JPH11213284A - Vehicle kind discrimination device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately discriminate a vehicle kind by an inexpensive device by detecting the candidate areas of the plural constituting elements of a vehicle and discriminating the vehicle kind by using the constituting elements whose kinds are decided based on the information on the position relation. SOLUTION: A constituting element detection means 3 refers to a constituting element dictionary 2 storing the attributes of the constituting elements of the vehicle and detects the constituting elements of an image picked-up vehicle from input images stored in an image input means 1 based on the attributes of the constituting elements. A constituting element decision means 5 refers to a constituting element arrangement dictionary 4 storing the position relation of the constituting elements of the vehicle and decides the kinds of the constituting elements of the image picked-up vehicle from the candidate areas of the constituting elements which are the detected result of the constituting element detection means 3 based on the position relation of the constituting elements. A vehicle dictionary 6 stores the relating information of the constituting elements for the respective vehicle kinds for discriminating the constituting elements for the respective vehicle kinds beforehand. A vehicle kind discrimination means 7 collates the decided result of the constituting element decision means 5 and the vehicle dictionary 6 and discriminates the kind of the image picked-up vehicle based on the relating information of the constituting elements for the respective vehicle kinds.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a vehicle type discriminating apparatus for discriminating a type of a vehicle from a captured image of the vehicle.

[0002]

2. Description of the Related Art FIG.
It is a block diagram which shows the schematic structure of the conventional vehicle type discrimination apparatus described in 08. In the figure, reference numeral 1 denotes image input means for digitizing an image signal to create an input image; 101, a license plate recognition means for detecting the position of the license plate from the input image and recognizing characters of the license plate image; Vehicle width detection sensor for detecting width, 10
Reference numeral 3 denotes a vehicle type discriminating means for discriminating the vehicle type from the input image based on the detection result of the license plate position and the detection result of the vehicle width. This is a search unit that performs a collation search. More specifically, the image input unit 1 includes an image capturing unit that captures an image of a vehicle and a frame memory that stores the captured image, and the license plate recognizing unit 101 includes a license plate indexing unit, a character indexing unit, and a character recognizing unit. The vehicle type discriminating means 103 includes a feature extracting unit, a collating unit, and a dictionary memory.

Next, the operation will be described. First, image input means 1
Digitizes a signal input from a camera or the like, creates an input image, and stores it in a frame memory. Next, the license plate recognition means 101 detects a license plate from the input image stored in the frame memory and recognizes characters in the license plate. Further, the vehicle width detection sensor 102
Detects the width of the vehicle on which the vehicle travels. Vehicle width detection sensor 10
Reference numeral 2 includes a photodiode array disposed at a gate on the road, a pressure sensor laid on the road, and the like.

[0004] Then, the vehicle type discriminating means 103 extracts the features of the area (processing area) other than the license plate in the input image, and compares it with the contents of the dictionary memory to determine the vehicle type. The feature extraction unit determines a processing area based on the vehicle width information obtained by the vehicle width detection sensor 102 and the position information of the license plate detected by the license plate recognizing means 101, and determines the processing area as left / right / up / down. To obtain a distribution of the number of black and white changing points, the number of continuous black points and the number of white points, and obtain a characteristic value. The matching unit compares and compares the feature value with the feature dictionary for each vehicle type stored in the dictionary memory, and determines the best matching vehicle type as a determination result.

Finally, the search unit 104 uses the recognition result of the license plate character obtained by the license plate recognizing means 101 and the vehicle type discrimination result obtained by the vehicle type discriminating means 103 and is registered in advance for supplementation. Search for the vehicle that is located.

[0006]

However, in a conventional vehicle type discriminating apparatus, a processing area used for vehicle type discrimination is determined according to vehicle width information and license plate position information.
If the license plate cannot be detected correctly, there is a problem that the vehicle type cannot be determined. Further, since a dedicated vehicle width detection sensor is required, there is a problem that the system becomes expensive. Further, when an image of a vehicle entering obliquely is captured, distortion (rotation) occurs in the image, and the position of the processing region is shifted from a desired position, so that there is a problem that the vehicle type cannot be correctly identified. Further, when the position of the processing area is corrected in accordance with the degree of distortion of the image in order to avoid the distortion of the image, there is a problem that the correction cannot be performed only by the vehicle width information and the position information of the license plate.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problems, and it is possible to accurately identify a vehicle type even when a specific component of the vehicle, such as a license plate, cannot be detected correctly. It is an object of the present invention to realize a vehicle type discriminating apparatus that can discriminate a vehicle type. Another object of the present invention is to realize a vehicle type discriminating apparatus that can image a vehicle that has entered obliquely and can accurately determine the vehicle type even if the image has distortion.

