CN115641724A - Patrol inspection identification method and system for on-road berth management and computer medium - Google Patents
Patrol inspection identification method and system for on-road berth management and computer medium Download PDFInfo
- Publication number
- CN115641724A CN115641724A CN202211119839.8A CN202211119839A CN115641724A CN 115641724 A CN115641724 A CN 115641724A CN 202211119839 A CN202211119839 A CN 202211119839A CN 115641724 A CN115641724 A CN 115641724A
- Authority
- CN
- China
- Prior art keywords
- license plate
- vehicle
- parking space
- information
- inspection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000007689 inspection Methods 0.000 title claims abstract description 95
- 238000000034 method Methods 0.000 title claims abstract description 26
- 238000004422 calculation algorithm Methods 0.000 claims description 46
- 238000001514 detection method Methods 0.000 claims description 41
- 238000001914 filtration Methods 0.000 claims description 12
- 238000004458 analytical method Methods 0.000 claims description 10
- 230000005540 biological transmission Effects 0.000 claims description 7
- 230000011218 segmentation Effects 0.000 claims description 6
- 238000012706 support-vector machine Methods 0.000 claims description 6
- 238000013135 deep learning Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims description 3
- 238000003384 imaging method Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 3
- 238000012790 confirmation Methods 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000012163 sequencing technique Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Images
Landscapes
- Traffic Control Systems (AREA)
Abstract
The invention provides a patrol inspection identification method, a system and a computer medium for in-road berth management, wherein the method comprises the following steps: the industrial personal computer of the inspection vehicle acquires the positioning information of the inspection vehicle through the integrated navigation equipment and associates the inspection vehicle with the target parking space according to the positioning information; the industrial personal computer simultaneously obtains images shot by the front-looking camera and the license plate camera; judging whether a vehicle is parked in the parking space or not according to an image shot by the front-view camera, identifying the license plate position of the vehicle when the parking space is judged to be parked with the vehicle, generating an identification result, acquiring three-dimensional point cloud picture information generated by the laser radar by the industrial personal computer when the parking space is judged not to be parked with the vehicle, and secondarily judging whether the vehicle is parked in the parking space or not according to the three-dimensional point cloud picture information; the patrol cloud platform receives the identification result and manages the vehicles in the parking places according to the identification result; the invention reduces the complexity of license plate recognition and improves the efficiency of parking management.
Description
Technical Field
The application relates to the technical field of wireless sensor information acquisition, in particular to a patrol inspection identification method, a system and a computer medium for on-road berth management.
Background
At present, with the rapid increase of the quantity of motor vehicles, the demand for parking is increasingly increased. For the management of berths in roads, in order to reduce the time cost of manual participation and improve the efficiency of road surface inspection, a plurality of intelligent road surface inspection vehicles appear in the market. And collecting the parking condition of parking in the road by using the inspection vehicle, and shooting and obtaining evidence for the vehicles at the parking positions.
The method is used for shooting and obtaining evidence of vehicles at the berth and simultaneously acquiring berth environment information and information of the vehicles and license plates thereof at the berth. For the acquisition of the panoramic information of the parking environment, a wide view field range is required, and a camera with a small focal length is required for shooting and evidence obtaining. However, for a camera with a small focal length, the display of the license plate information shot by the camera is small in the image, and the display is often unclear, so that the license plate recognition algorithm is inaccurate in recognition, and needs to be corrected by background customer service personnel or needs to be subjected to supplementary shooting by road surface workers for evidence collection. This adds virtually to the cost of human involvement.
The parking evidence obtaining and the license plate recognition in the prior art are completed by adopting images shot by the same camera, and the problems that: the scene picture shot by the evidence obtaining image is wide, so that not only the complete vehicle and the surrounding environment of the parking space where the vehicle is located are required to be shot, but also the parking space number is required to be clearly seen in the picture; meanwhile, the traditional detection technology is easy to carry out false detection on vehicles in the left side of the parking space and vehicles on right side of the parking space, so that false charging is caused.
Disclosure of Invention
The application mainly aims to provide a patrol inspection identification method, a patrol inspection identification system and a computer medium for on-road berth management, which are used for reducing the complexity of license plate identification, improving the speed and the precision of license plate identification and improving the efficiency of berth management.
