JP2013171452A - Car number recognition method and car number recognition device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To achieve a car number recognition method and a car number recognition device with a simple constitution capable of reducing car number recognition errors and shortening a recognition time as a whole.SOLUTION: A car number recognition device includes: a camera 51 installed outdoors, photographing an image including the number plate of an entering vehicle, and capable of setting the brightness of a photographed image in a plurality of stages; an image processing part 61 for processing the image to recognize a car number of the number plate; an illuminance sensor 52 for detecting an illuminance; and a photographic condition control part 81 for setting the priority order of a photographing parameter. The photographic condition control part sets a priority order of the photographing parameter on the basis of the brightness of a background and a character in the image of the number plate when recognition of the car number is successful. When the elapsed time after the last successful recognition of a car number is equal to or more than a prescribed value, the photographic condition control part performs day and night determination by a calendar function, sets such that a photographing parameter corresponding to the night has the highest priority when it is at night, and otherwise sets the priority of a photographing parameter according to an illuminance detected by the illuminance sensor.

Description

  The present invention relates to a vehicle number recognition method and a vehicle number recognition device, and in particular, an image including a license plate of an approaching vehicle is photographed with a camera installed outdoors, and the vehicle number (number) of the license plate is recognized by image processing. The present invention relates to a vehicle number recognition method and a vehicle number recognition device.

  A vehicle number recognition device for recognizing (automatically reading) the vehicle number (all characters) on a license plate of a vehicle is used for collecting a fee when entering and leaving a vehicle in a parking lot and managing a parking space. By using the car number recognition device in conjunction with parking lot equipment, (1) Realization of measures to enable the entry time to be specified by searching for car numbers when a user's parking ticket is lost, (2) Monthly contract Realization of traffic congestion mitigation measures by automatically opening and closing the gate by automatically recognizing the vehicle number of vehicles and pre-paid vehicles, (3) grasping the user area by calculating the utilization rate by land code of the vehicle number, etc. There is an advantage that it becomes possible.

The vehicle number recognition device captures a license plate with a camera, acquires digitized image data, performs image processing, and recognizes characters. The car number recognition device may be installed both indoors and outdoors. When the car number recognition device is installed in an indoor parking lot or an underground parking lot, sunlight does not enter the area where the camera shoots and illumination is installed, so that the illuminance of the shooting environment hardly changes. On the other hand, when the car number recognition device is installed outdoors, the illuminance of the shooting environment varies from a state close to zero to about 110,000 lux (lx: lm / m ² ) depending on the season, time, weather, etc. There can be. In this case, generally, the minimum illuminance is observed at night without moonlight, the maximum illuminance is observed during clear summer days, and the amount of change reaches several tens of thousands of times.

  An industrial camera used in a parking lot or the like does not have a sufficiently wide dynamic range for a normal industrial camera other than a dedicated expensive camera. Therefore, it is difficult to capture an image with appropriate brightness in the above-described shooting environment such as outdoors. If the brightness adjustment is not appropriate, the difference (contrast) between the characters on the license plate and the background (contrast) is not sufficient, resulting in a whitish or blackish image and the problem that the characters cannot be read occurs. In general, a whitish image is called “whiteout”, and a blackish image is called “blackout”.

  There are four types of shooting parameters for appropriately adjusting the luminance of an image to be shot: (1) exposure, (2) gain, (3) aperture, and (4) illumination intensity. (1) The exposure adjustment adjusts the length of time during which light is converted into electric charge by the image sensor. (2) Gain adjustment adjusts the magnification for converting the charge generated by the image sensor into a voltage. (3) The diaphragm increases or decreases the amount of light incident on the camera body. (4) The light intensity adjustment of the illumination is performed by adjusting the illuminance in the vicinity of photographing with the camera.

  Among the above shooting parameters, the “aperture” is not prepared as a function of the camera body in the case of an industrial camera, and it is necessary to add a part. This component is expensive if it can be adjusted electronically, and a fixed type that is manually adjusted is used, but it is not subject to automatic adjustment. In addition, “illumination” is an object of the shooting parameter when adjusting the light quantity of the illuminator, but the adjustment algorithm becomes complicated as the number of parameters of the shooting parameter increases. Suppose that the intensity of illumination is fixed.

  As described above, in the present invention, “exposure” and “gain”, which are easy to adjust, are used as shooting parameters.

  As described above, when a normal industrial camera is used for the car number recognition device in an outdoor parking lot, a good image cannot be obtained with fixed shooting parameters (exposure and gain). Therefore, in order to obtain a good image, shooting is performed with a plurality of different shooting parameters. A good image refers to an image that has a high contrast between a character on the license plate and a portion without a character (background) and is easy to perform image recognition processing. The multiple sets of shooting parameters are set so that a good image can be obtained with the illuminance of the shooting environment that each group is in charge of, and the entire range of illuminance of the shooting environment at the installation location is set with the multiple sets of shooting parameters. Cover. Due to cost constraints, there is one camera and one image processing unit, and it is not possible to simultaneously capture images with a plurality of cameras with different imaging parameters and perform image processing in parallel with a plurality of image processing devices. In other words, an object of the present invention is a configuration in which an image captured by one camera is processed by one image processing unit.