[0008]

SUMMARY OF THE INVENTION A vehicle type discriminating apparatus according to the present invention includes a component dictionary in which the attributes of the components of a vehicle are stored in advance, and a vehicle based on the attributes of the components with reference to the component dictionary. A component detecting means for detecting a component of an imaged vehicle from an input image which is imaged, a component arrangement dictionary in which positional relationships of components of the vehicle are stored in advance, and a configuration referring to the component arrangement dictionary A component determining unit that determines the type of the component from the candidate region of the component of the imaging vehicle detected by the component detecting unit based on the positional relationship of the component, and a vehicle for determining the component for each vehicle type in advance. The vehicle dictionary in which the relevant information of the components for each vehicle type is stored, and the components determined by the component determining means and the vehicle dictionary are compared with each other, based on the relevant information of the components for each vehicle type. In which and a vehicle type discriminating means for discriminating the type of the image vehicle.

Further, in the vehicle type discriminating apparatus according to the next invention, when the component determining means determines the type of the component by referring to the component arrangement dictionary, the position of the component determined in advance as the reference component is determined. The vehicle center line is determined based on the vehicle center line.

A vehicle type discriminating apparatus according to the next invention comprises a license plate recognizing means for recognizing a license plate from an input image, and a component dictionary, a component layout dictionary, and a component dictionary according to the recognition result of the license plate recognizing means. Alternatively, the reference range of the vehicle dictionary is limited.

Further, in the vehicle type discriminating apparatus according to the next invention, the license plate recognizing means recognizes an attribute of the license plate.

Further, in the vehicle type discriminating apparatus according to the next invention, the license plate recognizing means recognizes the vehicle type code of the license plate.

A vehicle type discriminating apparatus according to the next invention comprises a mark recognizing means for recognizing a mark or a character on a vehicle body from an input image, and a component dictionary and a component arrangement according to the recognition result of the mark recognizing means. This limits the reference range of the dictionary and / or vehicle dictionary.

A vehicle type discriminating apparatus according to the next invention comprises a license plate recognizing means for recognizing a license plate from an input image, and a mark recognizing means for recognizing a mark or a character on a vehicle body from the input image. The reference range of the component dictionary, the component arrangement dictionary, and / or the vehicle dictionary is limited according to the recognition result of the recognition unit and the mark recognition unit.

[0015]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiment 1 Hereinafter, a first embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of the first embodiment. In FIG. 1, reference numeral 1 denotes an image input unit that captures an image of a vehicle and digitizes a captured image signal to create an input image and inputs an image of the vehicle; 2 denotes a component dictionary in which attributes of the components of the vehicle are stored in advance; Reference numeral 3 denotes a component detection unit that detects a component of the imaged vehicle from an input image stored in the image input unit 1 based on the attribute of the component with reference to the component dictionary 2. Reference numeral 4 denotes a position of a component of the vehicle in advance. The component arrangement dictionary 5 in which the relation is stored, and the component element detection means 3 refer to the component arrangement dictionary 4 and based on the positional relation of the components.
A component determining means for determining the type of the component of the imaged vehicle from the component candidate area which is the result of the detection. The component determining means 6 stores in advance the relevant information of the component for each vehicle type for determining the component for each vehicle type. The vehicle dictionary 7 is a vehicle type discriminating means for collating the determination result of the component determining means 5 with the vehicle dictionary 6 and discriminating the type of the imaged vehicle based on the relevant information of the component for each vehicle type.

Next, the operation will be described with reference to FIGS.
FIG. 2 is a flowchart showing the flow of the operation of the present embodiment. FIG. 3 shows an input image stored in the image input means, which is an image of the front of the vehicle. FIG. 4 is an edge image created by performing edge detection from an input image.

The operation of the vehicle type discriminating apparatus is started, for example, by detecting that the vehicle has arrived within the imaging range of the camera. First, the image input means 1 inputs an image (step S1). Similar to the conventional example, for example, a camera captures an image of a vehicle in response to the arrival of the vehicle, digitizes an input signal of the vehicle, creates an input image, and stores the input image in a frame memory.

Next, the component detecting means 3 detects the components of the vehicle with reference to the component dictionary 2 (step S).
2). The constituent elements are individual areas forming a vehicle image. In the present embodiment, four types of windshield, headlight, turn signal, and license plate are targeted.

In the component detection, first, a candidate area having a possibility of a component is obtained from an input image. For example,
"License plate region extraction using pyramid hierarchical hi-speed Hough transform" (Transactions of the Institute of Electronics, Information and Communication Engineers, D, Vol.
J70-D, No. 7, pp. 1383 to 1389), an input image is subjected to Sobel conversion (Sobel conversion) and then binarized to create an edge image, and a region surrounded by the edge can be a component. And a candidate area having a certainty. FIG. 4 is an edge image created from the input image of FIG. 1, and is a diagram showing candidate regions for constituent elements.
In the present embodiment, it is assumed that 8 to 20 in FIG. 4 have been detected as the candidate regions of the component.

