JP5691588B2 - Vehicle recognition device, vehicle recognition method, and vehicle recognition program - Google Patents

Description

  The present invention relates to a vehicle recognition device and the like.

  A vehicle recognition system that captures a predetermined monitoring range with a camera installed on the road and recognizes a vehicle passing through the monitoring range using an image recognition technique has been put into practical use. This vehicle recognition system is used, for example, for parking lot management, road information providing systems such as traffic jam information, and automatic toll collection systems for highways.

  As a method for recognizing a vehicle, there is a method using a vehicle license plate. Since the number plate is unique to each vehicle, the vehicle can be recognized if the number plate can be read correctly.

  However, the method of identifying a vehicle using only the license plate has the following problems. First, when there is a shadow on a part of the license plate, or when a part of the license plate is blocked by a shield, characters cannot be read and the vehicle cannot be specified. Also, if the license plate characters are mistakenly recognized as similar characters, the vehicle may not be accurately identified. Moreover, vehicle recognition using a license plate cannot be performed for a vehicle with a counterfeit license plate.

  For this reason, there is a technique for recognizing a vehicle by using not only the information of the license plate but also the characteristics of the vehicle as an auxiliary. In this technique, a license plate and vehicle characteristics are registered in advance in a database in association with vehicle identification information. In this technique, when some characters of the license plate cannot be recognized, vehicle candidates are narrowed down based on the characteristics of the vehicle to be recognized and the characteristics registered in the database.

JP 62-090800 A JP 2000-222673 A JP 2008-033462 A

  However, in the above-described conventional technology, it is assumed that a database that associates the information of the license plate with the characteristics of the vehicle body is prepared in advance. However, when considering traffic monitoring on a general roadway, it is difficult to prepare a database in advance by predicting what kind of vehicle will pass.

  The disclosed technology has been made in view of the above, and an object of the present invention is to provide a vehicle recognition device, a vehicle recognition method, and a vehicle recognition program capable of reliably building a database used when a vehicle is recognized. .

  The vehicle recognition device disclosed in the present application includes, in one aspect, a recognition unit, a reliability adjustment unit, and a database management unit. A recognition part acquires the image of the vehicle of the imaging | photography range, and recognizes the number plate of the said vehicle, and the characteristic of the said vehicle. The reliability adjustment unit includes a first combination indicating a combination of the license plate recognized by the recognition unit and the vehicle characteristic, and a second combination indicating a combination of the license plate registered in the database and the vehicle characteristic. Compare. When the first combination and the second combination match, the reliability adjustment unit adds a predetermined value to the reliability with respect to the vehicle characteristics of the matched second combination. When the license plate of the first combination and the license plate of the second combination match, and the characteristics of the vehicle of the first combination do not match the characteristics of the vehicle of the second combination, A predetermined value is subtracted from the reliability for the feature of the second combination. The database management unit updates the feature of the second combination vehicle to the feature of the first combination vehicle based on the reliability of the feature of the second combination vehicle registered in the database.

  According to the vehicle recognition device, a database used for recognizing a vehicle can be constructed in a reliable state.

FIG. 1 is a diagram illustrating a configuration of a vehicle recognition device according to the present embodiment. FIG. 2 is a diagram illustrating an example of the data structure of the database. FIG. 3 is a diagram (1) for explaining the process by which the license plate recognition unit calculates the reliability. FIG. 4 is a diagram (2) for explaining the process by which the license plate recognition unit calculates the reliability. FIG. 5 is a diagram for explaining processing in which the vehicle feature recognition unit calculates the reliability. FIG. 6 is a diagram illustrating an example of processing of the reliability adjustment unit. FIG. 7 is a diagram illustrating an example of processing of the registration unit. FIG. 8 is a diagram illustrating an example of processing of the unauthorized vehicle determination unit. FIG. 9 is a diagram (1) illustrating an example of processing of the vehicle specifying unit. FIG. 10 is a diagram (2) illustrating an example of processing of the vehicle specifying unit. FIG. 11 is a diagram (3) illustrating an example of a process of the vehicle specifying unit. FIG. 12 is a diagram for explaining the processing of the vehicle specifying unit using all the features. FIG. 13 is a flowchart illustrating a processing procedure of the vehicle recognition device. FIG. 14 is a flowchart illustrating a processing procedure of the vehicle identification processing. FIG. 15 is a flowchart illustrating a processing procedure of the unauthorized vehicle determination process. FIG. 16 is a flowchart illustrating a processing procedure of database update processing. FIG. 17 is a diagram illustrating an example of a computer that executes a vehicle recognition program.

  Hereinafter, embodiments of a vehicle recognition device, a vehicle recognition method, and a vehicle recognition program disclosed in the present application will be described in detail with reference to the drawings. Note that the present invention is not limited to the embodiments.

  A configuration of the vehicle recognition apparatus according to the present embodiment will be described. FIG. 1 is a diagram illustrating a configuration of a vehicle recognition device according to the present embodiment. As illustrated in FIG. 1, the vehicle recognition device 100 includes an image acquisition unit 110, a communication unit 120, a database 130, a recognition unit 140, a management unit 145, an unauthorized vehicle determination unit 170, and a vehicle identification unit 180. In addition, the vehicle recognition device 100 is connected to the camera 10. The vehicle recognition device 100 is connected to the network 50.

  The camera 10 is installed on the road, for example, and takes an image of the shooting range. The camera 10 outputs the image data to the vehicle recognition device 100.

  The image acquisition unit 110 is a processing unit that acquires image data from the camera 10 and outputs the image data to the recognition unit 140. For example, the image acquisition unit 110 corresponds to an interface that connects the camera 10 and the vehicle recognition device 100. The communication unit 120 is a processing unit that performs data communication with a management device or the like via the network 50. The communication unit 120 corresponds to, for example, a communication card.

  The database 130 is a storage device that stores information on the license plate and the characteristics of the vehicle having the license plate in association with each other. The database 130 corresponds to, for example, a semiconductor memory element such as a random access memory (RAM), a read only memory (ROM), and a flash memory, or a storage device such as a hard disk or an optical disk.

  FIG. 2 is a diagram illustrating an example of the data structure of the database. As shown in FIG. 2, this database 130 stores vehicle ID (Identification), license plate information, and vehicle feature information in association with each other. The vehicle ID is identification information that uniquely identifies the vehicle.

  The license plate information includes a land branch, a classification, an application, and a serial number. Of these, the land branch corresponds to the land transportation branch office. The classification corresponds to the number written next to the land support on the license plate. The usage corresponds to one hiragana character written in the lower left of the license plate. The serial number corresponds to the numeric string described in the lower part of the license plate.

