JP4887104B2 - Parking lot management system - Google Patents

Description

  The present invention relates to a parking lot management system that reads a vehicle license plate and manages entry and exit of a vehicle based on a result of collation with a database when the vehicle enters and leaves the parking lot. The present invention relates to a determination method suitable for use in an unmanned parking lot that is devised to improve the recognition rate by combining pattern matching processing and optical character recognition (OCR) processing. .

  Various methods have been proposed for the management of parking lots, but these methods are inadequate with their respective advantages and disadvantages. For solving these drawbacks, for example, there are “vehicle number recognition device and recognition method” described in Patent Document 1 and “number plate recognition device” described in Patent Document 2.

  In the “vehicle number recognition device and recognition method” described in Patent Document 1 above, if the vehicle number plate cannot be recognized by imaging, the clerk inputs the number information by voice and converts it into a text to complement the number information. To improve the overall recognition accuracy.

Further, in the “number plate recognition device” described in Patent Document 2, in order to improve the recognition accuracy of the vehicle number, the number plate is imaged from different directions by a plurality of cameras.
JP-A-8-249415 JP-A-7-146916

  However, in the method of the above-mentioned patent document 1, since the improvement of recognition accuracy cannot be realized unless the clerk monitors the captured image, the labor cost cannot be saved. In addition, the method of Patent Document 2 has a problem that the overall cost increases because a plurality of cameras are used.

  Therefore, the technical problem of the present invention is to provide a parking lot management system devised so as to improve the recognition accuracy without incurring the cost of manpower and equipment as a means for improving the recognition accuracy of the license plate. .

In order to solve the above technical problem, the parking management system according to the present invention includes an entrance lane and an entrance lane in the parking lot, respectively , as described in claim 1, and the entrance lane includes the ticketing machine and the entrance lane. A gate device connected to the ticketing machine via a communication line, an imaging processing device, and a trigger sensor are provided, and a gate device, an imaging processing device, and a trigger sensor connected to the ticketing machine via a communication line are provided for the settlement lane, The imaging processing device is provided with a camera, an external input unit, an external communication unit, a memory, and a central processing unit, and is installed in a parking lot management center and connected to the imaging processing device via a LAN to perform OCR processing and OCR results. A central management PC having a transmission function, an image reception function, a DB search, and a corresponding vehicle presence / absence verification function is provided, and trigger sensors installed in the entrance lane and the exit lane are provided. If it is determined that a vehicle Ri, by said trigger signal to the external input unit of the image pickup apparatus is input to the imaging by the camera of the imaging and processing device, reads the license plate of the vehicle, entering the parking lot A parking lot management system for managing entry, and image data of a number of contracted vehicles and existing license plates stored in the memory of an imaging processing device installed in the entrance lane or entry lane of the parking lot, and the imaging The central processing unit of the processing device performs pattern matching processing with the image data of the license plate of the approaching vehicle generated from the captured image, and in parallel, the image of the vehicle that entered the lanes and the license plate are captured Is transmitted to the central management PC, and the central management PC performs OCR processing on the received image, and the OCR result and vehicle management And collating the chromatography data, whether the result corresponds to the resident vehicle or contract vehicles reply to the above-described imaging apparatus, in the imaging processing apparatus, the pattern matching process is the result of no matching, and When the OCR processing result indicates that there is a corresponding vehicle , the OCR result OK is transmitted to the ticket issuing machine or the checkout machine, and the ticket issuing machine or the checkout machine that has received this OK signal is connected to the gate device installed in the lane. In this case, the gate opening signal is output.

Moreover, parking management system of the present invention, as described in claim 2, the positional information of the license plate that is included in the image captured by the imaging apparatus installed in the parking lot entry lane or participate lane, the generated by the central processing unit of the imaging processing unit, wherein the position information of the license plate along with captured images and transmitted to the central management PC, the position information of the license plate of the captured image the central management PC is received Based on the above, the OCR process is performed only on the image of the area where the position of the license plate is specified.

  According to the parking lot management system according to claim 1, even if there is no matching in the pattern matching process, matching is performed if there is corresponding license plate data by checking with the in-vehicle DB or contracted car DB in the OCR process. If the recognition accuracy of the pattern matching process is not 100%, the accuracy can be supplemented by the OCR process, and the recognition accuracy close to 100% is unmanned and the cost of the apparatus. It can be realized without up.

  Furthermore, according to the parking lot management system according to claim 1, by allowing the central management PC to take charge of OCR processing with a long processing time, it is usually possible to operate in a quick processing time only by pattern matching processing, and no matching is performed. By judging from the reply from the central management PC only when it comes out, the judgment returns with a slight delay only when it is not recognized by the pattern matching process. Management can proceed smoothly.

  In addition, according to the parking lot management system according to the second aspect, the scope of the OCR process performed by the central management PC can be limited in advance using the license plate position information generated in the imaging processing apparatus, so the processing scale is reduced. Thus, even when waiting for OCR processing for no matching, the determination result is returned without causing a significant delay, and prompt processing is possible without annoying the user.