Next, the component detection means 3 refers to the component dictionary 2 to determine which component each candidate area is based on the attribute of the component, and determines any one of the four types of components. A flag indicating the component is added to the candidate area thus set, and the result is output as a result of component detection. The component dictionary 2 stores, for each component, an attribute of each component, for example, a maximum value and a minimum value of a color and a width / height.
The component detection unit 3 compares and compares the attribute of each candidate area that may be a component with the attribute in the component dictionary 2.
If there is a match, a flag indicating the component corresponding to the candidate area is added and output.

In this embodiment, a plurality of component flags may be added to one candidate area. Further, for components such as door mirrors having different colors depending on the vehicle, the determination based on the color attribute may not be performed. For example, the color attribute in the component dictionary 2 is set as a “collation exclusion attribute”, and the component is not determined based on the color attribute. Other components may also be excluded from matching for specific attributes according to their characteristics. In the present embodiment, among the candidate areas in FIG. 4, 9 is a windshield, 10 and 11 are door mirrors, 1
2 and 13 are headlights, 17 and 18 are turn signals, 1
It is assumed that 2 and 19 are determined to be license plates.

Next, the component determining means 5 refers to the component arrangement dictionary 4 and determines the components of the image pickup vehicle from the component candidate area detected by the component detecting means 3 based on the positional relationship of the components. The type is determined (step S3). The component arrangement dictionary 4 stores information indicating the positional relationship between the components of the vehicle. For example, for each component, as a relative position with respect to other components, the information that the windshield is above the headlight, the license plate is below the headlight, and the image between the components is The maximum value / minimum value of the distance is stored. The component determining means 5 determines the type of the vehicle component by comparing and comparing the combination of the components with the contents of the component arrangement dictionary 4.

When this step is completed, the flag of each component (flag indicating the type of component) is determined to be one. For example, in the present embodiment, the component detecting means 3 uses the license plates 12 and 19 in the image of FIG. 4, but the information in the component layout dictionary 4 indicates that “the license plate is below the headlight”. In this step, it is determined that 19 is a license plate.

Finally, the vehicle type discriminating means 7 compares the components of the imaged vehicle determined by the component determining means 5 with the vehicle dictionary 6 to determine the type of the vehicle in the input image (step S).
4). The vehicle dictionary 6 includes information related to components of each vehicle type,
For example, the shape (roundness, square, etc.), the size on the image (width, height, etc.), and the distance between components in the image are stored. This is compared with the constituent elements determined by the constituent element determining means 5 to determine the best matching vehicle type as the determination result, and the operation of the vehicle type determination ends.

As described above, the vehicle is imaged by the image input means 1 in response to the arrival of the vehicle, the input signal is digitized to create an input image, and is stored in the frame memory. Next, a candidate region of a component of the vehicle is detected from the input image stored in the component detection unit 3 based on attributes such as a maximum value and a minimum value of color and width / height. Further, the type of the component of the vehicle is determined by the component determining means 5 based on the positional relationship such as the relative position between the components and the distance between the components on the image. Finally, the determined component is compared with the related information of the component for each vehicle type stored in the vehicle dictionary, the size on the image, and the like, to determine the imaging vehicle type. Thereafter, for example, the determined vehicle is displayed on the display screen together with the vehicle type name, so that the determination result can be confirmed.

As described above, according to the vehicle type discriminating apparatus according to the present embodiment, the candidate areas of a plurality of constituent elements of the vehicle are detected, and the constituent elements whose types are determined based on the information on their positional relationship are determined. By using this to determine the vehicle type, each component of the vehicle can be correctly detected, and the vehicle type can be accurately determined with an inexpensive device. Further, even if there is a component that cannot be detected at all, the vehicle type can be determined with high accuracy from information on other components. For example, even if a specific component such as a license plate cannot be detected, the vehicle type can be determined from information on other headlights, windshields, and the like.

In the first embodiment, the input image is an image of the front of the vehicle. However, the input image is not limited to this as long as the image of the vehicle is captured. For example, a similar effect can be obtained for a rear surface image, a side image, and a top surface. In that case, the components may include other components of the vehicle such as a rear window and a tail lamp. Further, a plurality of images may be used as an input image. For example, the vehicle type may be determined using the front image and the side image of the vehicle as the input image. In this case, the processing time of the vehicle type determination may increase, but the accuracy of the vehicle type determination is improved.

Also, as a method of detecting a component candidate region in the component detecting means, a case has been described in which an edge image is created and a region surrounded by an edge is detected as a component candidate region. Not something. Another method may be used as long as the component can be detected from the input image, and the same effect can be obtained. Further, a plurality of methods may be used, or a different method may be used for each component.

Although the description has been given of the case where the components of the vehicle are a windshield, a headlight, a turn signal, and a license plate, other speed indicators such as a front grill, a bumper, and a speed display above a driver's seat of a large cargo vehicle are described. A lamp or the like may be used as a component, and a similar effect can be obtained. Further, by using more components for vehicle type determination, the accuracy of vehicle type determination can be further improved.