  The vehicle feature information includes features of a plurality of types of vehicles. The reliability is set for each feature. The greater the confidence, the more likely the feature is. The characteristics of the vehicle may be any characteristics as long as they represent the characteristics of the vehicle. The characteristics of the vehicle correspond to the color of the vehicle, the type of the vehicle, the size of the vehicle, the mounting position of the license plate, and the like.

  As shown in FIG. 2, the license plate information corresponding to the vehicle ID “1” is the land branch “Kawasaki”, the classification “300”, the usage “ha”, and the serial number “4249”. In the vehicle feature information corresponding to the vehicle ID “1”, the feature corresponding to the item “feature A” is “A1”, and the reliability corresponding to the item “feature A” is “100”. In the vehicle feature information corresponding to the vehicle ID “1”, the feature corresponding to the item “feature B” is “B1”, and the reliability corresponding to the item “feature B” is “90”. In the vehicle feature information corresponding to the vehicle ID “1”, the feature corresponding to the item “feature C” is “C1”, and the reliability corresponding to the item “feature C” is “40”. In the vehicle feature information corresponding to the vehicle ID “1”, the feature corresponding to the item “feature D” is “D1”, and the reliability corresponding to the item “feature D” is “30”. In the vehicle feature information corresponding to the vehicle ID “1”, the feature corresponding to the item “feature E” is “E1”, and the reliability corresponding to the item “feature E” is “60”. Further, in the vehicle feature information corresponding to the vehicle ID “1”, the feature corresponding to the item “feature F” is “F1”, and the reliability corresponding to the item “feature F” is “90”. In the vehicle feature information corresponding to the vehicle ID “1”, the feature corresponding to the item “feature G” is “G1”, and the reliability corresponding to the item “feature G” is “100”.

  License plate information and vehicle characteristic information are also stored in the vehicle IDs “2” and “3”. The feature and reliability of the vehicle feature information corresponding to the vehicle ID “4” is “−”. This indicates that the vehicle feature information corresponding to the vehicle ID “4” is not registered.

  Returning to the description of FIG. The recognition unit 140 includes a license plate recognition unit 140a and a vehicle feature recognition unit 140b. The license plate recognition unit 140a acquires image data from the image acquisition unit 110, analyzes the image data, and recognizes the license plate. In addition, the license plate recognition unit 140a calculates the reliability of the license plate recognition result.

  An example of processing in which the license plate recognition unit 140a recognizes the license plate will be described. The license plate recognition unit 140a holds images such as character strings, hiragana, and numeric strings of the land transportation branch office as templates. The license plate recognition unit 140a cuts out a license plate image from the image data, performs pattern matching between the cut-out image and the template, and recognizes the land support, classification, application, and serial number of the license plate.

  An example of a process in which the license plate recognition unit 140a calculates the reliability for the recognition result will be described. The license plate recognition unit 140a calculates the reliability for the recognition result of the land support, classification, application, and serial number. FIG. 3 and FIG. 4 are diagrams for explaining processing in which the license plate recognition unit calculates the reliability. Here, the case where the license plate recognition unit 140a calculates the reliability for the recognition result of the land branch will be described.

  In FIG. 3, 1a is a template of the land branch “Nasu”, and 1b corresponds to an area of image data recognized as “Nasu”. The license plate recognition unit 140a generates a difference image 1c between the template 1a and the image data area 1b. For example, the license plate recognizing unit 140a generates a difference image 1c configured by a difference in luminance between each pixel of the template 1a and each pixel of the region 1b. The average luminance of the difference image 1c takes a smaller value as the template 1a and the region 1b match. Therefore, the license plate recognition unit 140a calculates a value obtained by subtracting the average luminance of the difference image 1c from the maximum reliability value (for example, 100). In the example illustrated in FIG. 3, the difference image 1 c includes many images with low luminance. Therefore, the average luminance of the difference image 1 c takes a small value, and the reliability of the land branch recognition result “Nasu” is high.

  In FIG. 4, 1a is a template of the land branch “Nasu”, and 1d corresponds to an area of image data recognized as “Nasu”. The license plate recognition unit 140a generates a difference image 1e between the template 1a and the image data area 1d. For example, the license plate recognizing unit 140a generates a difference image 1e composed of the luminance difference between each pixel of the template 1a and each pixel of the region 1d. The average luminance of the difference image 1e takes a smaller value as the template 1a and the region 1d match. Therefore, the license plate recognition unit 140a calculates a value obtained by subtracting the average luminance of the difference image 1e from the maximum reliability value (for example, 100). In the example illustrated in FIG. 4, the difference image 1 e includes many images with high luminance. Therefore, the average luminance of the difference image 1 e takes a large value, and the reliability of the land branch recognition result “Nasu” is low.

  The license plate recognition unit 140a also calculates the reliability for the recognition result for the classification, application, and serial number in the same manner as the Nasu recognition result.

  The license plate recognition unit 140a includes license plate information in which the recognition result of the license plate and the reliability corresponding to the recognition result are associated with each other, the reliability adjustment unit 150, the registration unit 160, the unauthorized vehicle determination unit 170, the vehicle identification unit. Output to 180. Note that the license plate recognition unit 140a may set “?” As the recognition result whose reliability is equal to or less than a predetermined threshold. For example, the license plate recognition unit 140a sets the predetermined threshold value to 50.

  Returning to the description of FIG. The vehicle feature recognition unit 140b acquires image data from the image acquisition unit 110, analyzes the image data, and recognizes the feature of the vehicle. Further, the vehicle feature recognition unit 140b calculates the reliability of the recognition result of the vehicle features.

  An example of the process in which the vehicle feature recognition unit 140b recognizes the feature of the vehicle will be described. The characteristics of the vehicle correspond to the color of the vehicle, the type of the vehicle, the size of the vehicle, the mounting position of the license plate, and the like. The technology that the vehicle feature recognition unit 140b recognizes the feature of the vehicle uses a conventional image recognition technology.

  For example, the vehicle feature recognition unit 140b identifies the position of the vehicle from the image data, acquires RGB (Red Green Blue) values from each pixel of the vehicle, and averages the RGB values to recognize the color of the vehicle. . The vehicle feature recognition unit 140b holds the shape of the vehicle as a template for each vehicle type, and recognizes the type of the vehicle by performing pattern matching between the shape of the vehicle in the image data and the template. Further, the vehicle feature recognition unit 140b holds a table in which the proportion of vehicles in the image and the size of the vehicle are associated with each other, and the size of the vehicle is recognized using this table. The vehicle feature recognition unit 140b identifies the position of the vehicle and the position of the license plate from the image data, and recognizes the position from the upper end of the vehicle to the upper end of the license plate as the attachment position of the license plate.

