JP2002008186A - Vehicle type identification device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a vehicle type identification device capable of further increasing car model identification accuracy and being adjusted easily. SOLUTION: This vehicle type identification device 1 comprises a profile line extraction part 15a extracting a profile line by differentiating the image data of a vehicle to be identified, a feature quantity extraction part 15b extracting a feature quantity by calculating the frequency of component in a specified extracting direction of the profile line in the specified area, a storage part 13 storing a reference feature quantity and the information indicating a vehicle type in the state of being related to each other, and a feature quantity matching part 15c identifying the vehicle type by matching the feature quantity with the reference feature quantity.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a vehicle type identification device for identifying the type of various vehicles traveling on a road.

[0002]

2. Description of the Related Art Conventionally, a vehicle type identification device for identifying the type of various vehicles traveling on a road has been proposed and put into practical use. In particular, in recent years, ITS (Intelligent Transp
ort System) and ETC (Electronic T
oll Collection System) is being promoted. In such a system, it is necessary to reliably identify the vehicle type in order to automatically collect tolls on expressways. There is a growing demand for building.

[0003] Such a vehicle type discriminating apparatus radiates an ultrasonic wave or a microwave to a vehicle and uses a reflected wave thereof, or obtains an image of the vehicle and analyzes it by image processing to identify the vehicle type. What to do has been suggested. Among them, those using ultrasonic waves may cause erroneous recognition due to the influence of wind or the like, and those using microwaves have problems such as being regulated by the Radio Law.
Therefore, expectations for a vehicle type identification device by image processing, in which such a problem is relatively unlikely to occur, are increasing.

Conventionally, in a vehicle type identification apparatus using such image processing, a vehicle image acquired by a camera is
The binarization based on the luminance close to the luminance of the background such as the road surface separates the vehicle from the background. Then, the apparent size of the vehicle is determined based on the binarized image, and the vehicle type is identified by comparing the apparent size with known sizes of various vehicles.

[0005]

However, in such a conventional vehicle type discriminating apparatus based on image processing, especially when a part of the vehicle body is close to the luminance of a background such as a road surface, its apparent size is reduced. In some cases, it could not be obtained correctly, causing erroneous recognition. For example, when the luminance of the bed portion of a large vehicle is close to the luminance of the background, even if the image of the vehicle is binarized, the bed portion of the vehicle and the background cannot be distinguished from each other. Therefore, the apparent size of the vehicle is determined as a size obtained by removing the bed portion, and a large vehicle may be erroneously recognized as a normal vehicle.

[0006] Further, in the conventional vehicle type identification apparatus based on image processing, there is a problem that the adjustment requires time. That is, regarding the threshold value of luminance used for binarization,
Since the operator or the like has determined and input an appropriate value based on experience and the like in consideration of the identification time zone such as daytime and dusk, it takes time to determine this threshold, and as a result, the vehicle type identification device It took time to adjust.

[0007] The present invention has been made in view of the above problems, and can further improve the vehicle type identification accuracy.
It is an object of the present invention to provide a vehicle type identification device that can be easily adjusted.

[0008]

In order to achieve the above object, a vehicle type identification apparatus according to a first aspect of the present invention differentiates a contour line of a vehicle to be identified by differentiating image data of the vehicle to be identified. A contour extracting means for extracting, and a feature extracting means for calculating a frequency of a predetermined extraction direction component in a predetermined region of the contour extracted by the contour extracting means, thereby extracting a feature of the contour. And storage means for storing, in advance, a reference feature value of the vehicle serving as an identification reference and information indicating the vehicle type of the vehicle in association with each other, and a feature value extracted by the feature value extraction means. And a feature value matching means for matching the reference feature value stored in the storage means with each other to identify the type of the vehicle to be identified.

According to this configuration, the contour line of the vehicle is extracted by the contour line extracting means, and the characteristic amount of the contour line is extracted by the characteristic amount extracting means. Then, the feature amount matching unit performs matching between the feature amount and a reference feature amount acquired in advance to identify the type of the vehicle. Therefore, the vehicle type is identified based on the feature amount of the outline of each part of the vehicle, not just the apparent size as in the related art. Therefore, the vehicle type can be determined without considering the difference in brightness between the vehicle and the background. Can be identified, and the accuracy of vehicle type identification can be improved.