In order to solve the technical problem, the following technical scheme is adopted:
in a first aspect, the application provides an inspection identification system for in-road berth management, which comprises an inspection vehicle and an inspection cloud platform; the inspection vehicle includes:
the camera acquisition module is used for acquiring parking space images and license plate images;
the combined navigation module is used for acquiring positioning information of the inspection vehicle through combined navigation equipment and sending the positioning information to the industrial personal computer;
the laser radar module is used for acquiring three-dimensional point cloud picture information of the vehicle and assisting in judging whether the vehicle is parked at the parking position;
the industrial personal computer is used for judging whether a vehicle is parked in the parking space according to the parking space image shot by the front-view camera, identifying the license plate position of the vehicle when the parking space image shot by the front-view camera is judged to be parked in the target parking space, generating an identification result aiming at the target parking space, and sending the identification result to the patrol cloud platform; when the parking space is judged to be not parked, the industrial personal computer acquires three-dimensional point cloud picture information generated by the laser radar, and secondarily judges whether the parking space is parked with the vehicle or not according to the three-dimensional point cloud picture information;
the POE switch is used for respectively transmitting the parking space images and the license plate images acquired by the camera acquisition module to the industrial personal computer and the ARM calculation unit;
the ARM computing unit is provided with a GPU coprocessor and is used for convolution operation of a laser radar three-dimensional target detection algorithm and a license plate recognition algorithm;
the 4G router is used for being connected with the industrial personal computer, receiving the identification result sent by the industrial personal computer and transmitting the identification result to the routing inspection cloud platform;
and the inspection platform is used for receiving the identification result and managing the vehicles in the parking spaces according to the identification result. Optionally, the camera capturing module includes: the system comprises a front-view camera, a middle-view camera, a license plate camera and a rear-view camera; the camera is configured in the same local area network, and data are forwarded to the ARM computing unit and the industrial personal computer through the switch.
Optionally, different transmission network segments are set for external data transmission of the camera acquisition module, the laser radar module and the combined navigation module, so that data among the modules are isolated.
Optionally, a network segment where external data transmission of the camera acquisition module is located is 192.168.3.X; the laser radar module transmits external data in a network segment of 192.168.1.X; the combined navigation module transmits the external data in the network segment 192.168.2.X.
Optionally, the industrial computer acquires three-dimensional point cloud picture information for whether stop the vehicle on the parking stall and carry out the secondary and judge specifically including: starting a point cloud analysis algorithm to obtain a current latest point cloud frame; performing point cloud filtering according to a preset range, only retaining point cloud data of a target area, filtering out ground points by using laser radar height information and a plane estimation algorithm, performing target segmentation according to point cloud clustering, and classifying by using a support vector machine model; judging whether a current parking space area is occupied by a vehicle point cloud target or not according to the vehicle and the position of the parking space collected in advance, if the point cloud target is classified as a vehicle, judging that the vehicle is in the parking space, and if the point cloud target is not in the parking space, judging that the vehicle is not in the parking space.
Optionally, the recognizing the license plate according to the shot image, and the generating a recognition result for the target parking space specifically includes: the industrial personal computer takes out the license plate image information at the same moment after arriving at the parking space from the license plate camera image queue; and analyzing whether a license plate exists in the corresponding image sensing area or not, and identifying the license plate number if the license plate exists.
Optionally, the industrial personal computer identifies the position of a license plate of the vehicle through an identification algorithm, judges whether shielding exists or not, and identifies license plate information if the position of the license plate is identified and shielding does not exist; the industrial personal computer calls the image information collected by the three recent license plate cameras, synchronously identifies the license plate information of the images, and selects one image with the highest license plate information identification confidence coefficient as a license plate identification result.
In a second aspect, the application provides a patrol inspection identification method for in-road berth management, which is applied to patrol inspection vehicles, and the method comprises the following steps:
101, an industrial personal computer of the inspection vehicle acquires positioning information of the inspection vehicle through combined navigation equipment, and associates the inspection vehicle with a target parking space according to the positioning information;
102, when the inspection vehicle runs to a target parking space, the industrial personal computer simultaneously obtains images shot by the front-looking camera and the license plate camera;
103, judging whether a vehicle is parked in the parking space according to the image shot by the front-view camera, identifying the license plate position of the vehicle when the vehicle is parked in the target parking space, generating an identification result aiming at the target parking space, and sending the identification result to the patrol cloud platform; when the parking of the vehicle is judged to be stopped, the industrial personal computer acquires three-dimensional point cloud picture information generated by the laser radar, and secondary judgment is carried out on whether the vehicle is stopped at the parking space or not according to the three-dimensional point cloud picture information;
and 104, the routing inspection cloud platform receives the identification result and manages the vehicles in the parking spaces according to the identification result.
Optionally, pre-calibrating parking position calibration information, wherein the parking position calibration information comprises longitude and latitude and routing inspection course angle information; in the process of inspecting the inspection vehicle, the industrial personal computer receives the positioning information output by the combined navigation equipment in real time, compares the positioning information with the pre-calibrated parking space calibration information, judges whether the inspection vehicle enters a designated parking space acquisition area or not, and establishes a one-to-one correspondence relationship between the shot evidence-obtaining picture and the parking space so as to complete the association between the inspection vehicle and the target parking space.