  An example of processing in which the vehicle feature recognition unit 140b calculates the reliability of the recognition result of the vehicle features will be described. FIG. 5 is a diagram for explaining processing in which the vehicle feature recognition unit calculates the reliability. Here, it is assumed that the feature of the vehicle can be distinguished into feature 1 and feature 2. For example, the feature of the vehicle is the color of the vehicle body, the feature 1 is the color “red” of the vehicle, and the feature 2 is the color “blue” of the vehicle. The horizontal axis of FIG. 5 shows the size of the red component when focusing on the red component and the blue component of the colors of the vehicle. The vertical axis in FIG. 5 indicates the magnitude of the blue component when focusing on the red component and the blue component of the vehicle colors.

  The vehicle feature recognition unit 140b sets a boundary in advance based on a sample in which the correct answer of the vehicle feature is known. For example, a sample of red and blue cars is collected in advance, and as shown in FIG. 5, the size of the red component and the size of the blue component of the vehicle color of the red car are plotted with black circles. The size of the red component and the blue component of the vehicle color are plotted with white circles. From the plotted result, the boundary line 2a is set so that the red car and the blue car can be well separated. The boundary 2a may be set manually by a person, or may be set using an existing multivariate analysis such as discriminant analysis.

  When the vehicle feature recognition unit 140b calculates the reliability of the recognition result of the color of the vehicle, the point plotted by the size of the red component and the size of the blue component of the color of the vehicle to be recognized, and the boundary line 2a Is the reliability. For example, the vehicle feature recognition unit 140b determines that the color of the vehicle is “red”, and the point plotted by the size of the red component and the size of the blue component of the color of the vehicle is 2b in FIG. In this case, the vehicle feature recognizing unit 140b calculates the distance between the point 2b and the boundary line 2a, and sets the calculated value as the reliability for the color “red” of the vehicle.

  The vehicle feature recognition unit 140b outputs vehicle feature information in which the recognition result of the vehicle feature is associated with the reliability corresponding to the recognition result to the reliability adjustment unit 150 and the registration unit 160. In addition, the vehicle feature recognition unit 140b may normalize the magnitude of the reliability so that the magnitude of the reliability is included in 0 to 100.

  The management unit 145 is a processing unit that registers or updates the vehicle feature information in the database 130. The management unit 145 includes a reliability adjustment unit 150 and a registration unit 160. The reliability adjustment unit 150 is a processing unit that adjusts the reliability for each feature of the vehicle feature information in the database 130.

  The process of the reliability adjustment unit 150 will be specifically described. The reliability adjustment unit 150 acquires license plate information and vehicle feature information extracted from a vehicle having the same image data from the license plate recognition unit 140a and the vehicle feature recognition unit 140b. The reliability adjustment unit 150 determines whether or not the reliability of the license plate information for the land support, classification, application, and serial number are all equal to or greater than a predetermined threshold. For example, the predetermined threshold is 90. If any of the reliability levels is less than 90, the reliability level adjustment unit 150 waits until the next license plate information is acquired, and repeatedly executes the above processing.

  The reliability adjustment unit 150 corresponds to the land support, classification, use, and serial number of the license plate information when the reliability for the land support, classification, use, and serial number of the license plate information is all equal to or greater than a predetermined threshold. The vehicle feature information in the database 130 is specified. Here, the specified vehicle feature information is referred to as vehicle feature information 1A. The vehicle feature information acquired by the reliability adjustment unit 150 from the vehicle feature recognition unit 140b is referred to as vehicle feature information 1B.

  After identifying the vehicle feature information 1A, the reliability adjustment unit 150 compares each feature of the vehicle feature information 1A with each feature of the vehicle feature information 1B. As a result of the comparison, the reliability adjustment unit 150 adds a predetermined value to the reliability of the feature that matches the feature of the vehicle feature information 1B among the features of the vehicle feature information 1A. For example, the reliability adjustment unit 150 adds 1 to the reliability. In contrast, the reliability adjustment unit 150 subtracts a predetermined value from the reliability of features that do not match the features of the vehicle feature information 1B among the features of the vehicle feature information 1A. For example, the reliability adjustment unit 150 subtracts 1 from the reliability. In addition, the value added or subtracted from the reliability by the reliability adjustment unit 150 may be a value proportional to the reliability of the vehicle feature information 1B instead of a constant value.

  The reliability adjustment unit 150 adds or subtracts reliability for each feature of the vehicle feature information 1A, and then determines whether or not there is a feature whose reliability is less than the reference value in the vehicle feature information 1A. judge. For example, the reference value is 10. When all the reliability of the feature of the vehicle feature information 1A is 10 or more, the reliability adjustment unit 150 waits until the next license plate information is acquired, and repeatedly executes the above processing.

  When there is a feature whose reliability is less than 10 in the vehicle feature information 1A, the reliability adjustment unit 150 updates the content of the corresponding feature to the feature content of the vehicle feature information 1B. When the feature content is updated, the reliability adjustment unit 150 sets the reliability corresponding to the updated feature to an initial value. For example, the reliability adjustment unit 150 sets the initial value to 10. The reliability adjustment unit 150 waits until the next license plate information is acquired after updating the vehicle characteristic information 1A, and repeatedly executes the above processing.

  The process of the reliability adjustment unit 150 will be described with reference to the drawings. FIG. 6 is a diagram illustrating an example of processing of the reliability adjustment unit. The reliability adjustment unit 150 acquires the license plate information 1C from the license plate recognition unit 140a. Each reliability of the license plate information 1C is 90 or more. When the reliability adjustment unit 150 compares the license plate information 1C with the database 130, the vehicle ID “4” data including the license plate information corresponding to the license plate information 1C is hit. The vehicle feature information 1A is included in the data of the vehicle ID “4”.

  As shown in FIG. 6, the reliability adjustment unit 150 compares the vehicle feature information 1A in the database 130 with the vehicle feature information 1B acquired from the vehicle feature recognition unit 140b. As a result of the comparison, the features A to D are inconsistent between the vehicle feature information 1A and the vehicle feature information 1B. On the other hand, the features E to G match in the vehicle feature information 1A and the vehicle feature information 1B.

  Therefore, the reliability adjustment unit 150 adds a value proportional to the reliability of the vehicle feature information 1B to the reliability of the features E to G as shown in 3a of FIG. As shown in 3b, the reliability adjustment unit 150 subtracts a value proportional to the reliability of the vehicle feature information 1B from the reliability of the features B to D. Since the reliability corresponding to the feature A is less than the minimum reliability, the reliability adjustment unit 150 converts the content of the feature A into the feature “the content of the vehicle feature information 1B as shown in 3c”. Update to “A4” and set the reliability to the initial value “10”.

  The registration unit 160 is a processing unit that registers vehicle feature information corresponding to license plate information in the database 130 when vehicle feature information corresponding to license plate information is not registered in the database 130. The registration unit 160 acquires the license plate information and the vehicle feature information extracted from the same image data from the license plate recognition unit and the vehicle feature recognition unit 140b. The registration unit 160 registers the vehicle feature information in the database 130 when the reliability of the license plate identification information with respect to the land branch, classification, application, and serial number is all equal to or greater than a predetermined threshold.