  Furthermore, in the case of a car number recognition device installed at an entrance / exit of a parking lot or the like, the entering vehicle is moving at a certain speed, and if the car number cannot be recognized by processing the first shot image, Whether the image processing can be performed by changing the imaging parameter to an image that can be easily recognized and performing imaging again depends on the moving speed of the vehicle. When the first shot image is processed and the vehicle number cannot be recognized, and when the subsequent shooting and image processing cannot be performed, a vehicle number recognition error occurs. The vehicle number recognition error also occurs when the license plate is dirty or when a shadow is projected on the license plate. A parking lot manager who is a user is required to reduce the occurrence rate of vehicle number recognition errors and shorten the vehicle number recognition time as a whole.

  Therefore, priorities are assigned according to the degree of expectation that a good recognition result is obtained for the photographing parameters, and photographing and image processing are performed in the order of the photographing parameters of the priority order. The present invention relates to control for assigning this priority.

For vehicle number recognition devices used outdoors, various methods for reducing vehicle number recognition errors, reducing power consumption, and the like have been proposed.
Patent Document 1 describes a vehicle number recognition device installed at a predetermined outdoor location, which uses a timer function to reduce the amount of illumination light during the daytime to reduce the power consumption of the illumination device.

  Patent Documents 2, 3 and 7 are vehicle number recognition devices for recognizing the vehicle number of a vehicle traveling on the road surface, taking into account the installation point of the camera, and using a timer function, It is described that the brightness of the image is adjusted by aperture adjustment based on the determination result and the road surface image. Furthermore, Cited Document 2 describes that the aperture is adjusted with reference to image data taken in the immediately preceding predetermined period.

Patent Document 4 describes that an on-vehicle vehicle number recognition device detects a light lighting state of a vehicle and adjusts the gain of the camera according to the detection result.
Patent Document 5 is a vehicle number recognition device that recognizes the vehicle number of a vehicle traveling on a road surface, and determines whether sunlight becomes forward light or backlight with respect to photographing using a timer function. And adjusting the brightness of the image by adjusting the aperture based on the image of the road surface.
Patent Document 6 describes that the exposure sensitivity of a camera is adjusted according to the measured ambient illuminance in a car number recognition device.

  As described above, adjusting the aperture with reference to the image data taken in the immediately preceding predetermined period, adjusting the aperture according to the detected ambient illuminance, and adjusting the aperture according to the information on the installation location and the time It has been proposed to adjust.

JP 2003-300137 A JP 2005-354746 A Japanese Patent Laid-Open No. 8-124082 JP 2009-071521 A JP-A-4-364600 JP-A-3-276379 JP 2009-239501 A

However, in the methods proposed so far, the vehicle number recognition error cannot be sufficiently reduced, and the vehicle number recognition time as a whole cannot be sufficiently shortened.
An object of the present invention is to realize a vehicle number recognition method and a vehicle number recognition device that reduce vehicle number recognition errors with a simple configuration and to shorten the vehicle number recognition time as a whole.

  In order to solve the above problems, the vehicle number recognition method and the vehicle number recognition device of the present invention are based on the sunrise and sunset times calculated from the information on the brightness of the license plate detected immediately before, the time, the latitude and longitude of the installation location. By using the determination result and the ambient illuminance detected by the illuminance sensor, the priority of the imaging parameters is set with a simple algorithm, thereby reducing the vehicle number recognition error.

  That is, the vehicle number recognition method according to the present invention takes an image including a license plate of an entering vehicle with a camera installed outdoors, processes the image data captured by the camera, and recognizes the vehicle number of the license plate. Vehicle number recognition method, and based on the background and the brightness of the characters in the license plate image when the vehicle number recognition is successful, the priority of the shooting parameters that define the shooting conditions according to the illuminance of the shooting environment If the elapsed time from the last successful vehicle number recognition is greater than or equal to the specified value, the calendar function is used to make day / night judgments. It is set so that priority is given, and in the case of other than night, the shooting parameter is set according to the illuminance detected by the illuminance sensor.

  Also, the vehicle number recognition device of the present invention is a camera that is installed outdoors and captures an image including a license plate of an approaching vehicle, and can set brightness of the captured image in a plurality of stages, and image data captured by the camera Based on the image processing unit that recognizes the vehicle number of the license plate, the illuminance sensor that detects the illuminance near the camera, the image data processed by the image processing unit, and the illuminance detected by the illuminance sensor An imaging condition control unit that sets the priority of imaging parameters that define imaging conditions according to the illuminance of the imaging environment, and the imaging condition control unit is configured to detect the vehicle number by the image processing unit. Based on the background and the brightness of the characters in the license plate image when the recognition is successful, the priority of the shooting parameters is set, and finally the image processing unit recognizes the car number. If the elapsed time has exceeded the specified value, the day / night judgment is performed using the calendar function.In the case of night, the shooting parameters corresponding to the night are set to have the highest priority. Is characterized in that the priority order of the shooting parameters is set according to the illuminance detected by the illuminance sensor.