Optionally, the industrial personal computer judges that the inspection vehicle enters a designated parking position acquisition area, and simultaneously acquires images shot by the front-view camera and the license plate camera; the image shot by the front-view camera is the latest front-view image taken out from the front-view camera image queue, namely the parking space image information at the same moment after the parking space is reached; and analyzing the parking space state according to the parking space image shot by the front-view camera, wherein the analysis comprises the step of judging whether the vehicle stops by using a deep learning target classification recognition algorithm.
Optionally, if the target parking space is judged to be parked with a vehicle, the industrial personal computer identifies the license plate position of the vehicle through an identification algorithm and judges whether shielding exists or not, and if the license plate position is identified and no shielding exists, the industrial personal computer identifies license plate information; and meanwhile, the industrial personal computer calls the image information collected by three recent license plate cameras, synchronously carries out license plate information recognition on the images, selects one image with the highest confidence coefficient of a license plate information recognition algorithm as a license plate recognition result of the image shot by the license plate camera, carries out similarity comparison on the license plate recognition information of the images shot by the front-view camera and the license plate camera, if the comparison result is similar, the license plate recognition information of the image shot by the license plate camera is used as the license plate recognition result of the parking vehicle in the parking lot, and if not, the license plate recognition information of the image shot by the front-view camera is used as the license plate recognition result of the parking vehicle in the parking lot.
Optionally, if the position of the license plate cannot be identified or the license plate is judged to be blocked, the industrial personal computer controls the middle camera and the rear-view camera to perform supplementary shooting on the vehicle in the parking space along with the forward of the inspection vehicle and positions the position of the license plate; and if the position of the license plate can be positioned, identifying the license plate, and taking the license plate information with higher confidence coefficient of the identification algorithm as an identification result.
Optionally, when judging that there is not the vehicle that stops, the industrial computer obtains three-dimensional some cloud picture information for whether stop the vehicle on the parking stall and carry out the secondary and judge and specifically include: starting a point cloud analysis algorithm to obtain a current latest point cloud frame; performing point cloud filtering according to a preset range, only retaining point cloud data of a target area, filtering out ground points by using laser radar height information and a plane estimation algorithm, performing target segmentation according to point cloud clustering, and classifying by using a support vector machine model; judging whether a current parking space area is occupied by a vehicle point cloud target or not according to the vehicle and the position of the parking space collected in advance, if the point cloud target is classified as a vehicle, judging that the vehicle is in the parking space, and if the point cloud target is not in the parking space, judging that the vehicle is not in the parking space.
In a third aspect, the present application provides a computer medium, on which a computer program is stored, where the computer program, when executed by a processor, implements a method for routing inspection of in-road berth management according to the foregoing embodiments
Drawings
Fig. 1 is a schematic diagram of an architecture of an inspection system for in-road berth management according to an embodiment of the present invention;
fig. 2 is a schematic flowchart of a polling method for in-road berth management according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without inventive step based on the embodiments of the present invention, are within the scope of protection of the present invention.
The embodiment of the invention provides an inspection identification system for in-road berth management, which comprises an inspection vehicle and an inspection cloud platform, wherein the inspection vehicle is connected with the inspection cloud platform through a network; the inspection vehicle includes:
the camera acquisition module is used for acquiring parking space images and license plate images;
optionally, the camera capturing module includes: the system comprises a front-view camera, a middle-view camera, a license plate camera and a rear-view camera; the cameras are configured in the same local area network, and the data are forwarded to the ARM computing unit and the industrial personal computer through the switch.
Optionally, the front-view camera adopts a focal length of 4mm, the license plate camera adopts a focal length of 12mm, the middle-view camera adopts a focal length of 2.8mm, and the rear-view camera is also 4mm as the front-view camera; the license plate adopts a large-focus camera, so that the license plate of the vehicle can be amplified in the imaging of the picture.
Optionally, the camera installation principle is that the parking information to be collected can be shot along the vehicle routing inspection advancing direction, the shooting directions of the front panoramic camera and the license plate camera are consistent, and the license plate camera can shoot the complete license plate information of the vehicle parked at the parking position.
Optionally, the camera acquisition module is connected to the POE switch through a 192.168.3.X network segment, and transmits the shot image to the POE switch in an RTSP multicast manner.
Optionally, the forward-looking camera uses a camera with a smaller focal length to capture panoramic information of the environment including the parked vehicle for forensics.
The combined navigation module is used for acquiring positioning information of the inspection vehicle through combined navigation equipment and sending the positioning information to the industrial personal computer;
optionally, the combined navigation module outputs the positioning information of the inspection vehicle to an industrial personal computer through a network port at a frequency of 100HZ; and the combined navigation module is connected to a specified network port of the industrial personal computer through a 192.168.2.X network segment.
The laser radar module is used for acquiring three-dimensional point cloud picture information of the vehicle and assisting in judging whether the vehicle is parked at the parking position;
optionally, the laser radar module sends the three-dimensional point cloud image information to a specified port of an ARM computing unit through a 192.168.1.X network segment.