  As described above, according to the parking lot management system of the present invention, without increasing the cost of manpower and equipment, improve the accuracy of number reading recognition, eliminate congestion of the communication network in the parking lot, The advantage of facilitating data transactions can be exhibited.

  DESCRIPTION OF EMBODIMENTS Hereinafter, a parking lot management system according to an embodiment of the present invention will be described with reference to the drawings. FIG. 1 shows a configuration of a parking lot, where MP is a parking space, And second entry lanes 10a and 10b, and first and second entry lanes 20a and 20b, respectively, and each of these entrance and exit lanes 10a, 10b and 20a and 20b has a first loop. 11a, the second loop 11b, islands 12a and 12b, and one-side island 12c are provided.

  In each of these islands 12a, 12b, and 12c, 13 is a light emission trigger sensor (light emission side), 14 is a light reception trigger sensor (light reception side), 15 is a ticketing machine, 16 is a communication line, and 17 is a gate device provided with a gate bar 18. , 19 is an image pickup processing device, and 22 is a settlement machine. In FIG. 1, reference numeral 21 denotes a prior payment machine, 23 a communication line, 24 an Ethernet Ethernet (registered trademark), and 50 a and 50 b vehicles. Further, in FIG. 1, reference numeral 30 denotes a parking lot management center. The parking lot management center 30 is a server 31, a central processing unit (hereinafter referred to as a central management PC) 32 such as a personal computer, and a database (DB) 33. The database 33 includes a contract vehicle database (DB) 33a, an on-vehicle database (DB) 33b, and an image database (DB) 33c.

  As an example, a parking lot having two lanes for entry and two lanes for entry is shown, and the right side of the first and second entry lanes 10a, 10b is the islands 12a, 12b where the light receiving trigger sensor 14 is installed, respectively. A light emission trigger sensor 13 is installed on each of the front side of the central island 12b on the left side of the direction and the one side island 12c. An infrared beam is emitted from each light emission trigger sensor 13 to each light reception trigger sensor 14, and when this is blocked by the vehicle 50a or 50b, the trigger is turned on. However, in the first and second entrance lanes 10a and 10b, when the metal parts of the vehicles 50a and 50b are detected by the first loop 11a and it is determined that there is a vehicle, the trigger is determined to be on. It is also possible to have an AND configuration, and it is possible to prevent triggers from being turned on when a person or small animal passes.

  Further, each imaging processing device 19 is connected to the ticket issuing machine 15 via the communication line 16, and the ticket issuing machine 15 is connected to the gate device 17. The image pickup processing device 19 is connected to the image pickup processing device 19 and the central management PC 32 of other entrance lanes and entry lanes via the Ethernet 24 via a LAN.

  A pre-payment machine 21 is installed at one corner of the parking lot space MP, and is connected to the central management PC 32 via a communication line 23. Since the first and second entry lanes 20a and 20b are configured by rotating the directions of the first and second entry lanes 10a and 10b by 180 degrees, a detailed description of the structure thereof is omitted. However, in the islands 12a and 12b of the entry lanes 20a and 20b, a settlement machine 22 is installed instead of the ticket issuing machine 15.

  FIG. 2 shows the configuration of the imaging processing apparatus 19 described above. The imaging processing device 19 includes a central processing unit (CPU) 34, a memory 35, an external input unit 36, an external communication unit 37, a LAN board 38, a camera 39, a capture board 40, a flash 41, and an interface 42. Are connected via a bus 43. The imaging processing device 19 has the same configuration for the first and second entry lanes 10a and 10b and the first and second entry lanes 20a and 20b, but each device ID, IP address for LAN connection, etc. However, the processing program itself has a slightly different mechanism for entry and exit lanes.

  The light emission trigger sensor 13 always emits an infrared beam. When the light reception trigger sensor 14 is shielded by an object, a trigger signal is input to the imaging processing device 19 and the flash 41 is picked up and an image is taken by the camera 39. At this time, the flash 41 may be controlled using an optical sensor that measures the amount of light in the periphery, or an auto iris mechanism may be provided in the lens unit of the camera 39. In the embodiment, the captured image is converted into a still image from the camera 39 by the capture board 40, and pre-programmed application software operates on the CPU 34 to extract the license plate of the vehicle 50a or 50b, binarize, and compress the image compression data. The data is generated and stored in the memory 35.

  In the example of the imaging processing device 19 in the entrance lanes 10a and 10b, the memory 35 stores a large number of compressed image data of contracted vehicles. In the embodiment, data is stored in the memory 35, but a storage device such as an HDD or a compact flash (registered trademark) may be used. Application software programs and the like are stored in advance in a part of the memory 35 so as not to be rewritten.

  In the pattern matching processing, pre-programmed application software operates on the CPU 34 and performs processing by comparing the compressed image data generated from the captured image with the compressed image data stored in the memory 35. It is also possible to perform a series of operations such as license plate extraction, binarization, generation of compressed image data, and pattern matching processing by providing a dedicated image board.