  According to the present invention, (1) information on the brightness of the license plate detected immediately before, (2) day and night determination result based on sunrise and sunset times calculated from time, latitude and longitude of the installation location, and (3) illuminance sensor With the simple imaging parameter priority setting algorithm that organically uses the ambient illuminance detected in step 1, the vehicle number recognition error is reduced and the vehicle number recognition time is shortened as a whole.

  The camera can set the brightness of the image with a plurality of shooting parameters, and in the case of night, the shooting parameter corresponding to the case where the shooting environment is dark is given the highest priority.

  The calendar function calculates the sunrise and sunset times from the calculated time calculated by the built-in clock function and the latitude and longitude of the set location of the car number recognition device, and calculates the calculated times as sunrise and sunset times. To determine whether it is night.

  According to the present invention, the vehicle number can be reliably recognized by simple arithmetic processing simply by adding an illuminance sensor that detects the illuminance of the surroundings using a general function of an industrial camera. Number recognition time can be shortened.

FIG. 1 is a diagram showing a configuration of a parking lot system according to an embodiment of the present invention. FIG. 2 is a diagram illustrating a configuration of a vehicle number recognition device realized in the camera device of the embodiment. FIG. 3 shows a case where shooting parameters are switched in accordance with an algorithm for determining the next shooting parameter in accordance with the character of the license plate taken immediately before and the brightness of the background, and the license plate is fixed at a fixed position at intervals of 5 minutes. FIG. 6 is a diagram illustrating changes in luminance at a plurality of locations that are processed as images in FIG. 4 to recognize the background and characters of the license plate and are determined to be the background or character region. FIG. 4 is a flowchart illustrating priority order setting processing in the shooting parameter priority order setting unit of the embodiment.

  FIG. 1 is a diagram showing a configuration of a parking lot system according to an embodiment of the present invention, in which (A) shows the overall configuration of the parking lot system, (B) shows an entrance gate, an exit gate, a pre-settlement machine, and controls them It is a figure which shows the structure of the control system of the parking lot system which consists of a management apparatus.

  As shown in FIG. 1A, the parking system according to the embodiment includes a parking lot 1 surrounded by a fence and having a large number of parking spaces inside, and an entrance gate 2 provided at one place of the fence. And an exit gate 3 provided at one location of the fence, and a pre-payment machine 4.

  As shown in FIG. 1B, the entrance gate 2 includes photoelectric sensors 23 and 24, a parking ticket issuing machine 25, a camera device 26, and a gate device 27 that opens and closes the gate bar 28. The exit gate 3 includes photoelectric sensors 33 and 34, a fully automatic settlement machine 35, a camera device 36, and a gate device 37 for opening and closing the gate bar 38. The parking ticket issuing machine 25, the camera device 26 and the gate device 27 at the entrance gate 2, the fully automatic settlement machine 35, the camera device 36 and the gate device 37 at the exit gate 3, and the advance settlement machine 4 are connected via a communication path 41. The management apparatus 40 is communicably connected. The communication path 41 is realized by an RS485 interface, for example.

  As for the photoelectric sensors 23 and 24 of the entrance gate 2, one photoelectric sensor 24 outputs a light beam such as infrared rays toward the other photoelectric sensor 23. In a state where there is no shield between the photoelectric sensors 23 and 24, the photoelectric sensor 23 receives a light beam and outputs a high-level light reception signal. On the other hand, when the vehicle 5 or the like enters between the photoelectric sensors 23 and 24 and blocks the light beam, the reception signal at the photoelectric sensor 23 changes to a low level, so the vehicle 5 or the like enters the entrance gate 2. Can be detected. When the photoelectric sensor 23 detects that the amount of change in the received signal is equal to or greater than the threshold, the photoelectric sensor 23 notifies the camera device 26 of the occurrence of entry into the gate. When a person or the like other than the vehicle enters between the photoelectric sensors 23 and 24, the camera apparatus is similarly notified of the occurrence of entry into the gate. However, as will be described later, even if the camera device 26 operates in accordance with the photoelectric sensor 23, there is no particular problem because the license plate cannot be recognized. The operation of the photoelectric sensors 33 and 34 at the exit gate 3 is the same.

  When the camera devices 26 and 36 are notified of the entry to the gate from the photoelectric sensors 23, 24, 33, and 34, the camera devices 26 and 36 take an image including a vehicle license plate, and the attached image processing unit recognizes the vehicle number. Process. When the vehicle number is recognized, the recognized vehicle number is notified to the management device 40. As described above, when a vehicle other than the vehicle enters the gate or the license plate is dirty, the vehicle number is not recognized, and the vehicle number cannot be recognized.

  The parking ticket issuing machine 25 is operated by a person concerned with the vehicle that has entered the entrance gate 2, issues a parking ticket in accordance with the operation, and notifies the management device 40 of data related to the issuing. The management device 40 stores the issued parking ticket and the vehicle number recognized by the camera device 26 in association with each other, and manages the information. When the operation is performed, the parking ticket issuing machine 25 may communicate with the management device 40 and issue a parking ticket in accordance with an instruction from the management device 40 when the condition is satisfied. Also in this case, the management device 40 stores the issued parking ticket and the vehicle number recognized by the camera device 26 in association with each other, and manages the information.