[0010] The vehicle type identification device according to claim 2 is
2. The vehicle type identification device according to claim 1, wherein a reference contour extraction unit extracts a reference contour of the vehicle by differentiating the image data of the vehicle serving as the identification reference,
Calculating a frequency of a predetermined extraction direction component in a predetermined region of the reference contour extracted by the reference contour extraction means, thereby extracting a reference feature amount of the reference contour; Input means for inputting information indicating the type of the vehicle serving as an identification reference.

According to this configuration, the reference contour of the vehicle serving as the identification reference is extracted by the reference contour extracting means, and the reference feature of the reference contour is extracted by the reference feature extracting means. On the other hand, information indicating the type of the vehicle serving as the identification reference is input via the input unit. Therefore, by storing the reference feature amount and the information indicating the vehicle type in the storage unit, it is possible to create data serving as an identification reference. For this reason, when installing a new vehicle type identification device,
If the installation environment of the vehicle type identification device or the road surface conditions change,
Since the operator or the like can create identification reference data by judging only the type of the vehicle, it is not necessary to set a difficult numerical value such as determination of a luminance threshold unlike the related art. It will be easier.

Further, the vehicle type identification device according to claim 3 is
3. The vehicle type identification device according to claim 1, wherein the storage unit stores a plurality of groups of reference feature amounts classified according to conditions when acquiring image data of the vehicle, and the storage unit stores the reference feature amounts. And a selection unit for selecting a reference feature amount to be used for matching by the feature amount matching unit from among the plurality of groups of reference feature amounts.

[0013] According to this configuration, the reference feature to be used for matching can be selected from the plurality of groups of reference feature stored in the storage by the selector.
Therefore, as in the case of daytime or twilight, reference feature amounts for different conditions can be acquired in advance, and the reference feature amounts can be switched and used according to the conditions at the time of actual identification. For this reason, it becomes easy to use a more appropriate reference feature amount according to the condition, and the identification accuracy is further improved. Further, the setting of the reference feature amount is facilitated, and the adjustment of the vehicle type identification device is further facilitated.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a vehicle type identification device (this device) according to the present invention will be described below in detail with reference to the drawings. The present invention is not limited by the embodiment.

(Embodiment 1) FIG. 1 is a block diagram showing a configuration of a traffic monitoring system using the present apparatus according to Embodiment 1, and FIG. 2 is a block diagram showing an entire configuration of the present apparatus of FIG. . In the present embodiment, a vehicle type capable of identifying a vehicle type by roughly extracting a characteristic amount related to a contour of a vehicle and matching the characteristic amount with a characteristic amount of each vehicle acquired in advance. It relates to an identification device.

First, the overall configuration of the traffic monitoring system shown in FIG. 1 will be described. This traffic monitoring system uses the device 1
, A camera 2, a continuous image input device 3, and a traffic monitoring device 4 are connected as shown in the figure. The camera 2 is installed around an arbitrary road surface, such as near a tollgate on a highway, continuously captures images of vehicles traveling on the road surface, and outputs digitized continuous image data to the continuous image input device 3. Is what you do. The configuration of the camera 2 is arbitrary, and for example, an ITV camera can be used. The continuous image input device 3 captures a continuous image output from the camera 2 and outputs it as black and white discontinuous image data.

The traffic monitoring device 4 outputs data (identification result data) indicating the vehicle type identification result output from the device 1.
And performs various traffic monitoring controls based on the identification result data and any other data. Examples of such a traffic monitoring device 4 include a device that performs a toll settlement process on an expressway based on identification result data, and a device that monitors traffic conditions based on identification result data and vehicle speed data. it can. The traffic monitoring device 4 selects the timing of capturing images from the continuous image input device 3 to the present device 1 and selects an image to be identified by the vehicle.

A plurality of traffic monitoring systems configured as described above can be arranged according to the number of vehicles to be identified, lanes, and the like. Note that the correspondence between the present apparatus 1 and other system components is not limited to the illustrated one. For example, identification result data from a plurality of the present apparatuses 1 is taken into one traffic monitoring apparatus 4 to perform centralized monitoring. Can also be performed. Further, the traffic monitoring device 4 can be integrally formed with the present device 1. In addition, the system configuration including the device 1 can be arbitrarily modified without being limited to the configuration in FIG. 1, and the device 1 can be used for any purpose.