  FIG. 7 is a diagram illustrating an example of processing of the registration unit. The registration unit 160 acquires the license plate information 1C from the license plate recognition unit 140a. Each reliability of the license plate information 1C is 90 or more. When the registration unit 160 compares the license plate information 1C with the database 130, the data of the vehicle ID “4” including the license plate information corresponding to the license plate information 1C is hit. The vehicle feature information corresponding to the vehicle ID “4” is blank. Therefore, the registration unit 160 registers the vehicle feature information 1B as vehicle feature information corresponding to the vehicle ID “4”. At this time, for example, an initial value “10” is set as the reliability of the vehicle feature information corresponding to the vehicle ID “4”.

  Returning to the description of FIG. The unauthorized vehicle determination unit 170 is a processing unit that determines whether or not the vehicle included in the image data is an unauthorized vehicle. The unauthorized vehicle determination unit 170 notifies the management device of the ID of the unauthorized vehicle. The management device is connected to the network 50 and installed in a management center that manages vehicles. The unauthorized vehicle determination unit 170 is an example of a vehicle feature determination unit.

  Below, the process of the unauthorized vehicle determination part 170 is demonstrated concretely. The unauthorized vehicle determination unit 170 acquires the license plate information and the vehicle feature information extracted from the vehicle having the same image data from the license plate recognition unit 140a and the vehicle feature recognition unit 140b. The unauthorized vehicle determination unit 170 determines whether or not the reliability of the license plate information for land support, classification, usage, and serial number is all equal to or greater than a predetermined threshold. For example, the predetermined threshold is 90. If any of the reliability levels is less than 90, the unauthorized vehicle determination unit 170 waits until the next license plate information is acquired, and repeatedly executes the above processing.

  The illegal vehicle determination unit 170 responds to the land plate, classification, use, and serial number of the license plate information when the reliability of the land plate, classification, application, and serial number of the license plate information is all equal to or greater than a predetermined threshold. The vehicle feature information in the database 130 is specified. Here, the specified vehicle feature information is referred to as vehicle feature information 1A. The unauthorized vehicle determination unit 170 sets the vehicle feature information acquired from the vehicle feature recognition unit 140b as vehicle feature information 1B.

  After identifying the vehicle feature information 1A, the unauthorized vehicle determination unit 170 compares each feature of the vehicle feature information 1A with each feature of the vehicle feature information 1B. However, the unauthorized vehicle determination unit 170 compares features whose reliability is equal to or higher than a predetermined threshold. For example, the predetermined threshold is 90. The unauthorized vehicle determination unit 170 counts the number of different features among the compared features.

  If the number of different features is equal to or greater than a predetermined number, the unauthorized vehicle determination unit 170 determines that the vehicle corresponding to the vehicle feature information 1A or the vehicle corresponding to the vehicle feature information 1B is an unauthorized vehicle. And the unauthorized vehicle determination part 170 notifies vehicle ID corresponding to 1 A of vehicle characteristic information to a management apparatus.

  The processing of the unauthorized vehicle determination unit 170 will be described with reference to the drawings. FIG. 8 is a diagram illustrating an example of processing of the unauthorized vehicle determination unit. The unauthorized vehicle determination unit 170 acquires the license plate information 1C from the license plate recognition unit 140a. Each reliability of the license plate information 1C is 90 or more. When the unauthorized vehicle determination unit 170 compares the license plate information 1C with the database 130, the data of the vehicle ID “3” including the license plate information corresponding to the license plate information 1C is hit. The vehicle feature information 1A is included in the data of the vehicle ID “3”.

  As shown in FIG. 8, features having reliability of 90 or more in the vehicle feature information 1 </ b> A are feature C to feature G. Among the vehicle feature information 1B, features having a reliability of 90 or more are features C, D, feature F, and feature G. Therefore, the unauthorized vehicle determination unit 170 compares the feature C, feature D, feature F, and feature G of the vehicle feature information 1A with the feature C, feature D, feature F, and feature G of the vehicle feature information 1B. As a result of the comparison, different features are feature C, feature D, and feature F. When the unauthorized vehicle determination unit 170 counts the number of different features, the number of different features is three. For example, the unauthorized vehicle determination unit 170 determines that the vehicle is an unauthorized vehicle when the number of different features is two or more. In this case, the unauthorized vehicle determination unit 170 notifies the management device of the vehicle ID “3”.

  Returning to the description of FIG. The vehicle identification unit 180 is a processing unit that identifies the vehicle ID based on the license plate information and the characteristics of the vehicle when the reliability of the license plate information is low. The vehicle identification unit 180 is an example of a vehicle identification determination unit. Here, the low reliability of the license plate information means that one of the reliability of the license plate information, the classification, the application, and the serial number is less than the predetermined reliability. . For example, the predetermined reliability is 90. In addition, when the land support, classification, use, and serial number reliability of the license plate information are all equal to or higher than a predetermined reliability, the vehicle specifying unit 180 does not specify the vehicle. This is because the vehicle can be uniquely identified from the license plate information.

  The process of the vehicle specific | specification part 180 when the reliability of license plate information is low is demonstrated. The vehicle specifying unit 180 acquires the license plate information and the vehicle feature information extracted from the vehicle having the same image data from the license plate recognition unit 140a and the vehicle feature recognition unit 140b. The reliability adjustment unit 150 determines whether or not the reliability of the license plate information for the land support, classification, application, and serial number are all equal to or greater than a predetermined threshold. For example, the predetermined threshold is 90. If all of the reliability levels are 90 or more, the reliability adjustment unit 150 waits until the next license plate information is acquired, and repeatedly executes the above processing.

  When any of the reliability levels is less than 90, the vehicle specifying unit 180 specifies the vehicle feature information in the database 130 corresponding to the license plate information using a reliability level of 90 or more as a key. For example, it is assumed that the reliability of the land support, classification, and serial number among the land support, classification, use, and serial number of the license plate information is 90 or more. In this case, the vehicle specifying unit 180 specifies the vehicle feature information in the database 130 using the land support, classification, and serial number of the license plate information as keys. Here, the specified vehicle feature information is referred to as vehicle feature information 1D. The vehicle feature information 1D may include a plurality of vehicle feature information with different vehicle IDs. The vehicle feature information acquired by the vehicle specifying unit 180 from the vehicle feature recognition unit 140b is referred to as vehicle feature information 1B.