  The fully automatic settlement machine 35 is operated after a parking ticket is inserted by a vehicle person who has entered the exit gate 3, and information on the inserted parking ticket is notified to the management device 40. At this time, since the camera device 36 notifies the vehicle number of the recognized vehicle, the camera device 36 confirms whether it matches the vehicle number stored in association with the parking ticket issued when the entrance gate 2 is entered. For example, the fee corresponding to the elapsed time is notified to the fully automatic settlement machine 35. The fully automatic settlement machine 35 displays the notified fee and notifies the management device 40 of the completion of the settlement when the settlement is completed.

  The gate devices 27 and 37 perform an opening / closing operation in response to a command from the management device 40.

  The advance settlement machine 4 is operated after a parking ticket is inserted, and information on the inserted parking ticket is notified to the management device 40. The management device 40 notifies the advance settlement machine 4 of a fee corresponding to the elapsed time since the parking ticket was issued. The advance settlement machine 4 displays the notified fee, and when the settlement is completed, notifies the management device 40 of the settlement end.

Next, the operation and processing at the time of entering / exiting the parking lot will be described.
The vehicle 5 using the parking lot enters from the entrance gate 2. At this time, in the parking lot system of the embodiment, the vehicle number recognition device attached to the camera device 26 provided at the entrance gate 2 recognizes the vehicle number from the image of the license plate of the vehicle 5 entering. The recognized vehicle number is notified to the management device 40, and it is determined whether the recognized vehicle number is included in the vehicle number of the vehicle of the monthly contract.

  If included, that is, in the case of a monthly contract vehicle, the management device 40 instructs the gate device 27 to raise the gate bar, so that the monthly contract vehicle can enter the parking lot 1 and is predetermined. Park in an open or empty parking space. At this time, the management device 40 stores the entry time in association with the vehicle number of the vehicle of the entered monthly contract, and manages the information.

  When the vehicle is not included in the monthly contract vehicle, that is, when the vehicle 5 other than the monthly contract receives a parking ticket issued by the parking ticket issuing machine 25 of the entrance gate 2, the gate device 27 causes the management device 40 to raise the gate bar 28. Therefore, the user enters the parking lot 1 and parks in a vacant parking space. At this time, the management device 40 stores the issued parking ticket and the vehicle number recognized by the camera device 26 in association with each other, and manages the information.

  When the monthly contracted vehicle 5 exits the parking lot 1 and enters the exit gate 3, the vehicle number recognition device attached to the camera device 36 recognizes the vehicle number from the image of the number plate of the entering vehicle. It is determined whether or not the recognized vehicle number is included in the vehicle number of the vehicle of the monthly contract, and if included, that is, in the case of the vehicle of the monthly contract, the vehicle of the monthly contract of that vehicle number is based on the managed information. After confirming that the vehicle has entered the parking lot 1, the gate device 37 is instructed to raise the gate bar 38. In this way, the vehicle with the monthly contract goes out of the parking lot 1.

  When a vehicle 5 other than the monthly contract leaves the parking lot 1, there are two ways to settle the fee. One is a normal fee settlement method in which payment is made by the fully automatic settlement machine 35 of the exit gate 3. In this case, when the vehicle 5 leaving the parking lot 1 enters the exit gate 3, the vehicle number recognition device attached to the camera device 36 recognizes the vehicle number. Since the recognized vehicle number is not included in the vehicle number of the vehicle with the monthly contract, the gate bar 38 remains down. When the payment is completed by inserting the parking ticket into the fully automatic checkout machine, the management device 40 instructs the gate device 37 to raise the gate bar 38 after confirming the charge adjustment. In this way, the vehicle 5 goes out of the parking lot 1.

  Another method is a method using the pre-payment machine 4. In this case, the person concerned with the vehicle leaving the parking lot 1 inserts the parking ticket into the pre-payment machine 4 and performs settlement. The management device 40 stores the car number associated with the parking ticket for which the settlement has been completed as a settled car number after confirming the fee settlement in the prior settlement machine 4.

  When the vehicle 5 that has been paid in advance enters the exit gate 3, the vehicle number recognition device attached to the camera device 36 recognizes the vehicle number and notifies the management device 40 of the recognized vehicle number. The management device 40 determines whether the notified vehicle number is included in the adjusted vehicle number, and if included, that is, in the case of the adjusted vehicle, instructs the gate device 37 to raise the gate bar 38. In this way, the settled vehicle goes out of the parking lot 1. If it is not included, the above normal charge settlement method is performed.

  The above is the description of the parking lot system according to the embodiment. Such a system itself is widely used, and various modifications are also known. The vehicle number recognition device and the vehicle number recognition method of the present invention are realized in the camera devices 26 and 36. The vehicle number recognition device and the vehicle number recognition method of the present invention can also be applied to modifications other than the embodiment of the parking lot system.