Also, the case has been described where the attributes of the components stored in the component dictionary are the maximum and minimum values of the color and the width / height. A similar effect can be obtained. For example, an image serving as a reference of the turn signal is stored in the component dictionary, and the component detection unit compares and compares the image of the turn signal with the candidate region of each component detected from the input image to determine the component. .

Also, the case where the contents stored in the component arrangement dictionary are the relative vertical relationship and the maximum value / minimum value of the distance between the components on the image has been described. If it is a different item, another item, for example, a left-right positional relationship may be included, and the same effect can be obtained. Further, by using the component arrangement dictionary in which the positional relationships of more components are stored, the accuracy of the vehicle type determination can be further improved.

Also, the case has been described where the contents stored in the vehicle dictionary are the shapes of the components, the size on the image, and the distance between the components on the image. The vehicle type discriminating means may directly refer to the input image and perform the comparison and collation, including not only the information of the component to be detected but also information of another area set based on the component. .

The target of the vehicle type discrimination may be a coarse level such as a truck or an ordinary car, or a fine level including a common name and model of the vehicle. By changing the contents of each dictionary or changing the comparison and collation range of each dictionary according to the level to be determined, the vehicle type can be efficiently determined according to the determination level.

Embodiment 2 Hereinafter, a second embodiment of the present invention will be described with reference to the drawings. As in the first embodiment, FIG. 1 is a block diagram showing a schematic configuration of the present embodiment.

The operation of the present embodiment will be described with reference to FIGS.
This will be described with reference to FIG. FIG. 5 is a flowchart showing the flow of the operation of the present embodiment. FIG. 6 is an input image input by the image input means, and is an image of a front surface of a vehicle entering at an angle α with respect to a horizontal line. FIG.
Is an edge image created by performing edge detection from the input image.

The operation of the vehicle type discriminating apparatus is started, for example, by detecting that the vehicle has arrived within the imaging range of the camera. First, similarly to steps S1 and S2 of the first embodiment, an image is input (step T1), components are detected (step T2), and components are detected from the image of the vehicle.

After that, the component determining means 5 obtains the vehicle center line based on the position of the component which is determined in advance as the reference component among the components to be detected (step T3). In the present embodiment, a headlight is used as a reference component. FIG. 7 shows an edge image created from the input image shown in FIG. 6, and reference numerals 21 and 22 denote headlights detected by the component detecting means 3 in step T2.
The component determining means 5 obtains the vehicle center line in the input image based on the positions of the headlights 21 and 22.
Reference numeral 23 in FIG. 7 is the vehicle center line obtained by the component determining means 5.

Next, the component determining means 5 determines the vehicle center line 2
According to the position and the angle of No. 3, the type of the component of the imaging vehicle is determined from the component candidate area detected by the component detection means 3 (step T4). The operation of this step is almost the same as that of step S3 of the first embodiment. However, in this embodiment, it is assumed that one of the storage items of the component arrangement dictionary 4 has a pair line symmetry flag. The pair line symmetry flag indicates that there are two such components for one vehicle and that they are in line symmetry with respect to the vehicle center line. For example, headlights, turn signals,
Door mirror is applicable. The component determining means 5 also determines that the component having the pair line symmetry flag has a line symmetrical positional relationship with respect to the vehicle center line as the determination condition. The distance and direction between the components in the input image are obtained by performing rotation correction so that the direction of the vehicle center line is vertical.

Thereafter, as in the first embodiment, the vehicle type determining means 7 determines the type of the imaged vehicle by comparing the output of the component determining means 5 with the vehicle dictionary 6 (step T5). To end.

As described above, according to the vehicle type discriminating apparatus according to the present embodiment, in addition to the above-described effects, the center line of the vehicle is obtained, and the type of the component of the vehicle is determined according to the position and angle of the vehicle center line. Is determined, for example, a vehicle that has entered obliquely can be imaged, and the vehicle type can be accurately determined even when the image has distortion.

That is, for example, by adding the positional relationship with respect to the vehicle center line to the condition for determining the components, it is possible to improve the detection accuracy of components symmetric with respect to the vehicle center such as a turn signal, and to improve the detection accuracy of the vehicle center line. By calculating the distance between the components in the input image by performing the rotation correction based on the above, it is possible to further improve the detection accuracy of the components of the vehicle, and to accurately determine the vehicle type even if the image has distortion. it can.

In the second embodiment, there are two components to be determined using the vehicle center line in one vehicle, such as a turn signal, which are positioned symmetrically with respect to the vehicle center line. However, the present invention is not limited to this, and any device may be used as long as it has a line-symmetric positional relationship with respect to the vehicle center line. For example, only one vehicle per vehicle, such as a front grill, may be included which is symmetrical with respect to the vehicle center line.