Next, the configuration of the apparatus 1 of FIG. 2 will be described. As shown in FIG. 2, the present device 1 includes an input device 1
0, input IF (IF = interface) 11, output I
F12, the storage unit 13, and the control unit 14 are connected by a communication path such as a bus. The input device 10 is an input unit for inputting various data. As this data, for example, vehicle type data at the time of acquiring the reference feature amount can be cited (the contents will be described later). The configuration of the input device 10 is arbitrary, and for example, can be configured as a numeric keypad or a keyboard. The input IF 11 is connected to the continuous image input device 3.
The image data input from the input device and the input data input from the input device 10 are taken into the device 1. Output IF
Reference numeral 12 denotes the identification result data output from the control unit 14,
Output to the traffic monitoring device 4.

The storage unit 13 is a storage means for storing various data and programs, and at least associates the reference feature value extracted from the image data of each vehicle with the vehicle type data of each vehicle. In a non-volatile state. The specific configuration of the storage unit 13 is arbitrary, and for example, a ROM (Read Only Memory), a flash memory, or a hard disk can be used. In addition, the storage unit 13 may be configured as an external storage device that can read the contents of a storage medium storing the reference feature amounts so that the reference feature amounts can be easily changed according to road surface conditions and the like. it can.

The control section 14 is control means for controlling each section of the apparatus 1 and includes an identification section 15 and a reference data acquisition section 16. Among them, the identification unit 1
Reference numeral 5 denotes an identification means for identifying the vehicle, which includes an outline extraction unit 15a, a feature amount extraction unit 15b, and a feature amount matching unit 15c. In addition, the reference data acquisition unit 16 includes a reference contour line extraction unit 16a and a reference feature value extraction unit 16b.

First, each component of the identification unit 15 will be described. The contour extraction unit 15a is a contour extraction unit that extracts a contour of image data. Specifically, the contour line extraction unit 15a performs a differentiation process using a differentiation filter on the image data captured via the input IF 11, and outputs a contour line (a predetermined direction) of the vehicle included in the image data. An outline of the entire vehicle body and outlines of vehicle body components such as a bumper and a windshield included in the outline are extracted.

This differentiation processing is performed by a known method. Specifically, a predetermined differentiation filter (difference filter) is configured for the luminance value of each pixel of the image data or a block of a plurality of pixels. This is performed by calculating the product sum with the value of each pixel.

FIG. 3 shows a configuration example of such a differential filter.
It is shown in FIG. Among them, FIG. 3 shows a primary spatial differential filter for extracting a contour line in the vertical direction, and subtracts and adds together the luminance values of the pixels located on both sides of the column of the center pixel, thereby obtaining a horizontal direction. Are emphasized and extracted (= vertical contour). FIG. 4 shows a primary spatial differential filter for extracting a horizontal contour, which emphasizes and extracts a portion where the degree of change in the vertical direction is large (= horizontal contour). In addition, by using a plurality of differential filters, at least a contour line in a predetermined extraction direction described later is extracted. The differential filter is not limited to those shown in FIGS. 3 and 4, and for example, a secondary spatial differential filter such as a Laplacian filter can be used. Further, the number of dimensions of the differential filter is not limited to 3 × 3 as shown in FIGS.

The outline of the contour extracted by the contour extracting unit 15a will be described. FIG. 5 is a diagram illustrating extraction of a contour line of image data of a normal vehicle, and FIG. 6 is a diagram illustrating extraction of a contour line of image data of a large vehicle. As shown by the arrows in FIGS. 5 and 6, in the image data area, the outlines of the bumper, hood, window glass, and vehicle body ceiling of each vehicle are extracted.

Also, in FIG. 2, the feature amount extraction unit 15b
Is a feature amount extracting means for extracting the feature amount of the contour line.
Specifically, the feature amount extraction unit 15b includes the contour line extraction unit 15
A predetermined plurality of continuous areas are set for the contour extracted in a, and the frequency of the predetermined extraction direction component of the contour included in each area is calculated, so that the feature amount of the contour is calculated. Extract. The setting of the continuous area is performed by dividing the image data into n1 × n2 uniform matrix areas that are continuous with each other. Although the number of divisions n1 × n2 is arbitrary, the larger the number of divisions, the more the feature amount can be extracted, but the processing load increases. Therefore, an appropriate number of divisions may be set in consideration of these points. preferable. For example, FIGS. 5 and 6 show examples in which the number of divisions is set to 3 × 3. In FIGS. 5 and 6, the area divided by the dotted line is a continuous area.