The industrial personal computer is used for judging whether a vehicle is parked in the parking space according to the parking space image shot by the front-view camera, identifying the license plate position of the vehicle when the parking space image shot by the front-view camera is judged to be parked in the target parking space, generating an identification result aiming at the target parking space, and sending the identification result to the patrol cloud platform; and when the parking of the vehicle is judged not to exist, the industrial personal computer acquires three-dimensional point cloud picture information generated by the laser radar, and secondary judgment is carried out on whether the vehicle is parked on the parking space or not according to the three-dimensional point cloud picture information.
Optionally, pre-calibrating parking position calibration information, wherein the parking position calibration information comprises longitude and latitude and routing inspection course angle information; in the process of inspecting the inspection vehicle, the industrial personal computer receives the positioning information output by the combined navigation module in real time, compares the positioning information with the pre-calibrated berth calibration information, judges whether the inspection vehicle enters a designated berth acquisition area or not, and establishes a one-to-one correspondence relationship between the shot evidence-obtaining picture and the berth.
Optionally, the industrial personal computer judges that the inspection vehicle enters a designated parking acquisition area, and simultaneously acquires images shot by the front-view camera and the license plate camera; the image shot by the front-view camera is the latest front-view image taken out from the front-view camera image queue, namely the parking space image information at the same moment after the parking space is reached; and analyzing the parking space state according to the parking space image shot by the front-view camera, wherein the analysis comprises the step of judging whether the vehicle stops by using a deep learning target classification recognition algorithm.
Optionally, if the target parking space is judged to be parked with a vehicle, the industrial personal computer identifies the license plate position of the vehicle through an identification algorithm and judges whether shielding exists or not, and if the license plate position is identified and no shielding exists, the industrial personal computer identifies license plate information; and meanwhile, the industrial personal computer calls the image information collected by three recent license plate cameras, synchronously identifies the license plate information of the images, selects one image with the highest confidence coefficient of the license plate information identification algorithm as the license plate identification result of the image shot by the license plate camera, compares the similarity of the license plate identification information of the images shot by the front-view camera and the license plate camera, if the comparison result is similar, takes the license plate identification information of the image shot by the license plate camera as the license plate identification result of the parking vehicle at the parking lot, and if not, takes the license plate identification information of the image shot by the front-view camera as the license plate identification result of the parking vehicle at the parking lot.
Optionally, if the position of the license plate cannot be identified or the license plate is judged to be blocked, the industrial personal computer controls the middle camera and the rear-view camera to perform additional shooting on the vehicle in the parking space along with the advancing of the inspection vehicle, and positions the position of the license plate; and if the position of the license plate can be positioned, identifying the license plate, and taking the license plate information with higher confidence coefficient of the identification algorithm as an identification result.
Optionally, the industrial computer if judge that the parking stall has stopped the vehicle, will discern the vehicle and classify, categorised specifically including: the recognized vehicles are classified into cars, trucks and other vehicles with license plate information for parking management charging, and vehicles which cannot be subjected to charging management, such as tricycles, bicycles, trolleys and the like, are filtered.
Optionally, before analyzing the parking space state according to the parking space image shot by the front-view camera, performing association between the front-view camera and the license plate camera, where the association specifically includes: the similarity of the license plate image shot by the front-view camera and the license plate image shot by the license plate camera is compared, the threshold value of the similarity is set to be 60% in the example, namely, each character in the two license plate images is compared, if the similarity exceeds 60%, the correlation of parking vehicle information of parking positions shot by the two cameras is completed, and finally, the license plate identification information of the image shot by the license plate camera is taken as a license plate identification result.
Optionally, in the comparison of the license plate recognition information of the pictures shot by the front-view camera and the license plate camera, if the license plate recognition algorithm recognizes different license plate lengths, the position of each character in the license plate recognition result of the license plate camera can be compared with the corresponding position of the front-view camera and the characters in the front and rear positions of the front-view camera, and if one position is consistent, the recognized characters are determined to be the same.
Optionally, if the comparison similarity of the license plate identification information of the pictures shot by the front-looking camera and the license plate camera is lower than a set threshold, the vehicle color and the license plate color information of the pictures shot by the front-looking camera and the license plate camera can be further identified and compared, if the two images are consistent, the correlation of the parking vehicle information of the parking spaces shot by the two cameras is completed, and the license plate identification information of the pictures shot by the license plate camera is selected as an identification result.
Optionally, under the condition that the association fails, images shot by three license plate cameras can be collected again for license plate recognition, and similarity comparison is performed between the images and the front panoramic camera, so that the association success rate is improved.