Further, the present invention may be modified into various forms as in the first embodiment, and the above-described effects can be obtained according to the modifications. For example, a plurality of images may be used as the input image. In this case, the processing time of the vehicle type determination may be increased. However, in addition to the effect of the present embodiment, the effect that the accuracy of the vehicle type determination is further improved. Can be obtained.

Embodiment 3 FIG. Hereinafter, a third embodiment of the present invention will be described with reference to the drawings. FIG. 8 is a block diagram showing a schematic configuration of the present embodiment. In FIG. 8, the same components as those in the above-described embodiment are denoted by the same reference numerals, and description thereof will be omitted. Reference numeral 24 denotes a license plate recognizing unit that recognizes a license plate from the input image stored in the image input unit 1.

Next, the operation will be described with reference to FIGS. FIG. 9 is a flowchart showing the flow of the operation of the present embodiment. FIG. 10 is a schematic diagram showing types of license plates.

The vehicle type discriminating apparatus starts its operation, for example, when it detects that the vehicle has arrived within the imaging range of the camera. First, as in step S1 of the first embodiment, an image is input (step U1).

Next, the license plate recognizing means 24 detects the license plate from the input image stored in the image input means 1 and recognizes the attributes and characters of the license plate (step U2). Regarding license plate detection and recognition, see "Development of license plate recognition device"
(The IEICE Technical Report PRMU 96-46, pp. 65-72). As shown in FIG. 10, the license plate includes a large plate for a large vehicle and a medium plate for a medium vehicle. The number at the upper right of the license plate is a vehicle type code indicating the vehicle type. The license plate recognizing means 24 detects the license plate and outputs a license plate attribute (for example, one of a medium-sized plate and a large-sized plate) and a recognition result of a vehicle type code.

Next, the component detecting means 3 detects the components of the imaged vehicle with reference to the component dictionary 2 based on the recognition result of the license plate recognizing means 24 (step U3). In the present embodiment, it is assumed that the contents of the component dictionary 2 are stored in association with license plate attributes and vehicle type codes. For example, it is assumed that a medium-sized vehicle has a configuration in which the first digit of the vehicle type is 3 or 5 or 7, and a large-sized vehicle has a configuration in which the first digit of the vehicle type is 1 or 2. The component detecting means 3 detects the candidate area of the component of the imaged vehicle with reference to the component dictionary 2 limited to a range matching the attribute of the license plate recognized by the license plate recognizing means 24 and the vehicle type code. Otherwise, Embodiment 1
Is the same as step S2.

Next, based on the recognition result of the license plate recognizing means 24, the component determining means 5 refers to the component arrangement dictionary 4 and uses the component candidate area detected by the component detecting means 3 to pick up the image pickup vehicle. Is determined (step U4). In the present embodiment, it is assumed that the content of the component arrangement dictionary 4 is stored in association with the license plate attribute and the vehicle type code, similarly to the component dictionary 2. The component determining means 5 determines the type of the vehicle component by referring to the component arrangement dictionary 4 limited to the range matching the attribute of the license plate recognized by the license plate recognizing means 24 and the vehicle type code. Other than that,
This is the same as step S3 of the first embodiment.

Thereafter, in the same procedure as in the first embodiment, the vehicle type discriminating means 7 compares the components of the imaged vehicle determined by the component determining means 5 with the vehicle dictionary 6 to determine the type of the imaged vehicle ( Step U5), the operation of vehicle type discrimination ends.

As described above, according to the vehicle type discriminating apparatus according to the present embodiment, in addition to the above-described effects, the component dictionary and the component arrangement dictionary are referenced based on the recognition result of the license plate attribute and the vehicle type code. By limiting the scope,
It is possible to reduce erroneous determination of a vehicle type different from the actual type. Further, the speed of the vehicle type determination process can be increased.

In the third embodiment, the case where the reference range of each dictionary is limited based on the license plate attribute and the recognition result of the vehicle type code has been described. However, the reference range of the dictionary may be limited. However, the present invention is not limited to this. For example, the reference range of the dictionary may be limited only by the attribute of the license plate. In this case, there is no need to recognize the vehicle type code, so a simpler device and
Car type discrimination can be realized at higher speed. Further, for example, the reference range of the dictionary may be limited based on the license plate attribute, the vehicle type code, and the color. In this case, since the reference range can be narrower, erroneous determination can be further reduced.

Also, the case where the reference range of the component dictionary and the component arrangement dictionary is limited has been described, but the present invention is not limited to this. For example, the reference range of only the component dictionary may be limited, or the reference range of only the vehicle dictionary may be limited. Further, the reference range of all dictionaries may be limited.

Further, the present invention may be modified into various forms as in the above-described embodiments, and the above-described effects can be obtained according to the modifications. For example, a plurality of images may be used as the input image. In this case, the processing time of the vehicle type determination may be increased. However, in addition to the effect of the present embodiment, the effect that the accuracy of the vehicle type determination is further improved. Can be obtained.