  The vehicle identification unit 180 compares each feature of the vehicle feature information 1D with each feature of the vehicle feature information 1B. However, the vehicle identification unit 180 compares features whose reliability is equal to or higher than a predetermined threshold. For example, the predetermined threshold is 90. The vehicle identifying unit 180 counts the number of matching features among the compared features. The vehicle identification unit 180 includes the vehicle ID of the vehicle feature information when the number of matching features is the largest among the vehicle feature information included in the vehicle feature information 1D and the number of matching features is a predetermined number or more. Determine. For example, the predetermined number is 2. The vehicle specifying unit 180 notifies the management device of information on the specified vehicle ID.

  In addition, the vehicle specific | specification part 180 outputs an error, when multiple applicable vehicle characteristic information exists. The vehicle identifying unit 180 also outputs an error when there is no vehicle feature information in which the number of matching features is equal to or greater than a predetermined number.

  The processing of the vehicle specifying unit 180 will be described with reference to the drawings. 9-11 is a figure which shows an example of a process of a vehicle specific part. The vehicle identification unit 180 acquires the license plate information 1C from the license plate recognition unit 140a. As shown in FIG. 9, the reliability of the land support, classification, and serial number of the license plate information 1C is 90 or more. The reliability of the license plate information 1C is less than 90. In this case, the vehicle specifying unit 180 specifies the vehicle feature information ID of the database 130 having a set of the land branch “Kawasaki”, the classification “300”, and the serial number “4249” of the license plate information 1C. As shown in the lower part of FIG. 9, license plate information having a set of land support “Kawasaki”, classification “300”, and serial number “4249” corresponds to license plate information of vehicle ID “1, 2”.

  The vehicle identifying unit 180 refers to the reliability of the vehicle feature information 1B and identifies a feature having a reliability of 90 or more. As shown in the first row of FIG. 10, features having reliability of 90 or more in the vehicle feature information 1B are feature A, feature C, feature D, feature F, and feature G.

  The vehicle identifying unit 180 refers to the reliability of the vehicle feature information 1D and identifies a feature having a reliability of 90 or more. As shown in the second row of FIG. 10, features having reliability of 90 or more in the vehicle feature information 1D corresponding to the vehicle ID “1” are feature A, feature B, feature F, and feature G. As shown in the second row of FIG. 10, features with reliability of 90 or more in the vehicle feature information 1D corresponding to the vehicle ID “2” are feature A, feature B, feature C, and feature F.

  The vehicle specifying unit 180 compares the features of the vehicle feature information 1B and 1D having a reliability of 90 or more. For this reason, as shown in the third row of FIG. 10, the vehicle feature information 1D of the vehicle ID “1” is compared with the vehicle feature information 1B regarding the features A, F, and G. Further, the vehicle feature information 1D of the vehicle ID “2” is compared with the vehicle feature information 1B regarding the features A, C, and F.

  The description shifts to the description of FIG. The vehicle identification unit 180 compares the features A, F, and G of the vehicle feature information 1B with the features A, F, and G of the vehicle feature information 1D with the vehicle ID “1”. As a result of comparison, feature A, feature F, and feature G match each other. Therefore, the vehicle identification unit 180 sets the number of matches of the vehicle feature information 1D of the vehicle ID “1” to “3”.

  Further, the vehicle feature unit 180 compares the features A, C, and F of the vehicle feature information 1B with the features A, C, and F of the vehicle feature information 1D with the vehicle ID “2”. As a result of comparison, the feature C matches. Therefore, the vehicle identification unit 180 sets the number of matches of the vehicle feature information 1D of the vehicle ID “2” to “1”.

  As shown in FIG. 11, the number of matches of the vehicle feature information 1D of the vehicle ID “1” is “3”, and the number of matches of the vehicle feature information 1D of the vehicle ID “2” is “1”. The vehicle feature information 1D having the largest number of matches is the vehicle feature information of the vehicle ID “1”, and the number of matches is a predetermined number 2 or more. For this reason, the vehicle feature unit 180 notifies the management apparatus of the vehicle ID “1”.

  As described above, the vehicle specifying unit 180 specifies the number of matches by focusing on features whose reliability is equal to or higher than a predetermined reliability. In this way, it is useful to narrow down the features to be compared based on the reliability. If all the features are compared regardless of the reliability, the vehicle ID may not be narrowed down.

  FIG. 12 is a diagram for explaining the processing of the vehicle specifying unit using all the features. As illustrated in FIG. 12, the vehicle identification unit 180 compares the features A to G of the vehicle feature information 1B with the features A to G of the vehicle feature information 1D of the vehicle ID “1”. F and feature G match. On the other hand, when the vehicle identification unit 180 compares the features A to G of the vehicle feature information 1B with the features A to G of the vehicle feature information 1D of the vehicle ID “2”, the features B, C, and D are Match. Since the number of matches corresponding to the vehicle ID “1, 2” is both “3”, the vehicle identification unit 180 cannot narrow down the vehicle ID.

  The recognition unit 140, the management unit 145, the unauthorized vehicle determination unit 170, and the vehicle identification unit 180 correspond to an integrated device such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA). The processing units 140 to 180 correspond to electronic circuits such as a CPU (Central Processing Unit) and an MPU (Micro Processing Unit), for example.

  Next, the process procedure of the vehicle recognition apparatus 100 according to the present embodiment will be described. FIG. 13 is a flowchart illustrating a processing procedure of the vehicle recognition device. The process of FIG. 13 is executed when the vehicle recognition apparatus 100 acquires image data from the camera 10, for example. As shown in FIG. 13, the vehicle recognition apparatus 100 acquires image data (step S101), and recognizes a license plate included in the image data (step S102). Moreover, the vehicle recognition apparatus 100 recognizes the characteristics of the vehicle included in the image data (step S103).

  The vehicle recognition apparatus 100 determines whether the reliability of the license plate information is high (step S104). In step S104, the vehicle recognition apparatus 100 determines that the reliability of the license plate information is high when the reliability of the land support, classification, application, and serial number of the license plate information is all equal to or higher than a predetermined reliability.

  When it is determined that the reliability of the license plate information is not high (No at Step S104), the vehicle recognition device 100 executes a vehicle specifying process (Step S105) and ends the process. On the other hand, when it is determined that the reliability of the license plate information is high (step S104, Yes), the vehicle recognition device 100 performs an unauthorized vehicle determination process (step S106). For example, the vehicle recognition apparatus 100 may determine that the reliability is high when the reliability of the license plate information exceeds a predetermined threshold.

  If the vehicle recognition apparatus 100 determines that the vehicle is an unauthorized vehicle (step S107, Yes), the process ends. On the other hand, when the vehicle recognition device 100 does not determine that the vehicle is an unauthorized vehicle (No at Step S107), the vehicle recognition device 100 performs a database update process (Step S108) and ends the process.

  Next, the vehicle identification process shown in step S105 of FIG. 13 will be described. FIG. 14 is a flowchart illustrating a processing procedure of the vehicle identification processing. As illustrated in FIG. 14, the vehicle identification unit 180 extracts vehicle feature information 1D from the database 130 based on the license plate information (step S201).