FIG. 2 is a diagram illustrating a configuration of a vehicle number recognition device realized in the camera devices 26 and 36 of the embodiment.
As illustrated in FIG. 2A, the vehicle number recognition apparatus according to the embodiment includes a camera 51, an illuminance sensor 52, a shooting condition control unit 61, and an image processing unit 62. The photographing condition control unit 61 and the image processing unit 62 are realized in the processing circuit 60. The processing circuit 60 is preferably realized as a computer including a CPU, and the photographing condition control unit 61 and the image processing unit 62 are preferably realized in the processing circuit 60 by hardware and firmware. Although not shown, the camera devices 26 and 36 are provided with a camera control unit that performs overall control including the vehicle number recognition device. The camera control unit controls the image processing unit 62 to execute the vehicle number recognition process on the digital image data received from the camera 51 according to the signals from the photoelectric sensors 23 and 33.

  The camera 51 has a shooting direction and an image size so as to take a picture of a license plate of a vehicle entering the entrance and exit gates. Here, the camera 51 is a widely used industrial TV camera and will not be described in detail. Here, the camera 51 always outputs digital image data of a captured image, but when the photoelectric sensors 23 and 33 detect an entry to the gate according to the control signal from the camera control unit. Only digital image data can be output. The camera 51 is an infrared camera, and it is desirable to provide an illumination device that illuminates the photographing range with infrared rays as necessary. The lighting device may be of a flash type that emits light when the photoelectric sensors 23 and 33 detect an object entering the gate.

  Furthermore, the camera 51 adjusts exposure and gain as parameters for adjusting the brightness of the image to be captured. In the embodiment, by appropriately combining exposure and gain, shooting parameter 1 when the shooting environment is in a bright state, shooting parameter 2 when the shooting environment is in an intermediate state, and shooting parameter 3 when the shooting environment is in a dark state, Determine. The camera 51 selects three shooting parameters according to the set priority order, and performs shooting sequentially. However, the number of times that photographing can be performed is determined by the moving speed of the vehicle. Note that the camera control unit may control the shooting of the camera 51 in accordance with the shooting parameters.

  The illuminance sensor 52 is provided on the upper part of the camera 51, detects the illuminance near the shooting range of the camera 51, and outputs the detected illuminance data to the shooting condition control unit 61. It is desirable that the illuminance sensor 52 has a certain degree of directivity so as to detect the illuminance in the vicinity of the approach path of the vehicle, which is the shooting range of the camera 51.

  FIG. 2B is a diagram illustrating a configuration example of the illuminance sensor 52. The illuminance sensor 52 includes a photodiode 71, an amplifier (amplifier) 72, a resistor 73, and an A / D converter 74. The photodiode 71 changes the flowing current corresponding to the amount of incident infrared rays. The amplifier circuit formed by the amplifier (amplifier) 72 and the resistor 73 converts the current flowing through the photodiode 71 into a corresponding voltage and outputs the voltage. The A / D converter 74 converts the analog voltage output from the amplifier circuit into corresponding digital illuminance data.

  The image processing unit 62 receives digital image data captured by the camera 51 and executes a vehicle number recognition process on the received digital image data. The vehicle number recognition process performed here is a widely used process using a known vehicle number recognition algorithm, and detailed description thereof is omitted. However, the image processing unit 62 generates data indicating the average luminance level of the background and the text in the recognized license plate and outputs the data as plate luminance information. The average luminance level of the background and the character is, for example, the average value of the luminance levels at a plurality of locations in the region that is determined to be definitely the background or the character.

FIG. 2C is a functional block diagram of the imaging condition control unit 61 realized as a computer.
As illustrated, the shooting condition control unit 61 includes a shooting parameter priority setting unit 81, a timer 82, and a day / night determination processing unit 83.

  The timer 82 is realized by a calendar function in the computer and calculates the time. If calibration is performed about once every six months using a calendar function built in a normal computer, the time can be calculated with an accuracy within a maximum error of 5 minutes. The day / night determination processing unit 83 calculates the sunrise and sunset times with minute accuracy from the time calculated by the timer 82 and the latitude and longitude of the installation location of the parking lot system of the embodiment, and whether the time is night. It is determined whether or not, and the determination result is output. The latitude and longitude of the place where the parking lot system according to the embodiment is installed are the latitude and longitude of the place where the parking lot system is installed, which is detected by the system installer using a portable terminal equipped with a GPS function. Enter the longitude. Further, a GPS receiving device may be built in the vehicle number recognition device to detect the latitude and longitude of the installation location.

  The process of calculating the sunrise and sunset times from the latitude, longitude, and time in the day / night determination processing unit 83 is performed by firmware, and can be executed with little influence on the arithmetic processing capability of the computer. This process can be executed in a short calculation time and does not adversely affect other processes of the computer.

  The shooting parameter priority setting unit 81 includes success / failure information and plate luminance information of the vehicle number recognition from the image processing unit 62, illuminance data from the illuminance sensor 52 received via the I / O port 63, and day / night determination processing unit 83. Based on the determination result of whether or not it is the night of the camera, the priority order of the shooting parameters is set, and the set priority order is output to the camera 51. The camera 51 performs shooting by selecting shooting parameters according to the priority order. Note that the camera control unit described above may set the shooting parameters of the camera 51 according to the priority order.