  Pattern matching processing and OCR processing each have advantages and disadvantages (good and bad). In the pattern matching process, as a cause of no matching (disagreement) (weak point of pattern matching process), the vehicle approach angle (posture) at the time of entry and entry is, for example, the vehicle that entered the lane while turning in the opposite direction Image pattern matching processing tends to be no matching. This is likely to be a non-matching because the number of characters on each license plate is different due to the difference in distortion direction between the images of the license plates to be compared. In contrast, in the OCR process, each character is read and replaced with text, so even when a license plate image is distorted in the opposite direction when entering and exiting, the images are compared with each other. Because it is not a translation, the letters on the license plate can be recognized normally.

The weakness of the OCR process is that if the characters on the license plate are faint or dirty, it may not be recognized as a character, or other similar text will be applied, resulting in false recognition, but pattern matching Processing at entry and exit
In order to directly compare these images, it is determined whether or not the same vehicle is included, including the blurring and dirt on the letters on the license plate, and no erroneous recognition occurs.

  The pattern matching process is fast, and the pattern matching process in the entry lanes 20a, 20b only needs to select the best matching (the highest matching score = the highest number of matching points) from the data for the number of entries. Is limited to the number of vehicles present and the number of objects to be processed is small. In addition, in the entrance lanes 10a and 10b, the number of objects is limited to a regular contract vehicle. On the other hand, in the OCR process, since the recognition process is always performed for each character in consideration of the possibility of all characters, the entire process takes time. When both the pattern matching process and the OCR process are executed on the imaging processing device 19 side of the entrance lanes 10a and 10b and the entrance lanes 20a and 20b, the entire process takes time, and the CPU 34 has a fast clock frequency to cover it. And the capacity of the internal memory 35 must be increased, which increases the cost of the apparatus.

  As a defect of the pattern matching process, an example of a captured image is shown in FIG. It is assumed that the vehicle 50 that has entered the entrance lane 10 turns left and enters. When entering a right turn, it may pass the trigger sensor in such a posture. The image of the license plate 51 at that time is taken as shown in FIG. On the other hand, when the vehicle 50 participates, if it enters the participation lane 20 by turning leftward, it will turn right and enter the opposite direction. The image of the license plate 51 at that time is taken as shown in FIG. If the vehicle 50 has entered the entry lane 10 or the entry lane 20 straight, an image as shown in FIG. 3B is obtained for both the entry and entry lanes.

  FIG. 4 conceptually shows the license plate extraction process and the compressed image data generation process. Eventually, the pattern matching process is performed with a complex algorithm, but conceptually explained, the compressed image data is overlaid, and a score is assigned by the number of matching parts, which exceeds a preset threshold value. “Matching” if there is, “No matching” if the value is below the threshold value, and brute force processing between the imaged image and the stored image and selecting “Best Matching” from “Matching” .

  In the case of FIG. 4, (b ') and (b') or (b ') and (c') match, but (b ') and (c') indicate "no matching". May be. If the threshold value is lowered or “algorithm” is changed so as to be “matching”, there is a risk that “mismatching” will occur in which different license plates are regarded as “matching”. Accordingly, license plate images that differ greatly in the direction of image distortion due to a difference in the posture of the vehicle may be determined as “no matching J. If OCR processing is performed, (i), (b), ( In (c), the license plate information can be recognized normally, and for example, text data “Nagoya 400 wa 0444” can be obtained as a processing result.

  As a disadvantage of the OCR processing, when a parking lot system is constructed only by the OCR processing, there are cases where the license plate is misread as described above, and the OCR processing itself takes a processing time.

  In other words, the OCR process is a “pattern matching process” between a character part included in an image and characters in a preset form (font). However, since there are a large number of target set characters, “brute force” one character at a time. If you apply this, you will follow a huge number of processing steps. Therefore, since the processing time is long, high speed processing is required to install it in a device installed in the entrance / exit lane of the parking lot, and a device with a high CPU clock frequency is required. Installation increases costs. If there are other high-spec PCs, such as the central management PC 32, that are responsible for high-speed processing, the OCR processing is performed there, and the increase in the cost of the CPU affects the overall cost. It can also be made difficult.

  However, since it takes time to send images for OCR processing, only a system in which images from multiple lanes are intensively OCR processed at one location and the results returned to the entrance and exit lanes. In order to manage the entrance and exit of the parking lot, it is necessary to always wait several seconds for image transmission and result reply, and this alone is not practical (always slow response). Therefore, in view of the advantages and disadvantages of the pattern matching process and the OCR process described above, the parking lot management system according to the embodiment of the present invention executes processes as shown in FIGS.

  FIG. 5 shows an operation flow of the ticket issuing machine 15 when entering the vehicle. In step S1, it is determined whether the first loop 11a is on or off. If it is on, the process proceeds to step S2 and a trigger acceptance permission signal is shown in FIG. The image is transmitted to the imaging processing device 19 installed in the lane. Parking data including a matching result is received from the imaging processing device 19 shown in (B) in step S3, the matching result is judged in step S4, and if no matching, the process proceeds to step S5, and the imaging processing shown in (B '). If the OCR result is NO (no match) from the device 19, the process proceeds to step S6 and the guidance broadcast is started. When the ticketing button is pressed in step S7, a parking ticket is issued in step S8. In step S9, it is determined whether or not the parking ticket or commuter pass is extracted. If it is extracted, the gate bar 18 is opened in step S10. In step S11, it is determined whether or not the vehicle 50a or 50b has passed the second loop 11b. If it has passed, the gate bar 18 is closed in step S12, and the parking data such as entrance time and ticket number are updated in step S13. At the same time, a trigger reception prohibition command is transmitted to the imaging processing device 19 shown in FIG. If the ticket issuing button is not pressed in step S7, the process proceeds to step S14 to determine whether or not a commuter pass is inserted. If it is inserted, the process proceeds to step S9, and if it is not inserted, the process continues in step S7. Further, if matching is performed in step S4, the process immediately proceeds to step S9, and even if no matching is performed, if the OCR result is YES (match) from the imaging processing device 19 shown in (B ′) in step S5, At that time, the process proceeds to step S9.