Optionally, when judging that there is not the vehicle that stops, the industrial computer obtains three-dimensional some cloud picture information for whether stop the vehicle on the parking stall and carry out the secondary and judge and specifically include: starting a point cloud analysis algorithm to obtain a current latest point cloud frame; performing point cloud filtering according to a preset range, only retaining point cloud data of a target area, filtering out ground points by using laser radar height information and a plane estimation algorithm, performing target segmentation according to point cloud clustering, and classifying by using a support vector machine model; and judging whether a vehicle point cloud target occupies a current parking space area or not according to the vehicle and the pre-collected parking space position, if the vehicle point cloud target is classified as the vehicle, judging that the vehicle exists in the parking space, and if the vehicle point cloud target does not exist, judging that the vehicle does not exist in the parking space.
Optionally, a limiting condition is preset, and whether a vehicle is in a parking space or not and license plate information is acquired are judged for multiple times according to the limiting condition; the limiting conditions comprise whether a detection area of the parking space is exceeded or not, and the size of the detection area is set and adjusted according to the range from the detection starting position to the detection ending position, the length of the parking space and the running speed of the vehicle. And integrating the detection information of multiple rounds, and sequencing the detection results according to the confidence degrees.
Optionally, the industrial computer gathers the parking stall state, the license plate number and the corresponding confidence degree of the current parking stall of many times detection acquisition to carry out the check-up to it, the check-up specifically includes: and acquiring the summarized detection results of multiple rounds and the confidence degrees of the detection results, and starting the verification of the vehicle type when the confidence degree of the detection results is 0, wherein the fact that the vehicle is possibly not a normal passenger vehicle due to the fact that the license plate is not detected or recognized cannot charge, such as a tricycle for delivering express. When the confidence of the detection result is lower than 90, the reliability of the license plate recognition result is insufficient, secondary confirmation needs to be carried out by using different recognition algorithms, and if the secondary recognition result is consistent with the given result, the confidence of the first recognition of the modified license plate is not high, but the correct probability is higher, and the confidence of the license plate recognition is corrected. If the confidence coefficient is lower than a certain threshold value, parking space line detection is needed, whether false detection of vehicles on curbs is possible due to the fact that the vehicles are too far away from the parking space line is analyzed, and due to the fact that the imaging range of the license plate is small and the imaging effect of the license plate is poor in the general situation. The position relation between the vehicle detection area and the vehicle location line needs to be judged, if the two are not intersected, the result is false detection, and correction is needed.
Optionally, recognizing the license plate according to the shot image, and generating a recognition result for the target parking space specifically includes: the industrial personal computer encapsulates the corrected recognition result, sends the recognition result to the 4G router, and sends the detection result to the inspection cloud platform through the 4G router; the recognition result comprises license plate recognition information and evidence obtaining pictures shot by the cameras.
Optionally, the industrial personal computer is specifically designed by adopting an X86 platform, and at least has 3 network ports and 3 network segments, wherein the 192.168.0.X network segment is an external network segment connected with the 4G router, and uploads data to the patrol cloud platform; 192.168.2.X network segment is a network port connected with the integrated navigation module, and the data transmission frequency is 100HZ;192.168.3.X network segment is the net gape of being connected with POE switch, obtains the data of look ahead, center, back vision camera, and the industrial computer links to each other with ARM computational element simultaneously.
And the POE switch is used for forwarding the parking space images and the license plate images acquired by the camera acquisition module to the industrial personal computer and the ARM computing unit.
Specifically, the camera acquisition module forwards acquired image data of the front-view camera, the middle-view camera and the rear-view camera to the POE switch, and the POE switch sends the acquired image data to a specified network port of the industrial personal computer through a 192.168.3.X network segment; and the license plate image collected by the license plate camera is forwarded to the ARM computing unit through the POE switch.
The ARM computing unit is provided with a GPU coprocessor and is used for convolution operation of a laser radar three-dimensional target detection algorithm and a license plate recognition algorithm;
optionally, the laser radar three-dimensional target detection algorithm and the license plate recognition algorithm both involve a large amount of convolution operations and data parallel processing, so a GPU with a coprocessor is used for calculation.
Optionally, the ARM computing unit has at least 2 network ports, and the network segments in which the ARM computing unit is located are 192.168.3.X and 192.168.1.X, respectively; 192.168.3.X is used for obtaining license plate camera data, and 192.168.1.X is used for obtaining laser radar data.
And the 4G router is used for being connected with the industrial personal computer, receiving the identification result sent by the industrial personal computer and transmitting the identification result to the routing inspection cloud platform.
And the inspection cloud platform is used for receiving the identification result and managing the vehicles in the parking spaces according to the identification result. Optionally, the managing specifically includes: and the patrol cloud platform receives the recognition results containing the license plate recognition information and the evidence obtaining pictures shot by the cameras, and the recognition results are used as the basis for parking management charging and used for managing the vehicles in the parking spaces.
The embodiment of the invention provides a polling identification method for in-road berth management, which is applied to polling cars, wherein the polling cars comprise industrial personal computers; as shown in fig. 2, the method includes:
101, an industrial personal computer of the inspection vehicle acquires positioning information of the inspection vehicle through the integrated navigation equipment, and associates the inspection vehicle with a target parking space according to the positioning information.