  The vehicle identification unit 180 selects a feature with high reliability of the vehicle feature information 1B (step S202). The vehicle identification unit 180 selects a feature with high reliability in the vehicle feature information 1D (step S203). The vehicle identifying unit 180 compares the selected features and counts the number of matching features (step S204).

  The vehicle identification unit 180 determines whether or not the vehicle ID corresponding to the largest number of matches is single (step S205). If the vehicle ID is single (step S205, Yes), the vehicle identification unit 180 outputs the vehicle ID (step S206) and ends the process. On the other hand, when there are a plurality of vehicle IDs (step S205, No), the vehicle specifying unit 180 outputs an error (step S207) and ends the process.

  Next, the unauthorized vehicle determination process shown in step S106 of FIG. 13 will be described. FIG. 15 is a flowchart illustrating a processing procedure of the unauthorized vehicle determination process. As illustrated in FIG. 15, the unauthorized vehicle determination unit 170 extracts the vehicle feature information 1A corresponding to the license plate information from the database 130 (step S301).

  The unauthorized vehicle determination unit 170 selects a feature with high reliability in the vehicle feature information 1B (step S302). The unauthorized vehicle determination unit 170 selects a highly reliable feature from the vehicle feature information 1A extracted from the database 130 (step S303).

  The unauthorized vehicle determination unit 170 compares the selected features and determines whether or not they match (step S304). If the characteristics match (Yes in Step S305), the unauthorized vehicle determination unit 170 determines that the vehicle is not an unauthorized vehicle (Step S306), and ends the process. On the other hand, when the characteristics do not match (No at Step S305), the unauthorized vehicle determination unit 170 determines that the vehicle is an unauthorized vehicle, notifies the management device of the determination result (Step S307), and ends the process.

  Next, the database update process shown in step S108 of FIG. 13 will be described. FIG. 16 is a flowchart illustrating a processing procedure of database update processing. As shown in FIG. 16, the management unit 145 determines whether or not the vehicle feature information corresponding to the license plate information has been registered (step S401).

  When the vehicle feature information corresponding to the license plate information is not registered (No at Step S401), the management unit 145 registers the vehicle feature information (Step S402) and ends the process.

  When the vehicle feature information corresponding to the license plate information is registered (Yes at Step S401), the management unit 145 determines whether the vehicle feature information is the same as the database 130 (Step S403). When the vehicle feature information is the same as that in the database 130 (step S403, Yes), the management unit 145 increases the reliability of the feature of the matched vehicle feature information (step S404) and ends the process.

  On the other hand, when the vehicle feature information is different from the database 130 (No at Step S403), the management unit 145 lowers the reliability of the feature of the vehicle feature information that does not match (Step S405). The management unit 145 determines whether there is a feature whose reliability of the vehicle feature information is equal to or less than a reference value (step S406). For example, the reference value is 10.

  When there is no feature whose reliability of the vehicle feature information is less than or equal to the reference value (No in step S406), the management unit 145 ends the process. On the other hand, when there is a feature whose reliability of the vehicle feature information is equal to or less than the reference value (step S406, Yes), the management unit 145 updates the feature that is less than the reference value (step S407), and ends the process.

  Next, effects of the vehicle recognition device 100 according to the present embodiment will be described. The vehicle recognition device 100 acquires license plate information and vehicle feature information 1B from the license plate recognition unit 140a and the vehicle feature recognition unit 140b, and adjusts the reliability of the features registered in the database 130. Specifically, the vehicle recognition apparatus 100 compares the vehicle feature information 1A and the vehicle feature information 1B in the database 130, increases the reliability of matching features, and decreases the reliability of non-matching features. The vehicle recognition apparatus 100 updates the content of the feature with low reliability in the vehicle feature information 1A to the content of the feature of the vehicle feature information 1B. The vehicle recognition apparatus 100 performs the above process every time image data is acquired. As a result, the same feature extracted from a vehicle having the same license plate has high reliability, and the feature that is rarely extracted has low reliability. Therefore, features that are more reliable as vehicle features remain in the database 130, and features that are not certain as vehicle features do not remain in the database 130. For this reason, according to the vehicle recognition apparatus 100, the database used when recognizing a vehicle can be constructed in a reliable state.

  In addition, the vehicle recognition apparatus 100 compares the license plate information and vehicle feature information set recognized from the image data with the database 130 and determines whether or not the features of the vehicle feature information match. For this reason, according to the vehicle recognition apparatus 100, an unauthorized vehicle different from the feature of the database 130 can be determined.

  In addition, when the reliability of the license plate is less than the threshold, the vehicle recognition device 100 determines the vehicle identification information based on the characteristics of the vehicle extracted from the image data and the database 130. For this reason, even if the license plate cannot be accurately recognized due to the influence of a shadow or a shield, the vehicle can be specified.

  In addition, the vehicle recognition apparatus 100 adds or subtracts the reliability for each feature of the vehicle feature information included in the database 130, and determines the vehicle identification information using a feature whose reliability is equal to or higher than a predetermined threshold. For this reason, a vehicle can be determined more correctly.

  Moreover, the vehicle recognition apparatus 100 acquires license plate information and vehicle feature information, and registers the acquired vehicle feature information when there is no vehicle feature information corresponding to the license plate information in the database 130. For this reason, the vehicle characteristic information in the database 130 can be supplemented sequentially.

  The configuration of the vehicle recognition device 100 is not limited to that shown in FIG. For example, the database 130, the management unit 145, the unauthorized vehicle determination unit 170, and the vehicle identification unit 180 may be provided on the server of the management center. The server may notify the server of vehicle numbers and features recognized by the vehicle recognition device 100 and the camera 10 distributed in each area, and the server may detect or identify the vehicle.

  Incidentally, the processing of the vehicle recognition apparatus 100 and the like described in the above-described embodiments can be realized by executing a program prepared in advance by a computer system such as a personal computer or a workstation.

  Here, an example of a computer that executes a vehicle recognition program that realizes the same function as the processing by the vehicle recognition device 100 described in the above embodiment will be described with reference to FIG. FIG. 17 is a diagram illustrating an example of a computer that executes a vehicle recognition program.

  As illustrated in FIG. 17, a computer 200 that functions as the vehicle recognition device 100 includes a CPU 201 that executes various arithmetic processes, an input device 202 that receives input of data from a user, and a monitor 203. The vehicle recognition device 100 also includes a medium reading device 204 that reads a program and the like from a storage medium, and a network interface device 205 that exchanges data with other computers via a network. The computer 200 also includes a camera 206, a RAM 207 that temporarily stores various information, and a hard disk device 208. The devices 201 to 208 are connected to the bus 209.

  The hard disk device 208 stores a recognition program 208a, a management program 208b, and a determination program 208c that exhibit functions similar to the functions of the vehicle recognition device 100 described above. The hard disk device 208 stores vehicle data 208d. The vehicle data 208d corresponds to data stored in the database 130.