Next, a shooting parameter and an algorithm for setting the priority of the shooting parameter will be described.
As described above, in the embodiment, exposure and gain are adjusted as parameters for adjusting the luminance of the image to be captured. Here, in the embodiment, by appropriately combining exposure and gain, shooting parameter 1 when the shooting environment is in a bright state, shooting parameter 2 when the shooting environment is in an intermediate state, and shooting parameter when the shooting environment is in a dark state The three shooting parameters of 3 are set, and shooting is performed by selecting the shooting parameters according to the set priority order. However, the shooting parameters are not limited to three levels, and may be four levels or more.

In this embodiment, the luminance information of the background which is the part of the number plate without letters and characters of the license plate in the image that was taken immediately before and succeeded in recognition of the car number is mainly used, and it is inadequate if it is used. The information on whether it is night and the illuminance data are organically combined to appropriately set the priority order of the shooting parameters. However, regarding the luminance information of the plate, it is possible to use not only the immediately preceding data but also past data within a predetermined time period. For example, a value obtained by multiplying the past plate luminance data within a predetermined time by a weighting coefficient corresponding to the elapsed time may be added, and the added value may be used as plate luminance information, which is calculated by the following equation.
Weight with weight = previous brightness x weight of elapsed time from previous time + previous brightness x weight of elapsed time since last time + ... + last brightness x weight of elapsed time since last time A method of using the information on the license plate will be described.

  In the method using the information of the license plate photographed immediately before this, the priority order of the photographing parameters is set based on the brightness of the background, which is the portion of the license plate without the characters, from the license plate image recognized by photographing immediately before. If the brightness of the character is extracted and this brightness is darker than the reference brightness, shooting parameters that can be taken brighter in the next shooting are adopted, and conversely, if the brightness is brighter than the reference, In the next shooting, shooting parameters that can be shot darker are adopted.

  FIG. 3 shows the case where the shooting parameters are switched in accordance with the above main setting algorithm, the license plate for business use is fixed at a fixed position, the images continuously shot at 5-minute intervals are processed, and the background and characters of the license plate are displayed. It is a figure which shows the change of the brightness | luminance of several places which it recognized and was determined to be the area | region of a background or a character reliably. In FIG. 3, the solid line indicates the luminance of the character, and the dotted line indicates the luminance of the background. The horizontal axis is the time (minutes) and represents the elapsed time starting at 9 am, showing the case where the first process was performed at 9:20, the 9-hour process was performed, and the process ended at 18:20. . The vertical axis represents the luminance level, which can take 0 to 255 levels with 8-bit data.

  First, when 20 minutes have passed, the operation is started with the shooting parameter 2. At 145 minutes (11: 5) from the start, the brightness of the characters exceeds the upper threshold (240), so the shooting parameter 1 is switched.

At 245 minutes (13:05) from the start, the brightness of the background fell below the lower threshold (16), so the shooting parameter 2 was restored.
At 490 minutes (17:10) from the start, the brightness of the background fell below the lower threshold (16), so the shooting parameter 3 was switched to.

  As shown in FIG. 3, by switching the three-level parameters, the brightness of the character and the background does not saturate near 0 and 255, and the contrast between the character and the background is sufficiently large. It can be seen that the characters can be recognized.

  In FIG. 3, the setting of the shooting parameter that gives the highest priority has been described, but the second and third priority are set as follows, for example. When the shooting parameter 1 or 3 is given the highest priority, the shooting parameter 2 is set to the second priority, and the remaining is set to the third. When the shooting parameter 2 has the highest priority, for example, when the average value of the brightness levels of characters and background is equal to or greater than the intermediate value (128), the shooting parameter 1 is set to the second priority, and the shooting parameter 3 is set. If the average of the brightness levels of characters and background is less than the intermediate value (128), the shooting parameter 3 is set as the second priority, and the shooting parameter 1 is set as the third priority. And

  In this setting algorithm, a change in luminance is detected from the image taken immediately before, and the shooting parameter is made to follow the change. Since this method pinpoints the brightness of the license plate that is the final character recognition target, the control accuracy is high, and if the priority of the shooting parameters is controlled according to this method, recognition errors will be detected. The incidence can be reduced.

  However, this method has the following problems. (1) When the license plate and the vehicle number cannot be recognized for some reason, the illuminance of the environment to be followed is unknown. (2) If the time interval of vehicle entry is sufficiently short with respect to the change in illuminance of the shooting environment, the change during that time is small and the tracking can be performed correctly. However, if the time interval is long, the shooting environment is large during that time interval. There is a high possibility that it will change, and tracking may not be performed correctly. (3) Although related to (2), there are cases where the tracking cannot be performed correctly even when the time zone in which the vehicle does not enter is long.

  As a specific example of (1), the car number is recognized because the license plate is so dirty that the car number cannot be recognized, or because the license plate has a frame or decoration that hides part of the letters. It is a case where it cannot be done. Further, this is a case where a shadow of a building or the like is projected on a part of the license plate.

