d + k ₄ u + k ₅ r ² + k ₆ l ² + k ₇ d ² + k ₈ u ² + p ₀ (1)
In equation (1), variables r, l, d, and u indicate the vertical and horizontal divergence distances from the light distribution center, and k _{1 to} k ₈ are negative values that indicate the strength of attenuation due to the divergence distance. The constant term, p ₀ , indicates the in-picture luminance of the discrimination target license plate at the light distribution center.
Moreover, the regression formula of Formula (1) is not one way. Since the image processor is oblique imaging, the model is made with four independent axes here, but another attenuation model exists and can be handled in the same manner.
Further, in consideration of the possible deviation of the center of light distribution, this attenuation model (Equation (1)) is a two-dimensional model that follows both the first and second powers of the distance. When the estimated position of the light distribution center is deviated from the original center, the attenuation terms k _{5 to} k ₈ concerning the square of the distance are negative, and conversely, the first power terms k _{1 to} k ₄ are positive and can be absorbed. .

Furthermore, equation (1) is a one-dimensional multiple regression model, and if there is constant sample data, it can be solved by multiple regression analysis. The control unit 10 solves the multiple regression problem from the position information (r, l, du) and the average luminance information (p) for a predetermined number of samples stored in the storage unit 11, and k ₁ to k ₈ and to calculate the p _0. And the control part 10 correct | amends the average luminance information p of the discrimination | determination object license plate by the following formula | equation.
p ₀ = p− (k ₁ r + k ₂ l + k ₃ d + k ₄ u + k ₅ r ² + k ₆ l ² + k ₇ d ² + k ₈ u ² ) (2)
That is, the average luminance information p of the discrimination target license plate is converted (luminance normalization) into the in-picture luminance p ₀ when it is assumed that the image is captured at the center of the light distribution.
Note that the control unit 10 performs the above normalization process (conversion from p to p ₀ ) not only on the background part of the discrimination target license plate but also on the average luminance information of the character part.

(Determination phase by discriminant function)
Next, the discrimination phase by the discrimination function will be described.
By the light distribution center calculation phase and the luminance information correction phase, the influence of the light distribution can be subtracted from the average luminance information of the discrimination target license plate.
However, the luminance may fluctuate due to many factors other than the illumination light distribution.
Here, although there are totally unknown factors such as local weather changes (temporary shading) as factors affecting the brightness, they are known (can be estimated or can be sensed) and should be processed appropriately. In some cases, the effect can be subtracted (for example, plate attitude angle, body color tone (diffuse reflection effect), global weather change (day and night, etc.)).
However, in order to deduct the impact of these known factors individually, the way of the impact must be evaluated in advance, so when considering the effort and implementation for deduction of the impact, Not realistic.
Therefore, in the present embodiment, a specific classification vehicle is determined using a “discriminant analysis method” which is a statistical method.
In the present embodiment, by defining a factor group correlated with “light car-likeness” as “explanatory variable”, it is possible to make a judgment by comprehensive judgment without deducting in consideration of individual influence methods. .

First, before identifying a specific classification vehicle, data in which the actual affiliation (light car or other) is clearly known is prepared as a sample group. For example, when data collected in advance is used as a sample group (offline sample), the correct answer (light car or other) is confirmed by visual confirmation or the like. In addition, when data obtained in real time is used as a sample group (online sample), only data that can be reliably determined whether it is a light vehicle or any other method, such as sensing of vehicle specifications, is added to the sample group.
Then, when those sample groups are input via the information input unit 14 of the vehicle discrimination device 1 or the like, the control unit 10 stores the sample groups in the storage unit 11. Furthermore, as shown in FIG. 9, explanatory variables are defined for the sample group. In FIG. 9, samples 1 to N and explanatory variables 1 to M are prepared.