  The CPU 201 reads out the programs 208 a to 208 c from the hard disk device 208 and expands them in the RAM 207. The recognition program 208a functions as a recognition process 207a. The management program 208b functions as a management process 207b. The determination program 208c functions as a determination process 207c.

  The recognition process 207a corresponds to the recognition unit 140 in FIG. The management process 207b corresponds to the management unit 145 in FIG. The determination process 207c corresponds to the unauthorized vehicle determination unit 170 and the vehicle identification unit 180 of FIG. The CPU 201 executes the processes 207a and 207b, registers and updates the information of the vehicle data 208b, and executes the determination process 207c to determine the unauthorized vehicle and identify the vehicle.

  The programs 208a to 208c are not necessarily stored in the hard disk device 208 from the beginning. For example, each program is stored in a “portable physical medium” such as a flexible disk (FD), a CD-ROM, a DVD disk, a magneto-optical disk, and an IC card inserted into the computer 200. Then, the computer 200 may read and execute each program from these.

  The following supplementary notes are further disclosed with respect to the embodiments including the above examples.

(Additional remark 1) The recognition part which acquires the image of the vehicle of the imaging | photography range, and recognizes the number plate of the said vehicle, and the characteristic of the said vehicle,
Comparing the first combination indicating the combination of the license plate and the vehicle characteristics recognized by the recognition unit with the second combination indicating the combination of the license plate and the vehicle characteristics registered in the database;
When the first combination and the second combination match, a predetermined value is added to the reliability for the vehicle characteristics of the matched second combination,
If the license plate of the first combination and the license plate of the second combination match and the characteristics of the vehicle of the first combination do not match the characteristics of the vehicle of the second combination, A reliability adjustment unit that subtracts a predetermined value from the reliability of the feature;
A database management unit that updates the characteristics of the vehicle of the second combination to the characteristics of the vehicle of the first combination based on the reliability of the characteristics of the vehicle of the second combination registered in the database. The vehicle recognition apparatus characterized by the above-mentioned.

(Additional remark 2) The said recognition part further calculates the reliability of the recognized license plate, extracts the 2nd combination which has a license plate with a license plate with a reliability more than a threshold value from the said database, and extracted The vehicle is further characterized by further comprising a vehicle feature determination unit that determines whether or not the characteristics of the two combinations of vehicles match the vehicle features recognized by the recognition unit (140) and outputs a determination result. The vehicle recognition device according to 1.

(Supplementary note 3) The vehicle recognition according to supplementary note 1, wherein the recognition unit further calculates the reliability of the recognized license plate, and registers or updates the database only when the reliability of the license plate information is high. apparatus.

(Supplementary Note 4) The database stores vehicle identification information in association with the second combination,
Supplementary note 1 further comprising a vehicle identification determination unit that determines vehicle identification information based on the first combination and the second combination when the reliability of the license plate is less than a threshold value. The vehicle recognition device according to 2 or 3.

(Supplementary Note 5) The vehicle features included in the first combination have a plurality of vehicle features, and the vehicle features included in the second combination have a plurality of vehicle features, and the recognition The unit further calculates a reliability for each vehicle feature included in the first combination, and the reliability adjustment unit adds or subtracts the reliability for each vehicle feature included in the second combination. The vehicle identification determination unit is based on the characteristics of the vehicle whose reliability included in the first combination is equal to or higher than the threshold and the characteristics of the vehicle whose reliability included in the second combination is equal to or higher than the threshold. The vehicle recognition device according to appendix 4, wherein vehicle identification information is determined.

(Additional remark 6) When the 2nd combination which has a license plate corresponding to the license plate of the said 1st combination does not exist in the said database, the said database management part, the license plate contained in the said 1st combination, The vehicle recognition device according to any one of appendices 1 to 5, wherein the vehicle feature is registered in the database in association with the feature of the vehicle.

(Appendix 7) A vehicle recognition method executed by a computer,
Acquire an image of the vehicle in the shooting range, recognize the license plate of the vehicle and the characteristics of the vehicle,
Comparing the first combination indicating the combination of the license plate and the characteristics of the vehicle with the second combination indicating the combination of the license plate and the characteristics of the vehicle registered in the database;
When the first combination and the second combination match, a predetermined value is added to the reliability for the vehicle characteristics of the matched second combination,
If the license plate of the first combination and the license plate of the second combination match and the characteristics of the vehicle of the first combination do not match the characteristics of the vehicle of the second combination, Subtract a given value from the confidence for the feature,
Based on the reliability with respect to the characteristics of the second combination of vehicles registered in the database, the characteristics of the second combination of vehicles are updated to the characteristics of the first combination of vehicles. Recognition method.

(Additional remark 8) The said computer further calculates the reliability of the recognized license plate, extracts the 2nd combination which has a license plate whose reliability is more than a threshold value from the said database, and extracted vehicle of the 2nd combination 8. The vehicle recognition method according to appendix 7, wherein it is determined whether or not the characteristics of the vehicle and the recognized characteristics of the vehicle coincide with each other, and a determination result is output.

(Supplementary note 9) The vehicle recognition method according to supplementary note 7, wherein the reliability of the recognized license plate is further calculated, and the database is registered or updated only when the reliability of the license plate information is high.

(Supplementary Note 10) The computer further stores the database in association with vehicle identification information in the second combination,
The vehicle identification information is determined based on the first combination and the second combination when the reliability of the license plate is less than a threshold value. Vehicle recognition method.

(Supplementary Note 11) The vehicle features included in the first combination have a plurality of vehicle features, and the vehicle features included in the second combination have a plurality of vehicle features, and the computer Further calculates the reliability for each vehicle feature included in the first combination, adds or subtracts the reliability for each vehicle feature included in the second combination, and is included in the first combination. APPENDIX 10 characterized in that vehicle identification information is determined based on characteristics of a vehicle whose reliability is equal to or higher than a threshold and characteristics of a vehicle whose reliability included in the second combination is equal to or higher than a threshold. The vehicle recognition method described in 1.

(Additional remark 12) The said computer WHEREIN: When the 2nd combination which has a license plate corresponding to the license plate of the said 1st combination does not exist in the said database, the license plate contained in the said 1st combination and vehicle The vehicle recognition method according to any one of appendices 7 to 11, wherein a feature is associated and registered in the database.