  As a specific example of (2), the flow of clouds is fast and the sun and shade are frequently repeated. As a specific example of (3), the shooting parameters set based on the image of the vehicle that entered and exited the night before the night in a shopping center that is closed at night are used for the shooting of vehicles that enter and exit the next morning. This is the case.

  As described above, the priority setting method based on the main setting algorithm for detecting a change in luminance from the image taken immediately before and tracking the shooting parameter in accordance with the change is the vehicle in the cases (1) to (3) described above. There is a high possibility that a number recognition error will occur.

Next, advantages and disadvantages of day / night determination will be described.
Day / night judgment has the advantage that it is not affected by the state of the license plate or the time interval of vehicle entry.

  However, (1) Since there is no weather information, the influence of weather cannot be excluded, (2) The influence of shadows due to terrain and buildings cannot be excluded, (3) The day and night can be judged, but the illuminance of the shooting environment is a numerical value It has the disadvantage of not being obtained.

Further, advantages and disadvantages of the method using the illuminance data will be described.
The use of illuminance data has the advantage that (1) it is not affected by the state of the license plate or the time interval of vehicle entry, and (2) it can be measured including the effects of weather and shadows.

  However, (1) Since the brightness of the surroundings is measured without pinpointing the brightness of the license plate, it is affected by the color of the vehicle and the ground, and the control accuracy is low, and (2) it is affected by the headlight. , Has the disadvantages.

  In the present embodiment, the three methods having the advantages and disadvantages described above are organically used, and vehicle number recognition errors are reduced by a simple shooting parameter priority setting algorithm. Specifically, using the luminance information of the background that is the part of the number plate without letters and characters on the license plate in the image that was taken immediately before and successfully recognized the car number, if it is insufficient to use it, The priority of the shooting parameters is appropriately set using the information about whether it is night or the illuminance data.

FIG. 4 is a flowchart illustrating priority order setting processing in the imaging parameter priority order setting unit 81 according to the embodiment.
In S11, the photoelectric sensors 23 and 24 and 33 and 34 detect the entry of the vehicle, and if entered, the process proceeds to S12, and if not entered, the process proceeds to S21.
In S12, the camera 51 captures an image including the license plate of the vehicle that has entered in accordance with the set priority order, and the image processing unit 62 recognizes the license plate and the vehicle number.

In S13, it is determined whether the vehicle number recognition is successful or unsuccessful. If successful, the process proceeds to S14, and if unsuccessful, the process proceeds to S16.
In S <b> 14, the image processing unit 62 calculates the background of the license plate and the luminance of the characters (plate luminance), and transmits them to the shooting parameter priority setting unit 81.

  In S15, the shooting parameter priority setting unit 81 sets the shooting parameter priority according to the algorithm described above, and the process returns to S11. Specifically, if the background brightness of the license plate and the brightness of the characters are darker than the reference brightness, the shooting parameter for shooting one step brighter is given the highest priority, and conversely the brightness is brighter than the reference. If so, the shooting parameter for shooting one step dark is given the highest priority. As a result, it is possible to shoot following the illuminance change in the shooting environment. In normal times, the operations from S11 to S15 are repeated.

  In S16, the same shooting parameter priority order as the previous time is maintained, and the process returns to S11. In this case, it is considered that the license plate and characters could not be recognized due to the contamination of the license plate as described above. If a new priority order of shooting parameters is set for such an image, there is a possibility that the priority order of shooting parameters is not preferable. Therefore, in this case, the priority of the shooting parameters is not changed, and the next shooting is performed with the same shooting parameter priority. If the recognition fails, the elapsed time is measured assuming that no vehicle has entered.

In S21, it is determined whether the no entry time exceeds the threshold value. If it exceeds, the process proceeds to S22, and if not, the process returns to S11.
The normal operation is based on the premise that the time interval for entering the vehicle is shorter than the time during which the illuminance changes in the shooting environment. If the vehicle has not entered for a long time, the illuminance of the shooting environment will change greatly, and the letters on the previous license plate and the brightness of the background cannot be referenced. For this reason, the elapsed time from the previous vehicle entry is measured, and when the determined threshold value is exceeded, another process of S22 to S26 regarding “operation at return” is executed.

In S22, the day / night determination processing unit 83 calculates the sunrise or sunset time.
In S23, the day / night determination processing unit 83 compares the current time output by the timer 82 with the calculated sunrise or sunset time, and determines whether the current time is night, in other words, whether it is a dark state with no sun. If it is night, the process proceeds to S24. Otherwise, the process proceeds to S25.

  In S24, the shooting parameter 3 which is the shooting parameter in the dark state is set with the highest priority, and the process returns to S11. In this case, since the sun does not come out, even if the illuminance data indicates a bright illuminance, it is considered that it is due to a disturbance such as a headlight. Therefore, at night, the shooting parameter 3 is set regardless of the illuminance data and weather conditions.