Here, the explanatory variable is a factor group correlated with “light car-likeness”, and is an amount that directly or indirectly suggests a group to which the vehicle belongs (non-light car / other than light car). That is, these explanatory variables are used as weights in the specific classification vehicle determination.
Examples of explanatory variables include the following.
(A) A quantity that directly suggests a light car. The brightness value of the background of the license plate In a white system, if it is a light car for private use, the background color is yellow, otherwise it is white, so if it is a light car, the brightness of the background will be dark If it is not a light car, the brightness will be brighter (light (dark) ← → other than light (bright)).
2. The brightness value of the license plate character part In the white system, the color of the character part is black if it is a light car for private use, and the color of the character part is dark if it is a light car because it is green otherwise. If it is not a light car, the brightness will be brighter (light (dark) ← → other than light (bright)).
3. Presence / absence of emblems For example, certain automobile companies have automobile companies that manufacture more mini cars than large and ordinary cars. In this case, by identifying the emblem of the car company, the establishment of a light car increases (light (no emblem) ← → other than light (with emblem)).

(B) A quantity that indirectly suggests a light car (an amount that indirectly affects the judgment of a light car)
1. Plate tilt When the plate is tilted, it appears dark and easy to appear. That is, if the luminance value is the same, the probability of a white board (other than a light vehicle) is higher as the inclination is larger.
2. Estimated extraneous light The plate becomes darker and darker at night. That is, if the luminance value is the same, the probability of a white board (other than a light vehicle) is higher as the estimated extraneous light amount (which can be estimated in the captured time zone) is smaller.
In addition to the above examples, there are many candidates for explanatory variables that suggest the appearance / inappropriateness of minicars, but when adopting them as explanatory variables, there is a correlation with the affiliation group (minicars / other than minicars). It may be determined by evaluating the strength and weakness.

As described above, when the explanatory variable is input, the control unit 10 calculates the discriminant function in advance as follows and stores it in the storage unit 11.
Here, it is assumed that two luminance values of the background portion and the character portion are used as explanatory variables.
The control unit 10 calculates a discriminant function for discriminating the second group from the sample group according to a statistical method. As a calculation method, there are a method using least squares (multiple regression line (linear discriminant function), shown in FIG. 10) and a method using Mahalanobis distance (intermediate curve (nonlinear discriminant function), shown in FIG. 11).

Then, as described above, with the discrimination function stored in the storage unit 11, the control unit 10 determines whether or not the discrimination target vehicle is a specific classification vehicle as follows.
First, when the control unit 10 reads the average luminance information corrected in the luminance information correction phase, the control unit 10 reads the discriminant function stored in the storage unit 11. Then, as shown in FIG. 10, the control unit 10 calculates the position P1 in the sample group of the average luminance information to be discriminated, and based on the discriminant function, whether the position P1 belongs to the A group other than the light vehicle, It determines whether it belongs to the B group of a light vehicle. Since the position P1 belongs to the group B, the control unit 10 determines that the vehicle to be determined is a private light vehicle and outputs a determination result via the output unit 13.

As described above, according to the vehicle discrimination device 1 of the present embodiment, the average luminance of the discrimination target license plate is normalized based on the center of light distribution, so even a black and white camera can accurately determine whether it is a white plate or a yellow plate. Can be determined. Therefore, the driver can pass comfortably without using white light, and the specific classification vehicle can be discriminated quickly and with high accuracy.
In addition, since the weighting is performed by the explanatory variable when the specific classification vehicle is determined, the determination accuracy can be improved. At this time, it is not necessary to quantitatively clarify or subtract the ripple effect of individual explanatory variables.
Further, since the luminance value of the background portion of the license plate and the luminance value of the character portion are used as the explanatory variables, it is possible to improve the determination accuracy by using a particularly highly correlated value.
Although the method of setting appropriate explanatory variables depends on the design of the device, it is possible to obtain a discrimination accuracy of about 90% using several explanatory variables including the factors exemplified above. (Reference: In the method of simply discriminating from the threshold value of the plate portion luminance, the discrimination accuracy was about 60%).