(Supplementary note 13)
Acquire an image of the vehicle in the shooting range, recognize the license plate of the vehicle and the characteristics of the vehicle,
Comparing the first combination indicating the combination of the license plate and the characteristics of the vehicle with the second combination indicating the combination of the license plate and the characteristics of the vehicle registered in the database;
When the first combination and the second combination match, a predetermined value is added to the reliability for the vehicle characteristics of the matched second combination,
If the license plate of the first combination and the license plate of the second combination match and the characteristics of the vehicle of the first combination do not match the characteristics of the vehicle of the second combination, Subtract a given value from the confidence for the feature,
Vehicle recognition for executing a process of updating the characteristics of the vehicle of the second combination to the characteristics of the vehicle of the first combination based on the reliability of the characteristics of the vehicle of the second combination registered in the database. program.

(Additional remark 14) The reliability of the recognized license plate is further calculated, the 2nd combination which has a license plate whose reliability is more than a threshold value is extracted from the said database, and the characteristics of the vehicle of the extracted 2nd combination and recognition 14. The vehicle recognition program according to appendix 13, wherein the computer is further caused to determine whether or not the characteristics of the vehicle matched with each other and to output a determination result.

(Supplementary note 15) The vehicle recognition program according to supplementary note 13, wherein the reliability of the recognized license plate is further calculated, and the database is registered or updated only when the reliability of the license plate information is high.

(Supplementary Note 16) The database stores vehicle identification information in association with the second combination,
Appendix 13: causing the computer to further execute processing for determining vehicle identification information based on the first combination and the second combination when the reliability of the license plate is less than a threshold value. 14. A vehicle recognition program according to claim 14 or 15.

(Supplementary Note 17) The vehicle feature included in the first combination has a plurality of vehicle features, and the vehicle feature included in the second combination has a plurality of vehicle features, and the computer Further, the reliability for each vehicle feature included in the first combination is further calculated, and the reliability is added or subtracted for each vehicle feature included in the second combination, and included in the first combination. The vehicle identification information is determined based on the characteristics of the vehicle whose reliability is equal to or higher than the threshold and the characteristics of the vehicle whose reliability included in the second combination is equal to or higher than the threshold. The vehicle recognition program according to appendix 16.

(Supplementary Note 18) When a second combination having a license plate that matches the license plate of the first combination does not exist in the database, the license plate included in the first combination corresponds to the characteristics of the vehicle. Then, the vehicle recognition program according to any one of appendices 13 to 17, further causing the computer to execute registration in the database.

DESCRIPTION OF SYMBOLS 100 Vehicle recognition apparatus 110 Image acquisition part 120 Communication part 130 Database 140 Recognition part 145 Management part 170 Unauthorized vehicle determination part 180 Vehicle identification part

Claims (8)

A recognition unit that acquires an image of a vehicle in a shooting range and recognizes the license plate of the vehicle and the characteristics of the vehicle;
Comparing the first combination indicating the combination of the license plate and the vehicle characteristics recognized by the recognition unit with the second combination indicating the combination of the license plate and the vehicle characteristics registered in the database;
When the first combination and the second combination match, a predetermined value is added to the reliability for the vehicle characteristics of the matched second combination,
If the license plate of the first combination and the license plate of the second combination match and the characteristics of the vehicle of the first combination do not match the characteristics of the vehicle of the second combination, A reliability adjustment unit that subtracts a predetermined value from the reliability of the feature;
Based on the reliability adjusted by the reliability adjustment unit for the characteristics of the vehicle of the second combination registered in the database, the reliability of the vehicle of the second combination is less than the reference value . And a database management unit that updates the feature to the feature of the vehicle of the first combination. The recognition unit further calculates the reliability of the recognized license plate,
A second combination having a license plate that matches a license plate whose reliability is equal to or higher than a threshold value is extracted from the database, and the characteristics of the extracted second vehicle match the characteristics of the vehicle recognized by the recognition unit. The vehicle recognition apparatus according to claim 1, further comprising a vehicle feature determination unit that determines whether to perform the determination and outputs a determination result. The vehicle recognition device according to claim 1, wherein the recognition unit further calculates the reliability of the recognized license plate, and registers or updates the database when the reliability of the license plate information is equal to or greater than a threshold value. . The database stores vehicle identification information in association with the second combination,
A vehicle identification determination unit for determining vehicle identification information of the vehicle recognized by the recognition unit based on the first combination and the second combination when the reliability of the license plate is less than a threshold; The vehicle recognition device according to claim 2 , wherein the vehicle recognition device is provided. The vehicle features included in the first combination have a plurality of vehicle features, the vehicle features included in the second combination have a plurality of vehicle features, and the recognition unit The reliability for each vehicle feature included in the first combination is further calculated, and the reliability adjustment unit adds or subtracts the reliability for each vehicle feature included in the second combination, and the vehicle identification The determination unit is configured to recognize the recognition based on a feature of the vehicle in which the reliability included in the first combination is greater than or equal to a threshold and a feature of the vehicle in which the reliability included in the second combination is greater than or equal to the threshold. The vehicle recognition apparatus according to claim 4, wherein vehicle identification information of the vehicle recognized by the unit is determined.   The database management unit, when a second combination having a license plate that matches the license plate of the first combination does not exist in the database, the license plate included in the first combination and the characteristics of the vehicle The vehicle recognition device according to claim 1, wherein the vehicle recognition device is registered in the database in association with each other. A vehicle recognition method executed by a computer,
Acquire an image of the vehicle in the shooting range, recognize the license plate of the vehicle and the characteristics of the vehicle,
Comparing the first combination indicating the combination of the license plate and the characteristics of the vehicle with the second combination indicating the combination of the license plate and the characteristics of the vehicle registered in the database;
When the first combination and the second combination match, a predetermined value is added to the reliability for the vehicle characteristics of the matched second combination,
If the license plate of the first combination and the license plate of the second combination match and the characteristics of the vehicle of the first combination do not match the characteristics of the vehicle of the second combination, Subtract a given value from the confidence for the feature,
When the reliability is equal to or less than a reference value based on the reliability of the addition or subtraction adjustment for the characteristics of the vehicle of the second combination registered in the database, the vehicle of the second combination The vehicle recognition method is characterized in that the characteristic of the vehicle is updated to the characteristic of the vehicle of the first combination. On the computer,
Acquire an image of the vehicle in the shooting range, recognize the license plate of the vehicle and the characteristics of the vehicle,
Comparing the first combination indicating the combination of the license plate and the characteristics of the vehicle with the second combination indicating the combination of the license plate and the characteristics of the vehicle registered in the database;
When the first combination and the second combination match, a predetermined value is added to the reliability for the vehicle characteristics of the matched second combination,
If the license plate of the first combination and the license plate of the second combination match and the characteristics of the vehicle of the first combination do not match the characteristics of the vehicle of the second combination, Subtract a given value from the confidence for the feature,
When the reliability is equal to or less than a reference value based on the reliability of the addition or subtraction adjustment for the characteristics of the vehicle of the second combination registered in the database, the vehicle of the second combination The vehicle recognition program which performs the process which updates the characteristic of this to the characteristic of the vehicle of said 1st combination.

Images