In summary, for contract cars,
When the first loop 11a is turned on, a trigger permission signal is transmitted to the imaging processing device 19 as shown in FIG. In response to this, parking data including a matching result is output from the imaging processing device 19 as shown in FIG. When the vehicle 50a or 50b is a contract vehicle and the matching result is “matching”, a gate open signal is output, and when the vehicle 50a or 50b passes through the second loop 11b, a gate close signal is output. Thereafter, the updated data with the admission time added to the parking data is returned to the imaging processing device 19. At the same time, a trigger reception prohibition signal is transmitted to the imaging processing device 19.

For ordinary cars,
If the matching result is “no matching” in step S4 and the OCR result is “no match” in step S5, please press the ticketing button in the guidance broadcast and take out the parking ticket. If you have a commuter pass, "Please enter" is announced. If the ticket issuing button is pressed, a parking ticket is issued. When the parking ticket is extracted, a gate opening signal is output, and when the vehicle passes through the second loop 11b, a gate closing signal is output. After that, the admission time, ticket no. Etc. and the trigger acceptance prohibition signal are transmitted to the imaging processing device 19 as shown in FIG.

In case of no matching with contract car (1),
Even if it is a contract vehicle, if the matching result is “no matching” in step S4, the OCR result is transmitted from the imaging processing device 19 as shown in (B ′) in step S5, and the result matches with the OCR. In the case of OK determination that there is a contracted vehicle, the gate shifts to step S10, and when the vehicle passes through the second loop 11b, a gate closing signal is output. As shown in (C), the updated data added with the above and the trigger reception prohibition signal are transmitted to the imaging processing device 19.

In case of no matching with contract vehicle (2),
Even if it is a contracted vehicle, if the matching result is “no matching” in step S4 and the OCR result is “no match” in step S5, the guidance broadcast is started and the ticketing button is pressed in step S7 or in step S14. If the commuter pass is inserted and is valid while the commuter pass is inserted or is waiting, a gate opening signal is output when the commuter pass is extracted, and the vehicle passes through the second loop 11b. A gate closing signal is output. After that, the admission time and commuter pass no. As shown in (C), the updated data added with the above and the trigger reception prohibition signal are transmitted to the imaging processing device 19.

  As described above, even if there is no matching for some reason at the entrance of the contracted vehicle, if the OCR result is received by OK, the gate is opened and processed at that point in the same manner as the matching. The time from when the result of no matching is received in (B) until the OCR result OK is obtained again in (B ′) varies depending on the degree of congestion of the LAN, but the recent high-speed LAN and high-spec central management PC 32 With, it won't take as long as the driver gets frustrated.

  FIG. 6 shows an operation flow of the image pickup processing device 19 when entering the vehicle. In step S21, it is determined whether the trigger input is on or off. If it is off, it is installed in that lane in step S22 as shown in (A). When the trigger acceptance permission signal is received from the ticket issuing machine 15, but not received, the trigger input determination in step S21 is repeated. If the trigger input is on in step S21, it is determined in step S23 whether the trigger is permitted or prohibited. If it is in the permitted state, flash emission is appropriately captured and image capture is performed in step S24, and a license plate is extracted in step S25. Processing is performed, and parking data and an image including number position information are transmitted to the central management PC 32 as shown in (D) in step S26. In step S27, compressed image data is generated, and in step S28, pattern matching processing is executed. Thereafter, the process proceeds to step S29, the matching result is determined, and parking data including the matching result is transmitted to the ticket issuing machine 15 as shown in (B) in step S30. If the vehicle enters the parking lot while waiting as it is, the process proceeds to step S31, and as shown in (C), updated parking data including the entry time and ticket number are issued from the ticket issuing machine 15 and a trigger acceptance prohibition command. In step S32, the parking data and the compressed image data are transmitted to the central management PC 32 as shown in (E). On the other hand, if there is no matching in step S29, the process proceeds to step S33, where it is determined whether or not an OCR result has been received from the central management PC 32 as shown in (E). Parking data including the result is transmitted to the ticket issuing machine 15.