The predetermined extraction direction is an arbitrary number of directions set for each of the continuous areas. Although the number n3 of the predetermined extraction directions is arbitrary, it is preferable to set an appropriate number of directions in consideration of the same points as the number of divisions. Further, the angle of each extraction direction is also arbitrary, but in the present embodiment, the angles are set to be equal to each other. That is, as shown in the lower part of FIG. 5, for each region, a horizontal direction D1, a direction D2 at 30 degrees to the horizontal direction D1, a direction D3 at 60 degrees to the horizontal direction D1, and a vertical direction D4 Are set (the number of directions n3 = 4). From these facts, the feature amount is calculated as the number n1 × n
2 (n1 × n2 × n)
It is expressed in the dimension of 3).

The feature amount matching unit 15c matches the feature amount extracted by the feature amount extracting unit 15b with each of the plurality of reference feature amounts stored in the storage unit 13, and performs the most matching. This is a feature amount matching unit that identifies the type of the vehicle to be identified by selecting a reference feature amount with a high rate. This matching can be performed using a known statistical method. For example, a minimum distance between patterns (for example, a Mahalanobis distance or a Levenshtein distance) between a feature amount and each reference feature amount is determined, and the distance is the shortest. Can be selected as the reference feature having the highest matching rate.

Next, each component of the reference data acquisition unit 16 will be described. First, the reference contour extraction unit 16a
Similar to the above-described contour line extraction unit 15a, it is a reference contour line extraction unit that extracts a contour line from image data of a vehicle. Such image data is obtained in advance by an administrator, a maintenance person, or the like of the apparatus 1 before actual identification starts. Alternatively, the image data acquired at the time of identifying the implementation can be used as appropriate. At this time, various vehicle data is selected from a large number of image data so that identification can be performed with high accuracy.
It is preferable to extract a contour line.

The reference feature extracting unit 16b is a reference feature extracting unit that extracts a feature from the contour extracted by the reference contour extracting unit 16a in the same manner as the feature extracting unit 15b described above. is there. That is, similarly to the feature value, the reference feature value is data expressed in the dimension of the product (n1 × n2 × n3) of the number of divisions n1 × n2 of the continuous area and the number n3 of directions in the predetermined extraction direction. . This reference feature amount extraction unit 16
The reference feature value extracted in b is stored in the storage unit 13 in association with the vehicle type data input by the input device 10.

Next, the processing contents of the present apparatus 1 will be described. In the present apparatus 1, reference data acquisition processing is performed in advance by the reference data acquisition unit 16, and thereafter, the identification unit 1
5 performs vehicle type identification processing. In this case, the vehicle is referred to as "large car", "regular car", or
An example is shown in which the vehicle is identified as “motorcycle”, and vehicles that could not be identified in any of them are classified as “unknown”. However, the present invention is not limited to this example, and the vehicle type can be identified in more detail.

First, reference data acquisition processing will be described. FIG. 7 is a flowchart of the reference data acquisition process. In the reference data acquisition processing, first, image data of one vehicle serving as a reference (reference setting vehicle) is taken into the reference contour line extraction unit 16a (step S7-).
1). This image data is obtained, for example, by photographing a vehicle running on the road with the camera 2 and taking in a continuous image output from the camera 2 with the continuous image input device 3 to obtain monochrome discontinuous image data. Can be obtained by: The acquisition timing of this image data is arbitrary, but can be determined by the traffic monitoring device 4. Alternatively, the timing can be determined by recording a continuous image captured as described above on a video tape and instructing the timing of generating a digital image from the video tape by a person. However, since this image data is only used for the reference setting, image data obtained by any other method can be used.

Then, a reference contour line is extracted by the reference contour line extraction unit 16a from the image data thus captured (step S7-2). As described above, the calculation of the reference contour line is performed by calculating the product sum of the luminance of each pixel or a block of a plurality of pixels of the image data and the value of each pixel constituting a predetermined differential filter. It is.

After that, the data of the contour line thus extracted is input to the reference feature value extraction unit 16b, and the reference feature value is calculated (step S7-3). Specifically, a continuous area is set for the contour extracted by the reference contour extracting unit 16a, and the frequency of a predetermined extraction direction component of the contour included in each area is calculated. The feature amount of the contour is used. For example, when the extraction direction components of the above four directions D1 to D4 are set, there are four contours in the direction D1, one contour in the direction D2, one contour in the direction D3, and When three contours in the direction D4 are included, each of these numbers is calculated as the frequency in each direction. Note that a contour line in a direction that does not completely match the extraction direction component can be handled as the contour line of the closest extraction direction component, or can be simply ignored.