Optionally, after acquiring the positioning information of the inspection vehicle, the combined navigation device outputs the positioning information of the inspection vehicle to the industrial personal computer through the network port at the frequency of 100HZ; and the combined navigation module is connected to a specified network port of the industrial personal computer through a 192.168.2.X network segment.
Optionally, the inspection vehicle inspects along the inspection track at a speed of 20-25 kilometers per hour, a combined navigation device can be equipped on the inspection vehicle, and berth calibration information is calibrated in advance, wherein the berth calibration information comprises longitude and latitude and inspection course angle information; in the process of inspecting the inspection vehicle, the industrial personal computer receives the positioning information output by the combined navigation module in real time, compares the positioning information with the pre-calibrated parking space calibration information, judges whether the inspection vehicle enters a designated parking space acquisition area, and establishes a one-to-one correspondence relationship between the shot evidence-obtaining picture and the parking space so as to complete the association between the inspection vehicle and the target parking space.
And step 102, when the patrol vehicle runs to the target parking space, the industrial personal computer simultaneously obtains images shot by the front-view camera and the license plate camera.
Optionally, when the inspection vehicle is about to travel to a position with a parking distance of about 2.5 meters to be photographed, the industrial personal computer simultaneously collects images photographed by the front-view camera and the license plate camera.
103, judging whether a vehicle is parked in the parking space according to the image shot by the front-view camera, identifying the license plate position of the vehicle when the vehicle is parked in the target parking space, generating an identification result aiming at the target parking space, and sending the identification result to the patrol cloud platform; and when the parking of the vehicle is judged not to exist, the industrial personal computer acquires three-dimensional point cloud picture information generated by the laser radar, and secondary judgment is carried out on whether the vehicle is parked on the parking space or not according to the three-dimensional point cloud picture information.
Optionally, the industrial personal computer judges that the inspection vehicle enters a designated parking position acquisition area, and simultaneously acquires images shot by the front-view camera and the license plate camera; the image shot by the front-view camera is the latest front-view image taken out from the front-view camera image queue, namely the parking space image information at the same moment after the parking space is reached; and analyzing the parking space state according to the parking space image shot by the front-view camera, wherein the analysis comprises the step of judging whether the vehicle stops by using a deep learning target classification recognition algorithm.
Optionally, if the target parking space is judged to be parked with a vehicle, the industrial personal computer identifies the position of the license plate of the vehicle through a license plate identification algorithm, judges whether shielding exists, and if the position of the license plate is identified and no shielding exists, the industrial personal computer identifies license plate information; and meanwhile, the industrial personal computer calls the image information collected by three recent license plate cameras, synchronously identifies the license plate information of the images, selects one image with the highest confidence coefficient of the license plate information identification algorithm as the license plate identification result of the image shot by the license plate camera, compares the similarity of the license plate identification information of the images shot by the front-view camera and the license plate camera, if the comparison result is similar, takes the license plate identification information of the image shot by the license plate camera as the license plate identification result of the parking vehicle at the parking lot, and if not, takes the license plate identification information of the image shot by the front-view camera as the license plate identification result of the parking vehicle at the parking lot.
Optionally, if the position of the license plate cannot be identified or the license plate is judged to be blocked, the industrial personal computer controls the middle camera and the rear-view camera to perform supplementary shooting on the vehicle in the parking space along with the forward of the inspection vehicle and positions the position of the license plate; and if the position of the license plate can be positioned, identifying the license plate, and taking the license plate information with higher confidence coefficient of the identification algorithm as an identification result.
Optionally, the industrial computer if judge that the parking stall has stopped the vehicle, will discern the vehicle and classify, categorised specifically including: the recognized vehicles are classified into vehicles such as cars and vans containing license plate information for parking management charging, and vehicles such as tricycles, bicycles and trolleys which cannot be subjected to charging management are filtered.
Optionally, before analyzing the parking space state according to the parking space image shot by the front-view camera, performing association between the front-view camera and the license plate camera, where the association specifically includes: the similarity comparison is carried out on the license plate image shot by the front-view camera and the license plate image shot by the license plate camera, the threshold value of the similarity is set to be 60% in the example, namely, each character in the two license plate images is compared, if the similarity exceeds 60%, the correlation of parking space parking vehicle information shot by the two cameras is completed, and finally the license plate identification information of the image shot by the license plate camera is taken as a license plate identification result.
Optionally, in the comparison of the license plate recognition information of the pictures shot by the front-view camera and the license plate camera, if the license plate recognition algorithm recognizes different license plate lengths, the position of each character in the license plate recognition result of the license plate camera can be compared with the corresponding position of the front-view camera and the characters in the front and rear positions of the front-view camera, and if one position is consistent, the recognized characters are determined to be the same.