Embodiment 4 Hereinafter, a fourth embodiment of the present invention will be described with reference to the drawings. FIG. 11 is a block diagram showing a schematic configuration of the present embodiment. In FIG. 11, the same components as those in the above-described embodiment are denoted by the same reference numerals, and description thereof will be omitted. Reference numeral 25 denotes mark recognition means for recognizing marks and characters on the vehicle body from the input image stored in the image input means 1.

Next, the operation will be described with reference to FIGS. FIG. 12 is a flowchart showing the flow of the operation of the present embodiment. FIG. 13 is a schematic diagram showing marks and characters on the vehicle body.

The operation of the vehicle type discriminating apparatus is started, for example, by detecting that the vehicle has arrived within the imaging range of the camera. First, an image is input as in step S1 of the first embodiment (step V1).

Next, the mark recognizing means 25 recognizes marks and characters on the vehicle body from the input image stored in the image input means 1 (step V2). In FIG. 13, reference numeral 26 denotes a mark on the vehicle body, and 27 denotes a character on the vehicle body. The number of marks may be one or more. Regarding mark recognition, for example, “Character pattern extraction and recognition from scene images” (IEICE Transactions D, Vol. J71-D, No. 6, p.
p. 1037 to 1047). The mark recognizing means 25 detects a mark on the vehicle body from the input image and outputs a mark recognition result.

Next, based on the recognition result of the mark recognizing means 25, the component detecting means 3 refers to the component dictionary 2 to detect the components of the imaged vehicle (step V3). In the present embodiment, it is assumed that the contents of the component dictionary 2 are stored in association with the marks. Component detection means 3
Detects the candidate area of the component of the imaged vehicle with reference to the component dictionary 2 limited to the range matching the mark recognized by the mark recognizing means 25. Otherwise, Embodiment 1
Is the same as step S2.

Next, based on the recognition result of the mark recognizing means 25, the component determining means 5 refers to the component arrangement dictionary 4 and uses the component candidate area detected by the component detecting means 3 to detect the image pickup vehicle. The type of the component is determined (step V4). In the present embodiment, the component arrangement dictionary 4 also
Similar to the component dictionary 2, it is assumed that the contents are stored in association with the marks. The component determining means 5 determines the type of the component of the image pickup vehicle with reference to the component arrangement dictionary 4 limited to a range matching the mark recognized by the mark recognizing means 25. Otherwise, it is the same as step S3 of the first embodiment.

Then, in the same procedure as in the first embodiment, the vehicle type discriminating means 7 compares the vehicle components determined by the component determining means 5 with the vehicle dictionary 6 to determine the type of the imaged vehicle (step V5). ), The operation of discriminating the vehicle type ends.

As described above, according to the vehicle type discriminating apparatus according to the present embodiment, in addition to the above-described effects, the reference range of the component dictionary and the component arrangement dictionary is limited based on the recognition result of the mark on the vehicle body. By doing so, it is possible to reduce erroneous determination of a vehicle type different from the actual type. Further, the speed of the vehicle type determination process can be increased.

In the fourth embodiment, the case where the reference range of the component dictionary and the component arrangement dictionary is limited has been described, but the present invention is not limited to this. For example, the reference range of only the component dictionary may be limited, or the reference range of only the vehicle dictionary may be limited. Further, the reference range of all dictionaries may be limited.

Further, the present invention may be modified into various forms as in the above-described embodiments, and the above-described effects can be obtained according to the modifications. For example, a plurality of images may be used as the input image. In this case, the processing time of the vehicle type determination may be increased. However, in addition to the effect of the present embodiment, the effect that the accuracy of the vehicle type determination is further improved. Can be obtained.

Embodiment 5 Hereinafter, a fifth embodiment of the present invention will be described with reference to the drawings. FIG. 14 is a block diagram showing a schematic configuration of the present embodiment. In FIG. 14, the same components as those in the above-described embodiment are denoted by the same reference numerals, and description thereof will be omitted.

Next, the operation will be described with reference to FIGS. FIG. 15 is a flowchart showing the flow of the operation of the present embodiment.

The vehicle type discriminating apparatus starts its operation, for example, when it detects that the vehicle has arrived within the imaging range of the camera. First, an image is input as in step S1 of the first embodiment (step W1).

Next, the license plate recognizing means 24 detects and recognizes the license plate in the same procedure as in the third embodiment (step W2), and outputs the attribute of the license plate and the vehicle type code as a recognition result. The mark recognizing means 25 detects and recognizes the mark on the vehicle body in the same procedure as in the fourth embodiment (step W3), and outputs the mark on the vehicle body.