Then, by calculating the frequency for all the regions, the feature amount can be calculated. It is expected that such feature amounts will differ depending on the vehicle type. Specifically, in the case of a large vehicle, since there are many bumpers and glass parts on the front part, there is a tendency that one or more contours in the direction (horizontal direction) D1 can be extracted in a wide area, while a motorcycle is In the case of, since the width is narrow, the contour line in the direction D1 tends to be hardly extracted in any region. Note that, in this case, identification can be similarly performed using not only the front surface but also the image of the rear surface of the vehicle.

Thereafter, the manager or the like visually identifies the vehicle type of the reference setting vehicle at an arbitrary timing or the like, and inputs predetermined vehicle type data corresponding to the vehicle type via the input device 10. The vehicle type data is stored in the storage unit 13 in a state where the vehicle type data is associated with the reference feature amount acquired in step S7-3 (step S7-4). Below, "large car",
Until the reference feature amount is obtained for all the types of the “normal automobile” and the “motorcycle”, a series of processes from step S7-1 to S7-4 is repeated (step S7-5). The data acquisition process ends.

Next, the vehicle type identification processing will be described. FIG. 8 is a flowchart of the vehicle type identification processing. In this vehicle type identification processing, first, image data of one vehicle to be identified (identification vehicle) is captured by the contour line extraction unit 15a (step S8-1). This image data can also be obtained by capturing the identification vehicle with the camera 2 and using the continuous image input device 3 as discontinuous image data, similarly to the image data of the reference setting vehicle. This fetch timing is determined by the traffic monitoring device 4. Then, a contour line is extracted from the image data thus captured by the contour line extraction unit 15a (step S8-2), and a feature value is calculated by the feature value extraction unit 15b (step S8-).
3). These correspond to steps S7-2 and S7-3, respectively.
Is performed in the same manner as described above.

Next, the feature value matching unit 15c matches the feature value of the vehicle for identification with the reference feature value of each reference setting vehicle stored in the storage unit 13 (step S8-4). Then, one of the reference feature amounts of each reference setting vehicle is determined, and one of the reference feature amounts closest to the feature amount of the identification vehicle is determined, and the vehicle type data stored in association with the reference feature amount is determined. It is obtained as identification result data indicating the vehicle type (step S8-5). However, as a result of the matching, if the reference feature amount closest to the feature amount of the identification vehicle cannot be determined for some reason, it is determined that the type of the identification vehicle is “unknown”.
Finally, the identification result data thus obtained is output to the traffic monitoring device 4 via the output IF 12 (step S8-6), and the vehicle type identification processing ends.

According to such an apparatus, since the type of the vehicle is identified based on the characteristic amount of the contour line of the vehicle instead of the apparent size as in the related art, the accuracy of the type identification can be improved. Can be. Further, since the operator or the like can create identification reference data by judging only the type of the vehicle, there is no need to perform difficult setting such as determination of a luminance threshold unlike the related art. Adjustment can be easily performed.

(Embodiment 2) FIG. 9 is a block diagram showing an overall configuration of a vehicle type identification apparatus according to Embodiment 2.
The present embodiment relates generally to a vehicle type identification device that stores a plurality of groups of reference feature amounts and can switch between them according to the situation. In addition, regarding Embodiment 2,
Configurations and processes that are not particularly described are the same as those of the first embodiment, and the same configurations and processes are denoted by the same reference numerals.

In the present embodiment, the storage section 13 stores first and second reference feature amounts and first and second vehicle type data. Among these, the first reference feature amount is
The first feature data is a reference feature amount acquired based on image data of a vehicle running in the daytime, and the first vehicle type data is data indicating a vehicle type corresponding to the first reference feature amount. Similarly, the second reference feature amount is a reference feature amount of a vehicle running at twilight, and the second vehicle type data is data indicating a vehicle type of the vehicle corresponding to the second reference feature amount. The reason why a plurality of reference feature amounts and vehicle type data are obtained in this way is that the amount of sunlight and the like differ depending on the identification time zone, and the number of contour lines that can be obtained by image processing may differ.

In this embodiment, the input device 10
Is provided with a selection unit 10a. This selector 10a
Is a selection unit for selecting a reference feature used for matching in the feature matching unit 15c from the first and second reference features. Specifically, the selection unit 1
0a can be arbitrarily configured as long as the reference feature value used for matching can be specified,
For example, it can be configured as a changeover switch or a numeric keypad.