In summary, for contract cars,
Although one trigger input is monitored at step S21, the trigger reception is prohibited at step S23, and the process does not proceed further. If the trigger acceptance is permitted in step S23, the flash data is emitted, the image is captured, the image is captured, the license plate is extracted, the parking data to which the license plate position information is added, the captured parking data, and the parking data to which the captured image is added. The transmission of the captured image is started to the central management PC 32 in (D). Since it takes time to transmit the image, the process immediately proceeds to generation of compressed data in parallel, and a matching result is obtained through pattern matching processing. Here, the image transmission is processed by the LAN board 38 of the image pickup processing device 19, and the image matching data generation and pattern matching processing are processed by the application software running on the internal CPU 34. Yes. If the matching result is matching, a matching result (score, contract car license plate information, commuter pass No., etc.) is further added to the parking data and transmitted to the ticket issuing machine 15 in (B). Then, while waiting for the vehicle to pass, if (C) receives the update data in which the entry time is given to the parking data from the ticket issuing machine 15 of that lane, the parking management data and the image compression data are sent to the central management PC 32 (F). Send to.

For ordinary cars,
In step S29, the matching result is no matching, but the matching result (no matching score, etc.) is assigned to the parking data in the same manner as the matching, and is transmitted to the ticket issuing machine 15 in (B). On the other hand, even if there is no matching, it waits for the OCR result from the central management PC 32 in (E), and when it receives the result, again (B ′) gives the OCR result to the parking lot data and transmits it. In the case of a general vehicle, the text of the license plate and the search result “not applicable” of the contract DB 33a are transmitted as the OCR result.

If there is no matching with the contract car,
The process is the same as that for a general car, but the license plate text data and the search result “applicable” in the contract DB 33a are transmitted as the OCR result. As shown in the contents of the data, “land transportation branch office name, vehicle model number, hiragana application name, serial number of up to 4 digits” is given and transmitted to the ticket issuing machine 15 in (B ′). Then, when the update data in which the entry time is added to the parking data is received from (C) while waiting for the vehicle to pass, the parking data is transmitted to the central management PC 32 in (F).

  FIG. 7 shows an operation flow of the central management PC 32 when entering the vehicle. In step S41, parking data and images including number position information are received from the imaging processing device 19 as shown in (D), and OCR is received in step S42. Execute the process. In step S43, the contracted car DB 33a is searched, and the process proceeds to step S44. In step S44, it is determined whether or not there is a corresponding vehicle. If there is no corresponding vehicle, only the license plate information obtained by the OCR process is set in the data in step S45, and the OCR result is displayed in step S46 as shown in (E). To the imaging processing device 19. In step S47, the image is stored in the image DB 33c. In step S48, parking data and compressed image data are received from the imaging processing device 19 as shown in (F), and in-vehicle information is stored in the inventory DB 33b in step S49. In step S50, it is determined whether or not the vehicle is a contract vehicle. If the vehicle is not a contract vehicle, it is determined in step SSS53 whether or not there is an accumulated number. If there is an accumulated number, the process proceeds to step S51. If it is timed out, the process proceeds to step S51, where the number of stored items and time-out are initialized. In step S52, as shown in (F '), the parking data and the compressed image data are obtained from the imaging processing device 19 at each exit. Send to. If there is a corresponding vehicle in step S44, the process proceeds to step S55, the license plate information obtained by the OCR process and the information extracted from the contracted car DB are set in the data, and the process proceeds to step S46. If it is a contract vehicle in step S50, it will progress to step S51.

  In summary, as shown in (D), parking data and an image including license plate position information are transmitted from the imaging processing device 19 in a certain lane. In the OCR processing, character number extraction and character recognition in the surrounding area are performed with reference to the position information of the license plate, so that processing for searching for the number from the entire image can be omitted. Based on the number information obtained as a result of the OCR process, the contracted vehicle DB 33a in the server 31 provided alongside the central management PC 32 is searched in step S43. As a result of searching in step S44, if there is data of the corresponding vehicle, not only the number information but also the commuter pass number and the expiration date are stored in the contracted vehicle DB 33a in step S50, so that they are incorporated in the reply data as necessary. In step S46, the image data is transmitted to the imaging processing device 19 as shown in (E).

  If there is no corresponding vehicle data as a result of the search in step S44, only the information whose number was read in step S45 is incorporated, and the information is transmitted to the imaging processing device 19 shown in (E) in step S46. The image is stored in the image DB 33c of the server 31. Thereafter, in (F), the parking data and the compressed image data are received again from the same imaging processing device 19, and the information is stored in the on-board DB 33b. Next, if the entering vehicle at this time is a contract vehicle in step S50, the accumulated number and timeout count are initialized (set to 0) in step S51, and the parking data and the image compression data in step S52. Is transmitted to the imaging processing device 19 of each participating lane.

  If the vehicle entered at this time is not a contract vehicle in step S50, it is determined in step S53 whether the number of stored vehicles is N or in step S54, the timeout is M seconds or more, and both have reached the set values. If not, continue counting time-out. If the next admission vehicle arrives at this timing, it is sufficient to shift to the operation from START again. In this way, the parking lot data and the compressed image data are not distributed until the set number or time elapses when the vehicle is not a contracted vehicle. And compressed image data is distributed to the imaging processing device 19 of each entry lane, so that the traffic on the LAN is congested as compared to the case where the data is distributed to the imaging processing device 19 of a plurality of exit lanes at each entry. It is possible to avoid as much as possible, and to perform time-consuming image transmission / reception as fast as possible.