Next, based on the recognition results of the license plate recognizing means 24 and the mark recognizing means 25, the component detecting means 3 refers to the component dictionary 2 to detect the components of the imaged vehicle (step W4). In the present embodiment, it is assumed that the contents of the component dictionary 2 are stored in association with license plate attributes, vehicle type codes, and marks.
The component detecting means 3 includes a license plate recognizing means 24.
The candidate area of the component of the imaged vehicle is detected by referring to the component dictionary 2 limited to the range matching the attribute, the vehicle type code and the mark recognized by the mark recognizing means 25. Otherwise, it is the same as step S2 of the first embodiment.

Next, the component determining means 5 refers to the component arrangement dictionary 4 on the basis of the recognition results of the license plate recognizing means 24 and the mark recognizing means 25 and detects the component detected by the component detecting means 3. The type of the component of the imaging vehicle is determined from the candidate area (step W5). In the present embodiment, it is assumed that the contents of the component arrangement dictionary 4 are stored in association with the license plate attribute, the vehicle type code and the mark, similarly to the component dictionary 2. The component determining unit 5 refers to the component arrangement dictionary 4 to limit the range of the attribute, the vehicle type code recognized by the license plate recognizing unit 24 and the mark recognized by the mark recognizing unit 25, and refers to the component arrangement dictionary 4. Determine the type. Otherwise, it is the same as step S3 of the first embodiment.

Then, in the same procedure as in the first embodiment, the vehicle type discriminating means 7 compares the components of the imaged vehicle determined by the component determining means 5 with the vehicle dictionary 6 to determine the type of the vehicle in the input image. It is determined (step W6), and the operation of the vehicle type determination ends.

As described above, according to the vehicle type discriminating apparatus according to the present embodiment, in addition to the effects described above, based on the recognition result of the license plate attribute, the recognition result of the vehicle type code, and the recognition result of the mark on the vehicle body. By limiting the reference range of the component dictionary and the component arrangement dictionary in this way, it is possible to further reduce erroneous determination of a vehicle type different from the actual vehicle type. Further, the speed of the vehicle type determination process can be further increased.

In the fifth embodiment, the case where the license plate recognizing means recognizes the license plate attribute and the vehicle type code has been described, but the present invention is not limited to this. For example, only the attributes of the license plate may be recognized. In this case, since it is not necessary to recognize the vehicle type code, the vehicle type determination can be realized with a simpler device and at higher speed. Further, for example, the attribute, the vehicle type code, and the color of the license plate may be recognized. In this case, since the reference range can be narrower, erroneous determination can be further reduced.

The case where the reference range of the component dictionary and the component arrangement dictionary is limited has been described, but the present invention is not limited to this. For example, the reference range of only the component dictionary may be limited, or the reference range of only the vehicle dictionary may be limited. Further, the reference range of all dictionaries may be limited.

Further, the present invention may be modified into various forms as in the above-described embodiments, and the above-described effects can be obtained according to the modifications. For example, a plurality of images may be used as the input image. In this case, the processing time of the vehicle type determination may be increased. However, in addition to the effect of the present embodiment, the effect that the accuracy of the vehicle type determination is further improved. Can be obtained.

[0076]

As described above, according to the vehicle type discriminating apparatus of the present invention, candidate areas of a plurality of components of a vehicle are detected,
By determining the vehicle type using the components whose type has been determined based on the information on the positional relationship, each component of the vehicle can be correctly detected, and the vehicle type can be accurately determined with an inexpensive device. Further, even if there is a component that cannot be detected at all, the vehicle type can be determined with high accuracy from information on other components.

According to the vehicle type discriminating apparatus of the next invention,
In addition to the above-described effects, when determining the type of component of the imaging vehicle, the vehicle center line is obtained, and by using the vehicle center line, for example, an image of a vehicle that has entered obliquely is captured,
Even if the image has distortion, the vehicle type can be determined with high accuracy.

According to the vehicle type discriminating apparatus of the next invention,
In addition to the above-described effects, by limiting the reference range of the component dictionary, the component layout dictionary, and / or the vehicle dictionary based on the recognition result of recognizing the license plate, it is possible to reduce erroneous determination of a vehicle type different from the actual type. Can be. Also,
It is possible to speed up the process of vehicle type determination.

Further, according to the vehicle type discriminating apparatus of the next invention, in addition to the above-mentioned effects, the reference range of the component dictionary, the component arrangement dictionary and / or the vehicle dictionary is determined based on the recognition result obtained by recognizing the attributes of the license plate. , It is possible to reduce erroneous determination of a vehicle type different from the actual vehicle type. Further, the speed of the vehicle type determination process can be increased.

Further, according to the vehicle type discriminating apparatus of the next invention, in addition to the above-described effects, the component dictionary, the component arrangement dictionary, and / or the vehicle dictionary are referred to based on the recognition result obtained by recognizing the vehicle type code of the license plate. By limiting the range, it is possible to reduce erroneous determination of a vehicle type different from the actual type. Further, the speed of the vehicle type determination process can be increased.