In the above-described embodiment, the discrimination target is a personal light vehicle. However, the present invention is not limited to this, and it may be a private vehicle or a commercial vehicle, a large vehicle, or the like. That is, the yellow plate is discriminated from the white system, but the white plate may be discriminated, or the blackboard or the green plate may be discriminated from the black system. For example, since the blackboard has a lower luminance than the green plate if it is black, the blackboard can be determined from the black as in the above embodiment.
The technical scope of the present invention is not limited to the above embodiment, and various modifications can be made without departing from the spirit of the present invention.

It is explanatory drawing which shows a mode that embodiment of the vehicle discrimination device which concerns on this invention was installed in the road. It is a block diagram which shows a vehicle discrimination device for every function. It is a flowchart which shows the process of the vehicle discrimination device in a light distribution center calculation phase. This is an open space where there is no illumination reflection from other than the floor surface, and shows the distribution of illumination light distribution and luminance change when shooting a normal white board. It is a graph which shows the relationship between the positional information of a sample number plate, and the average luminance information of only a background, and shows a mode when the horizontal axis is made to respond | correspond to the position of the horizontal direction of an imaging region. It is a graph which shows the relationship between the position information of a sample number plate, and the average luminance information of only a background, and shows a mode when the horizontal axis is made to respond | correspond to the position of the vertical direction of an imaging region. It is a flowchart which shows the process of the vehicle discrimination device in a brightness | luminance information correction phase. It is explanatory drawing which models and shows the relationship between illumination light distribution and brightness | luminance attenuation | damping. It is a table | surface which shows the relationship between an explanatory variable value and a sample group. It is explanatory drawing which shows the positional relationship between the discriminant function calculated by the least square, and a sample group. It is explanatory drawing which shows the positional relationship between the discriminant function calculated by Mahalanobis distance, and a sample group.

Explanation of symbols

1 vehicle discriminating apparatus 10 control unit (light distribution center computing unit, correction unit, discriminating unit, discriminant function computing unit)
12 Imaging unit C Vehicle S Imaging region

Claims (5)

A vehicle discrimination device for discriminating a specific classification vehicle from a vehicle license plate,
An imaging unit that captures an imaging region that is a predetermined virtual plane region and acquires monochrome image data including a license plate of the vehicle that passes through the imaging region;
A light distribution center calculation unit for calculating a light distribution center in the imaging region based on luminance information of the license plate in the monochrome image data acquired by the imaging unit and position information of the license plate in the monochrome image data; ,
Based on the position information of the license plate in the monochrome image data to be determined and the light distribution center calculated by the light distribution center calculation unit, the luminance information of the license plate in the monochrome image data to be determined is corrected. A correction unit;
A vehicle discrimination device comprising: a discrimination unit that discriminates the specific classification vehicle based on luminance information of the license plate in the monochrome image data to be discriminated corrected by the correction unit. A discriminant function computing unit that computes a discriminant function according to an explanatory variable correlated with the specific classification vehicle,
The discrimination unit is
The specific classification vehicle is discriminated based on a discriminant function calculated by the discriminant function calculator from brightness information of the license plate to be discriminated corrected by the corrector. The vehicle discriminating device described. The explanatory variables are
The vehicle determination device according to claim 2, comprising a luminance value of a background portion of the license plate. The explanatory variables are
The vehicle determination device according to claim 2, further comprising a luminance value of a character portion of the license plate. A vehicle discrimination method for discriminating a specific classification vehicle from a vehicle license plate,
An imaging step of imaging an imaging area, which is a predetermined virtual plane area, and obtaining monochrome image data including a license plate of the vehicle passing through the imaging area ;
A light distribution center calculating step for calculating a light distribution center in the imaging region based on luminance information of the license plate in the monochrome image data acquired by the imaging step and position information of the license plate in the monochrome image data; ,
Based on the position information of the license plate in the monochrome image data to be determined and the light distribution center calculated in the light distribution center calculation step, the luminance information of the license plate in the monochrome image data to be determined is corrected. A correction step;
A vehicle determination method comprising: a determination step of determining the specific classified vehicle based on luminance information of the license plate in the monochrome image data to be determined corrected in the correction step.

Images