  FIG. 8 shows the operation flow of the checkout machine 22 when entering the vehicle. In step S61, it is checked whether the first loop 11a of the lane is on or off. If it is on, a trigger acceptance permission command is issued in step S62. It transmits to the imaging processing device 19 installed in the lane shown in (G). In step S63, parking data including the matching result is received as shown in (H), and the process proceeds to step S64 to check the matching result. If there is no matching, the process proceeds to step S65, and if the OCR result is NO (no match) is received from the imaging processing device 19 shown in (H '), the process proceeds to step S66 to start the guidance broadcast, and the parking ticket is received in step S67. It is determined whether or not a fee has been inserted. If it has been inserted, it is determined in step S68 whether or not a fee has been generated. If a fee has been generated, the process proceeds to step S69 to execute a fee settlement process. In step S70, the exit gate bar 18 is opened. Next, in step S71, it is checked whether or not the vehicle 50a or 50b has passed through the second loop 11b. If so, the gate bar 18 is closed in step S72, and in step S73. Parking data such as the entry time is updated, and a trigger reception prohibition instruction is transmitted to the imaging processing device 19 as shown in (J). If a parking ticket is not inserted in step S67, it is determined whether or not a commuter pass is inserted in step S74. If a commuter pass is not inserted, the process returns to step S67, and if it is inserted, the process proceeds to step S75. If matching is found in step S64, the process proceeds to step S68. Even if no matching is performed, it is checked in step S65 whether or not the OCR result from the imaging processing device 19 shown in (H ') matches. If so, the process proceeds to step S68. .

  In summary, the operation is basically the same as the operation of the ticket issuing machine 15 in FIG. 5, but whether the gate proceeds to step S70 depending on whether or not a charge is generated in step S68 even if the matching result is matching in step S64. , It will be divided whether proceed to fee settlement processing. Even if it is a regular car, even if it matches with the car in existence, if it is not yet settled, the fee settlement process is necessary. This can be determined from the status of the matching vehicle parking data.

  FIG. 9 shows an operation flow of the imaging processing device 19 when entering the vehicle. In step S81, it is determined whether the trigger input is on or off. If it is off, the lane shown in FIG. The reception permission command is received from the installed checkout machine 22, but if not received, the trigger input determination in step S81 is repeated. If the trigger input is turned on in step S81, it is determined in step S83 whether the trigger reception is permitted or prohibited. If the trigger input is permitted, in step S84, flash emission is appropriately captured and image capture is performed. Perform the extraction process. In step S86, the parking data and the image including the number position information are transmitted to the central management PC 32 shown in (K). In step S87, compressed image data is generated, and pattern matching processing is performed in step S88. In step S89, it is determined whether or not there is a matching result. The process proceeds to step S90, and parking data including the matching result is transmitted to the settlement machine 22 shown in (H).

  If the vehicle enters the parking lot while waiting as it is, the process proceeds to step S91, where the entry time and ticket number. The updated parking data and the trigger reception prohibition command such as are received from the settlement machine 22 shown in (J), and it is determined whether or not the vehicle is a contracted vehicle in step S92. If it is not a contract vehicle, the parking data and the compressed image data are deleted in step S93, and a parking data deletion instruction is transmitted to the imaging processing device 19 and the central management PC 32 of each participating lane shown in (M) in step S94. On the other hand, if there is no matching in step S89, the process proceeds to step S95 to check whether or not the OCR result is received from the central management PC 32 shown in (L), and in step S96, the parking data including the OCR result is (H ' ) To the checkout machine 22 shown in FIG. Furthermore, if it is a contract vehicle in step S92, it will progress to step S97 and will transmit parking data to the imaging processing apparatus 19 of each entrance lane shown to (N), and will return to step S93. If the trigger acceptance is prohibited in step S83, the process proceeds to step S98, the distribution acceptance process is performed, and the process returns to step S81.

  In summary, one trigger input is constantly monitored in step S81, but the trigger acceptance prohibition in step S83 does not proceed further. If the trigger acceptance is permitted in step S83, flash emission, imaging, image capture, license plate extraction processing, parking data with license plate position information and the captured image itself are sent to the central management PC 32 in (K). Start sending.

  When the matching result is matching in step S89, the matching result is further given to the parking data and transmitted to the settlement machine 22 in (H). Then, while waiting for the vehicle to pass, if the update data in which the participation time is added to the parking data is received from the settlement machine 22 of (J) in step S91, the step is performed if the vehicle that participated at that time is a contract vehicle. In S97, an instruction to update the parking data is transmitted to the imaging processing device of each entrance lane. The parking data update instruction transmitted at that time is that the compressed image data of the contracted vehicle stored in the imaging processing device of the entrance lane is “in-vehicle = invalid” after the entrance process (to prevent unauthorized re-entry )) Return the item to “Entered = Valid” and prepare for matching when the contracted car enters next time.

  If the vehicle that participated at that time in step S92 is a general vehicle, the parking data and the compressed image data in its own memory are deleted in step S93, and a similar deletion instruction is issued in step S94 for the imaging process of each participating lane. The data is transmitted to the device 19 and the central management PC 32.