In such a configuration, the operator of the apparatus 1 selects one of the first and second reference feature amounts by operating the selector 10a in accordance with the weather at the time of identifying the vehicle type. . Then, any one of the selected first and second reference feature amounts and the feature amount extracted by the feature amount extraction unit 15b are matched by the feature amount matching unit 15c. Then, vehicle type data corresponding to the matching result is obtained as identification result data from any of the first and second vehicle type data corresponding to the selected first and second reference feature amounts. Alternatively, the discrimination time zone is automatically determined by the selection unit 10a, and the reference feature value to be used is automatically selected from the first and second reference feature values in accordance with the time frame. Is also good.
According to such an apparatus, it is easy to use an appropriate reference feature amount according to conditions, and the identification accuracy is further improved. Further, the setting of the reference feature amount is facilitated, and the adjustment of the vehicle type identification device 1 is further facilitated.

[0044]

As described above, according to the vehicle type identification apparatus of the present invention (claim 1), the outline of the vehicle is extracted by the outline extraction means, and the outline is extracted by the feature quantity extraction means. Are extracted, and the two are matched by the feature amount matching means to identify the type of the vehicle. Therefore, instead of just the apparent size as before,
Since the type of the vehicle is identified based on the feature amount of the outline of the vehicle, the accuracy of the type identification can be improved.

According to the vehicle type identification apparatus of the present invention, the reference contour is extracted by the reference contour extracting means, and the reference feature is extracted by the reference feature extracting means. Information indicating the vehicle type of the reference vehicle is input via input means. Therefore, the operator or the like can create identification reference data by judging only the type of the vehicle, so that it is not necessary to perform difficult setting such as determination of a luminance threshold unlike the related art. Becomes easier.

Further, according to the vehicle type identification apparatus of the present invention (claim 3), a reference feature value to be used for matching is selected by the selecting device from a plurality of groups of reference feature values stored in the storage device. Therefore, it is easy to use an appropriate reference feature amount according to conditions, and the identification accuracy is further improved. Further, the setting of the reference feature amount is facilitated, and the adjustment of the vehicle type identification device is further facilitated.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a traffic monitoring system using a vehicle type identification device according to a first embodiment of the present invention.

FIG. 2 is a block diagram showing the overall configuration of the apparatus of FIG. 1;

FIG. 3 is a diagram illustrating a configuration example of a differential filter.

FIG. 4 is a diagram illustrating a configuration example of a differential filter.

FIG. 5 is a diagram illustrating contour line extraction and the like of image data of a normal vehicle.

FIG. 6 is a diagram showing contour line extraction and the like of image data of a large vehicle.

FIG. 7 is a flowchart of reference data acquisition processing.

FIG. 8 is a flowchart of a vehicle type identification process.

FIG. 9 is a block diagram illustrating an overall configuration of a vehicle type identification device according to a second embodiment of the present invention.

[Explanation of symbols]

 Reference Signs List 1 vehicle type identification device 2 camera 3 continuous image input device 4 traffic monitoring device 10 input device 10a selection unit 11 input IF 12 output IF 13 storage unit 14 control unit 15 identification unit 15a contour extraction unit 15b feature extraction unit 15c feature matching Unit 16 reference data acquisition unit 16a reference contour line extraction unit 16b reference feature value extraction unit

Claims (3)

[Claims] 1. A contour line extracting means for extracting a contour line of a vehicle to be identified by differentiating image data of the vehicle to be identified, and a contour region extracted by the contour line extracting means within a predetermined area of the contour line. A feature amount extracting means for extracting the feature amount of the contour line by calculating the frequency of the predetermined extraction direction component, and information indicating the reference feature amount of the vehicle serving as the identification criterion and the vehicle type of the vehicle acquired in advance A storage unit that stores the identification object in association with each other, and a feature amount extracted by the feature amount extraction unit and a reference feature amount stored by the storage unit. And a feature amount matching means for identifying a vehicle type of the vehicle. 2. A reference contour extracting means for extracting a reference contour of the vehicle by differentiating the image data of the vehicle serving as the identification reference, and a reference contour extracted by the reference contour extracting means. By calculating the frequency of a predetermined extraction direction component in a predetermined area of the line, a reference feature amount extraction unit for extracting a reference feature amount of the reference contour line, and information indicating a vehicle type of the vehicle serving as the identification reference are input. The vehicle type identification device according to claim 1, further comprising: an input unit for inputting the information. 3. The storage means stores a plurality of groups of reference feature amounts divided for each condition when acquiring image data of a vehicle, and stores the plurality of groups of reference feature amounts stored in the storage means. 3. The vehicle type identification device according to claim 1, further comprising a selection unit that selects a reference feature amount to be used for matching by the feature amount matching unit from the inside.