Optionally, if the comparison similarity of the license plate identification information of the pictures shot by the front-looking camera and the license plate camera is lower than a set threshold, the vehicle color and the license plate color information of the pictures shot by the front-looking camera and the license plate camera can be further identified and compared, if the two images are consistent, the correlation of the parking vehicle information of the parking spaces shot by the two cameras is completed, and the license plate identification information of the pictures shot by the license plate camera is selected as an identification result.
Optionally, under the condition that the association fails, images shot by three license plate cameras can be collected again for license plate recognition, and similarity comparison is carried out between the images and the front panoramic camera, so that the association success rate is improved.
Optionally, when judging that there is not the vehicle that stops, the industrial computer obtains three-dimensional some cloud picture information for whether stop the vehicle on the parking stall and carry out the secondary and judge and specifically include: starting a point cloud analysis algorithm to obtain a current latest point cloud frame; performing point cloud filtering according to a preset range, only keeping point cloud data of a target area, filtering ground points by using laser radar height information and a plane estimation algorithm, performing target segmentation according to point cloud clustering, and classifying by using a support vector machine model; and judging whether a vehicle point cloud target occupies a current parking space area or not according to the vehicle and the pre-collected parking space position, if the vehicle point cloud target is classified as the vehicle, judging that the vehicle exists in the parking space, and if the vehicle point cloud target does not exist, judging that the vehicle does not exist in the parking space.
Optionally, when the secondary judgment determines that the vehicle is parked in the target parking space, the industrial personal computer continuously judges whether the vehicle is parked in the parking space or not according to the image shot by the front-view camera and carries out license plate position recognition on the vehicle.
Optionally, a limiting condition is preset, and whether a vehicle is in a parking space or not and license plate information is acquired are judged for multiple times according to the limiting condition; the limiting conditions comprise whether a detection area of the parking space is exceeded or not, and the size of the detection area is set and adjusted according to the range from the detection starting position to the detection ending position, the length of the parking space and the running speed of the vehicle. And integrating the detection information of multiple rounds, and sequencing the detection results according to the confidence degrees.
Optionally, the industrial computer gathers the parking stall state, the license plate number and the corresponding confidence degree of the current parking stall of many times detection acquisition to carry out the check-up to it, the check-up specifically includes: and acquiring the gathered detection results of multiple rounds and the confidence degrees of the detection results, and starting the verification of the vehicle type when the confidence degree of the detection results is 0, wherein the vehicle is possibly not a normal passenger vehicle and cannot be charged because the license plate is not detected or identified, such as a tricycle for delivering express. When the confidence of the detection result is lower than 90, the reliability of the license plate recognition result is insufficient, secondary confirmation needs to be carried out by using different recognition algorithms, and if the secondary recognition result is consistent with the given result, the confidence of the first recognition of the modified license plate is not high, but the correct probability is higher, and the confidence of the license plate recognition is corrected. If the confidence coefficient is lower than a certain threshold value, parking space line detection is needed, and whether false detection of vehicles on curbs is possible due to the fact that the vehicles are too far away from a parking space line is analyzed, because the imaging range of the license plate is small and the imaging effect of the license plate is poor under the condition. The position relation between the vehicle detection area and the vehicle location line needs to be judged, if the two are not intersected, the result is false detection, and correction is needed.
Optionally, after the inspection vehicle is associated with the target parking space, sending indication information containing the number of the target parking space to an industrial personal computer, a laser radar module and an ARM (advanced RISC machines) computing unit, wherein the industrial personal computer receives the indication information and is used for analyzing the parking space state of the target parking space and judging whether the target parking space has a vehicle or not; after receiving the indication information, the laser radar module generates three-dimensional point cloud picture information aiming at the target parking space; after the ARM computing unit receives the indication information, convolution operation is carried out on the laser radar three-dimensional target detection algorithm and the license plate recognition algorithm to judge whether a vehicle exists in the target parking space and recognize the license plate of the vehicle. The industrial personal computer collects the detection results of the multiple detections, and sorts the results of the multiple detections according to the confidence degree.
And 104, the routing inspection cloud platform receives the identification result and manages the vehicles in the parking spaces according to the identification result.
Optionally, the managing specifically includes: and the patrol cloud platform receives the recognition result containing the license plate recognition information and the evidence obtaining picture shot by each camera, and the recognition result is used as the basis for parking management charging and is used for managing the vehicles in the parking spaces.
The invention provides a scheme of adopting two cameras, wherein the image of one camera is used for comprehensive evidence obtaining, and the image of one camera is specially used for license plate recognition. The results of the two cameras are synchronized through a fusion algorithm, and then are calibrated and filtered through a result checking module. Finally, the evidence obtaining image and the license plate recognition image are in one-to-one correspondence, so that the effect of the evidence obtaining image is guaranteed, the accuracy of license plate recognition is guaranteed, and the manual intervention rate is reduced. In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be physically included alone, or two or more units may be integrated into one unit. The integrated unit may be implemented in the form of hardware, or in the form of hardware plus a software functional unit.