  In this way, regarding the deletion and update instructions when the vehicle has passed, it is only necessary to send the parking data corresponding to the parking data to be deleted and updated and the compressed image data with a header indicating the deletion and update instructions, Since the amount of data is extremely small, it is distributed each time without being stored and sent. If the matching result is no matching in step S89, a matching result (such as a no matching score) is assigned to the parking data in the same manner as the matching, and the result is transmitted to the checkout machine 22 in (H). On the other hand, even if there is no matching, the OCR result from the central management PC 32 of (L) is waited, and when the result is received, the OCR result is added to the parking lot data and transmitted to the checkout machine 22 again (H ′). As a result, the settlement machine 22 receives the no-matching result once, but when the OCR result is returned as OK, it can shift to the same processing as the matching.

  FIG. 10 is an operation flow of the central management PC 32 at the time of participation. In step S101, license plate position information and an image are received from the imaging processing device 19 shown in (K), and OCR processing is performed in step S102. In step S103, the data in the in-vehicle DB 33b is searched, and in step S104, it is determined whether or not there is a corresponding vehicle. If there is no corresponding vehicle, only the OCR result is set as data in step S105, and the OCR result is transmitted to the imaging processing device 19 shown in (L) in step S106. In step S107, the image is stored in the image DB 33c. In step S108, an instruction command for deleting the parking data and the compressed image data is received from the imaging processing device 19 shown in (M), and the parking data is deleted from the on-board DB 33b in step S109. On the other hand, if there is a corresponding vehicle in step S104, the process proceeds to step S110, the number information obtained by the OCR process and the information extracted from the on-board DB are set in the data, and the process proceeds to step S106.

  In summary, as shown in (K), parking data and an image including license plate position information are transmitted from the imaging processing device 19 of a certain entry lane. In the OCR processing, character number extraction and character recognition in the surrounding area are performed with reference to the position information of the license plate, so that processing for searching for the number from the entire image can be omitted. Based on the number information obtained as a result of the OCR process, the in-vehicle DB 33b in the server 31 provided along with the central management PC 32 is searched in step S103. As a result of the search in step S104, if there is data of the corresponding vehicle, not only the number information but also the ticket number and the entry time are stored in the on-board DB 33b in step S110. In S106, as shown in (L), the image is transmitted to the imaging processing device 19 of the participating lane. As a result of the search in step S104, if there is no data of the corresponding vehicle, only the information whose number was read in step S105 is incorporated, and in step S106, as shown in (L), it is transmitted to the imaging processing device 19 of the entry lane, Thereafter, the image is stored in the image DB 33c of the server 31. As shown in (M), when the parking data and the compressed image data are received again from the imaging processing device 19 of the same entry lane, the parking data is deleted from the on-board DB 33b in step S109, but the data is moved to the history DB (not shown) as it is. You may save it.

  FIG. 11 is a distribution acceptance processing operation flow. In step S111, it is determined whether additional data has been received. If not, the process proceeds to step S112, in which it is determined whether deletion data has been received. If no deletion data has been received, the process returns. If additional data has been received in step S111, the process proceeds to step S113, where parking data and compressed image data are received from the central management PC 32 shown in (F '), and parking data and compressed image data are stored in step S114. If the deletion data has been received in step S112, a parking data deletion command is received from another imaging processing device shown in (M) in step S115, and the parking data and the compressed image data are deleted in step S116.

  In summary, as shown in step S98 in FIG. 9, the imaging processing device 19 in the entry lane is always in the distribution acceptance state when in the idle state. If neither additional data nor deleted data is received, the system returns to the original trigger wait state and constantly monitors them. If the additional data is received in step S111, the parking data and the compressed image data at that time are received and stored in the memory. If the deletion data is received in step S112, the corresponding parking data and compressed image data are deleted from the memory according to the instruction.

  According to the present invention, even in the case of no matching in the pattern matching process, there are many cases where the license plate is normally imaged. In that case, if the corresponding information is found by searching the vehicle presence information from the OCR process result, “ Handled in the same way as “matching” (= some matches). Also, the pattern matching process is performed by the imaging processing device 19 for the entrance lane and the entry lane, and at the same time, the image is transmitted to the central management PC 32 via the LAN. The internal CPU is fully operational when the pattern matching processing is being executed by the imaging processing device 19, but communication processing such as image transmission is performed by a dedicated LAN board 38 (or simply a LAN dedicated circuit), so parallel processing is performed. Even if an image is transmitted with, the pattern matching process does not slow down.

  Since the load on the CPU is large in the OCR process, an image is transmitted to the central management PC 32 equipped with a high-spec CPU, the OCR process is performed based on the image, and only the result is returned to the imaging processor 19. When the imaging processing device 19 transmits an image to the central management PC 32, the position data of the license plate included in the image is added and transmitted. The central management PC 32 can shorten the time of OCR processing by extracting the license plate image and performing OCR processing only for the image area including the license plate position information. When the image matching data for pattern matching processing of the entered vehicle is distributed (sent to a plurality of devices) to the imaging processing devices 19 of all the participating lanes 20a and 20b, the number of the imaging processing devices 19 of the entering lanes The information is stored in a single area or transmitted to the central management PC 32 and distributed from the central management PC 32 to the image processing apparatus 19 at a certain timing. At that time, a time-out value is set in the process of accumulating several items, and if a time-out occurs, distribution is performed even if a plurality is not accumulated. As a result, LAN traffic is free and image transmission can be performed smoothly.