According to the vehicle type discriminating apparatus of the next invention,
In addition to the effects described above, by limiting the reference range of the component dictionary, component layout dictionary, and / or vehicle dictionary based on the recognition result of recognition of marks and characters on the vehicle body, erroneous discrimination to a vehicle model different from the actual model is performed. Can be reduced. Further, the speed of the vehicle type determination process can be increased.

According to the vehicle type discriminating apparatus of the next invention,
In addition to the above-described effects, by limiting the reference range of the component dictionary, the component arrangement dictionary, and / or the vehicle dictionary based on the recognition result of recognizing the license plate and the recognition result of recognizing marks and characters on the vehicle body. In addition, it is possible to reduce erroneous determination of a vehicle type different from the actual type. Further, the speed of the vehicle type determination process can be increased.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a schematic configuration of a vehicle type identification device according to Embodiments 1 and 2.

FIG. 2 is a flowchart showing a flow of an operation according to the first embodiment.

FIG. 3 is a schematic diagram illustrating an input image according to the first embodiment;

FIG. 4 is a schematic diagram illustrating an edge image and a candidate region of a component according to the first embodiment;

FIG. 5 is a flowchart showing an operation flow according to the second embodiment.

FIG. 6 is a schematic diagram illustrating an input image according to the second embodiment.

FIG. 7 is a schematic diagram illustrating an edge image, a reference component, and a vehicle center line according to the second embodiment.

FIG. 8 is a block diagram illustrating a schematic configuration of a vehicle type identification device according to a third embodiment.

FIG. 9 is a flowchart showing an operation flow according to the third embodiment.

FIG. 10 is a schematic diagram for explaining types of license plates according to the third embodiment.

FIG. 11 is a block diagram illustrating a schematic configuration of a vehicle type identification device according to a fourth embodiment.

FIG. 12 is a flowchart showing a flow of an operation according to the fourth embodiment.

FIG. 13 is a schematic diagram for explaining a mark according to the fourth embodiment.

FIG. 14 is a block diagram illustrating a schematic configuration of a vehicle type identification device according to a fifth embodiment.

FIG. 15 is a flowchart showing an operation flow in the fifth embodiment.

FIG. 16 is a block diagram illustrating a schematic configuration of a conventional vehicle type identification device.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 Image input means 2 Component dictionary 3 Component detecting means 4 Component arrangement dictionary 5 Component determining means 6 Vehicle dictionary 7 Vehicle type discriminating means 24 Number plate recognizing means 25 Mark recognizing means 101 Number plate recognizing means 102 Vehicle width detecting means 103 Vehicle type determination means 104 Search unit

Claims (7)

[Claims] 1. A configuration of an imaged vehicle from a component dictionary in which the attributes of the components of the vehicle are stored in advance, and an input image in which the vehicle is imaged based on the attributes of the components with reference to the component dictionary. Component detecting means for detecting an element, a component arrangement dictionary in which positional relationships of the components of the vehicle are stored in advance, and the component based on the positional relationship of the components with reference to the component arrangement dictionary A component determining unit for determining the type of the component from the candidate region of the component of the imaging vehicle detected by the detecting unit, and a component for each vehicle type of the vehicle for determining the component for each vehicle type in advance. The vehicle dictionary in which the related information is stored is compared with the component determined by the component determining means and the vehicle dictionary, and based on the related information of the component for each vehicle model, A vehicle type determining device for determining the type of the imaging vehicle. 2. The method according to claim 1, wherein when determining the type of the component by referring to the component arrangement dictionary, the vehicle center line is determined based on a position of the component determined in advance as a reference component. The vehicle type discriminating apparatus according to claim 1, wherein the vehicle center line is determined and the vehicle center line is used. 3. A license plate recognizing means for recognizing a license plate from the input image, and a reference range of the component dictionary, the component layout dictionary and / or the vehicle dictionary according to a recognition result of the license plate recognizing means. 3. The method according to claim 1, wherein
The vehicle type discriminating apparatus according to 1. 4. The vehicle type discriminating apparatus according to claim 3, wherein said license plate recognizing means recognizes an attribute of the license plate. 5. The vehicle type discriminating apparatus according to claim 3, wherein said license plate recognizing means recognizes a vehicle type code of the license plate. 6. A mark recognizing means for recognizing a mark or a character on a vehicle body from the input image, wherein the mark of the component dictionary, the component layout dictionary and / or the vehicle dictionary is stored in accordance with a recognition result of the mark recognizing means. The vehicle type discriminating apparatus according to claim 1 or 2, wherein the reference range is limited. 7. A license plate recognizing means for recognizing a license plate from the input image, and a mark recognizing means for recognizing a mark or a character on a vehicle body from the input image, wherein the license plate recognizing means and the mark recognizing means. 3. The vehicle type discriminating apparatus according to claim 1, wherein a reference range of the component dictionary, the component arrangement dictionary, and / or the vehicle dictionary is limited according to a recognition result of the vehicle type.