In S25, the shooting parameter priority setting unit 81 reads the illuminance data.
In S26, the priority order of the shooting parameters is set based on the illuminance data, and the process returns to S11. In this case, if the illuminance data is higher than the first threshold (when bright), the shooting parameter 1 is given the highest priority, and if the illuminance data is lower than the first threshold and higher than the second threshold (medium brightness). ), The shooting parameter 2 is given the highest priority, and if the illuminance data is lower than the second threshold value (when dark), the shooting parameter 3 is given the highest priority.

  As described above, the parking lot system of the embodiment and the vehicle number recognition device used therein have been described. In the parking lot system described above, when the vehicle number recognition device of the present embodiment that dynamically changes the priority order is used. The results of 24-hour fixed-point observation using the case where the vehicle number recognition apparatus in which the priority order of the imaging parameters is fixed in the order of the imaging parameters 3, 2, 1 are used will be described. This observation took place in the spring, and the weather at the time of the observation was fine. During the 24 hours, 899 vehicles entered and exited the parking lot.

  According to the results of the above observation, when the car number recognition apparatus of this embodiment is used, in most cases, the recognition succeeds with one shooting, and the time required for the recognition is about 0.5 seconds. Yes, it was about 0.8 seconds at the longest. On the other hand, when the priority order is fixed, the number of times of recognition time between 1.6 and 2 seconds is about 530 times, and the number of times of recognition time between 0.5 and 1.2 seconds is about. It was 250 times, and the case where the recognition time was 2 seconds or more was about 100 times. A recognition time of 1.6 to 2 seconds corresponds to a case where the recognition succeeds in two shootings, a recognition time of 0.5 to 1.2 seconds corresponds to a case where the recognition succeeds in one shooting, A recognition time of 2 seconds or more corresponds to a case where the recognition succeeds after three shootings.

  As described above, by using the vehicle number recognition apparatus of the embodiment, the vehicle number of most vehicles can be recognized within 0.8 seconds, and vehicle number recognition errors can be reduced, and the vehicle number recognition time as a whole. Was shortened.

  The embodiment has been described above, but it goes without saying that various modifications are possible. For example, as described above, the shooting parameters may be four or more stages. In this case, the number of threshold values for determining the luminance information and illuminance data of the license plate when shooting is successful is determined to be night. Corresponding to the darkest shooting parameter.

  The present invention is applicable to a vehicle number recognition device and a vehicle number recognition method for recognizing a vehicle number of a license plate from an image taken by a camera installed outdoors.

DESCRIPTION OF SYMBOLS 1 Parking lot 2 Entrance gate 3 Exit gate 4 Advance adjustment machine 23,24,33,34 Photoelectric sensor 25 Parking ticket issuing machine 26,37 Camera apparatus 27,37 Gate apparatus 35 Fully automatic adjustment machine 40 Management apparatus 41 Communication path 51 Camera 52 Illuminance Sensor 60 Processing Circuit 61 Imaging Condition Control Unit 62 Image Processing Unit 81 Imaging Parameter Priority Setting Unit 82 Timer 83 Day / Night Determination Processing Unit

Claims (4)

Take an image including the license plate of the approaching vehicle with a camera installed outdoors,
A vehicle number recognition method for processing image data captured by the camera and recognizing a vehicle number of the license plate,
Based on the brightness of the background and characters in the image of the license plate when the vehicle number is successfully recognized, set the priority of the shooting parameters that define the shooting conditions according to the illuminance of the shooting environment,
When the elapsed time since the last recognition of the car number is greater than or equal to a predetermined value, the day / night judgment is performed by the calendar function. In the case of night, the parameter corresponding to the night is the highest priority. The vehicle number recognition method is characterized in that the priority order of the shooting parameters is set in accordance with the illuminance detected by the illuminance sensor in cases other than night. A camera that is installed outdoors and captures an image including a license plate of an approaching vehicle, and the brightness of the image to be captured can be set in multiple stages;
An image processing unit that processes image data captured by the camera and recognizes a vehicle number of the license plate;
An illuminance sensor for detecting illuminance in the vicinity of the camera to be photographed;
A shooting condition control unit that sets priority of shooting parameters that define shooting conditions according to the illuminance of the shooting environment based on the image data processed by the image processing unit and the illuminance detected by the illuminance sensor. A vehicle number recognition device,
The photographing condition control unit
Based on the background and the brightness of the characters in the image of the license plate when the vehicle number is successfully recognized by the image processing unit, set the priority order of the shooting parameters,
Finally, when the elapsed time since the recognition of the vehicle number by the image processing unit is greater than or equal to a predetermined value, day / night determination is performed by the calendar function. A vehicle number recognition device that is set so as to be given the highest priority and sets the priority of the photographing parameters in accordance with the illuminance detected by the illuminance sensor when it is other than night.   3. The vehicle number recognition according to claim 2, wherein the camera can set a plurality of shooting parameters for the brightness of the image, and in the case of night, the shooting parameter corresponding to a case where the shooting environment is dark is given the highest priority. apparatus.   The calendar function calculates the sunrise and sunset times from the calculated time calculated by the built-in clock function and the latitude and longitude of the set location of the vehicle number recognition device, and calculates the calculated times to the sunrise and sunset. The vehicle number recognition apparatus according to claim 2 or 3, wherein it is determined whether it is night compared to time.

Images