  In addition, in the central management PC 32, when pre-clearing is performed with the parking ticket of the vehicle corresponding to the data being accumulated, the accumulated data is sent to the imaging device of each entry lane before the above timeout or accumulation setting number. Even so, distribution will start immediately. As a result, there is no problem that when the entry, advance settlement, and entry are performed quickly, the data that has been prepaid is being accumulated in the central management PC and it is slow to reach the entry lane device.

  Furthermore, the accumulated data is limited to ordinary cars. When a regular car enters, the accumulated data is distributed at that time. Thus, even if a delivery vehicle or the like mainly enters and exits quickly, the data delivery to the exit lane is not delayed as described above.

The block diagram of the parking lot management system by the Example of this invention. The block diagram which shows the structure of the imaging processing apparatus used with the parking lot management system by the Example of this invention. Explanatory drawing which shows the captured image of a pattern matching process. Explanatory drawing which shows the plate extraction process of a pattern matching process, and an image compression process. The operation | movement flowchart of the ticket issuing machine used with the parking lot management system by the Example of this invention. The operation | movement flowchart of the imaging processing apparatus of the entrance lane in the parking lot management system by the Example of this invention. The processing operation | movement flowchart at the time of the vehicle entrance of the central management PC used with the parking lot management system by the Example of this invention. The operation | movement flowchart of the account machine used with the parking lot management system by the Example of this invention. The operation | movement flowchart of the imaging process seismicization of the entry lane in the parking lot management system by the Example of this invention. The processing operation | movement flowchart at the time of the vehicle participation of the central management PC used with the parking lot management system by the Example of this invention. The loyalty reception process operation | movement flowchart of the imaging processing apparatus of the participation lane of central management PC used with the parking lot management system by the Example of this invention. The figure which shows the content of the parking data in the parking lot management system by the Example of this invention.

10a, 10b Admission lane
11a, 11b loop
12a, 12b, 12c island
13 Light emission trigger sensor
14 Light receiving trigger sensor
15 Ticketing machine
16 Communication line
17 Gate device
18 Gate bar
19 Imaging processing device
20a, 20b participation lane
21 Advance payment machine
22 Checkout machine
23 Communication line
24 Ethernet
30 Administration Center
31 servers
32 Central management PC
33 Database (DB)
33a Contract car DB
33b In-vehicle DB
33c Image DB
34 CPU
35 memory
36 External input section
37 External Communication Department
38 LAN board
39 Camera
40 catcher board
41 flash
42 interface
43 Bus
50a, 50b Vehicle

Claims (2)

There are entrance lanes and entry lanes in the parking lot,
The entrance lane is provided with a ticketing machine, a gate device connected to the ticketing machine via a communication line, an imaging processing device, and a trigger sensor,
The entry lane is provided with a payment machine, a gate device connected to the payment machine via a communication line, an imaging processing device, and a trigger sensor.
The imaging processing apparatus includes a camera, an external input unit, an external communication unit, a memory, and a central processing unit,
A central management PC installed in a parking lot management center and connected to the imaging processing device via a LAN and having an OCR process, an OCR result transmission function, an image reception function, a DB search, and a corresponding vehicle presence / absence verification function is provided. When it is determined that there is a vehicle by the trigger sensors installed in the entrance lane and the entrance lane, a trigger signal is input to the external input unit of the imaging processing device and the vehicle is captured by the camera of the imaging processing device. It is a parking lot management system that reads the license plate and manages entry and exit to the parking lot,
Image data of many contracted vehicles and existing license plates stored in the memory of the imaging processing device installed in the entrance lane or entry lane of the parking lot, and the central processing unit of the imaging processing device generates from the captured image Perform pattern matching processing with the image data of the license plate of the approaching vehicle, and in parallel,
Send an image of the vehicle that entered the lanes and the license plate to the central management PC,
The central management PC performs OCR processing on the received image, compares the OCR result with vehicle management data, and returns a result of whether or not the vehicle is present or a contracted vehicle to the imaging processing device. And
In the imaging processing apparatus, when the pattern matching process is a result of no matching and the result of the OCR process is a corresponding vehicle , the OCR result OK is transmitted to the ticket issuing machine or the checkout machine, and the OK signal The parking lot management system according to claim 1, wherein the ticket issuing machine or the checkout machine performs a process of outputting a gate opening signal to the gate device installed in the lane. The position information of the license plate that is included in the image captured by the imaging apparatus installed in the entrance lane or competed lanes of the parking lot, produced by the central processing unit of the imaging apparatus,
Position information of the license plate along with the image of which an image is captured, be transmitted to the central management PC,
The central management PC performs the OCR processing only on an image of an area where the position of the license plate is specified based on the received license plate position information of the captured image. The described parking management system.

Images