The integrated unit implemented in the form of a software functional unit may be stored in a computer readable storage medium. The software functional unit is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute some steps of the transceiving method according to various embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a portable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other media capable of storing program codes.
While the preferred embodiments of the present invention have been described, it should be understood that modifications and adaptations to those embodiments may occur to one skilled in the art without departing from the principles of the present invention and are within the scope of the present invention.
Claims (14)
1. The utility model provides an in-road berth management's identification system that patrols and examines which characterized in that: the system comprises an inspection vehicle and an inspection cloud platform; the inspection vehicle includes:
the camera acquisition module is used for acquiring parking space images and license plate images;
the combined navigation module is used for acquiring positioning information of the inspection vehicle through combined navigation equipment and sending the positioning information to the industrial personal computer;
the laser radar module is used for acquiring three-dimensional point cloud picture information of the vehicle and assisting in judging whether the vehicle is parked at the parking position;
the industrial personal computer is used for judging whether a vehicle stops in the parking space according to the parking space image shot by the front-view camera, identifying the position of a license plate of the vehicle when the vehicle stops in a target parking space, generating an identification result aiming at the target parking space, and sending the identification result to the patrol cloud platform; when the parking of the vehicle is judged to be stopped, the industrial personal computer acquires three-dimensional point cloud picture information generated by the laser radar, and secondary judgment is carried out on whether the vehicle is stopped at the parking space or not according to the three-dimensional point cloud picture information;
the POE switch is used for respectively transmitting the parking space images and the license plate images acquired by the camera acquisition module to the industrial personal computer and the ARM calculation unit;
the ARM computing unit is provided with a GPU coprocessor and is used for convolution operation of a laser radar three-dimensional target detection algorithm and a license plate recognition algorithm;
the 4G router is used for being connected with the industrial personal computer, receiving the identification result sent by the industrial personal computer and transmitting the identification result to the inspection cloud platform;
and the inspection platform is used for receiving the identification result and managing the vehicles in the parking spaces according to the identification result.
2. The in-road berth management patrol inspection identification system according to claim 1, characterized in that: the camera acquisition module includes: the system comprises a front-view camera, a middle-view camera, a license plate camera and a rear-view camera; the cameras are configured in the same local area network, and the data are forwarded to the ARM computing unit and the industrial personal computer through the switch.
3. The in-circuit berth management inspection recognition system according to claim 2, characterized in that: different transmission network segments are set for external data transmission of the camera acquisition module, the laser radar module and the combined navigation module, so that data among the modules are isolated.
4. The in-circuit berth management inspection recognition system according to claim 3, characterized in that: the network segment where the external data transmission of the camera acquisition module is located is 192.168.3.X; the network segment 192.168.1.X where the laser radar module transmits external data; the combined navigation module transmits the external data in the network segment 192.168.2.X.
5. The in-road berth management patrol inspection identification system according to claim 1, characterized in that: when judging not stopping there is the vehicle, the industrial computer acquires three-dimensional some cloud picture information for whether stop there being the vehicle in the parking stall to carry out the secondary and judge, specifically include: starting a point cloud analysis algorithm to obtain a current latest point cloud frame; performing point cloud filtering according to a preset range, only keeping point cloud data of a target area, filtering ground points by using laser radar height information and a plane estimation algorithm, performing target segmentation according to point cloud clustering, and classifying by using a support vector machine model; judging whether a current parking space area is occupied by a vehicle point cloud target or not according to the vehicle and the position of the parking space collected in advance, if the point cloud target is classified as a vehicle, the parking space is judged to be occupied, and if the point cloud target is not existed, the parking space is judged to be not occupied.
6. The in-circuit berth management inspection recognition system according to claim 1, characterized in that: the recognizing the license plate according to the shot image and the generating a recognition result aiming at the target parking space specifically comprise: the industrial personal computer takes out the license plate image information at the same moment after reaching the parking space from the license plate camera image queue; and analyzing whether a license plate exists in the corresponding image sensing area or not, and identifying the license plate number if the license plate exists.
7. The in-circuit berth management inspection recognition system according to claim 6, characterized in that: the industrial personal computer identifies the position of a license plate of a vehicle through an identification algorithm, judges whether shielding exists or not, and identifies license plate information if the position of the license plate is identified and shielding does not exist; the industrial personal computer calls the image information collected by the three recent license plate cameras, synchronously identifies the license plate information of the images, and selects one image with the highest license plate information identification confidence coefficient as a license plate identification result.
8. A patrol inspection identification method for on-road berth management is characterized in that: the method is applied to the inspection vehicle and comprises the following steps:
101, an industrial personal computer of the inspection vehicle acquires positioning information of the inspection vehicle through integrated navigation equipment, and associates the inspection vehicle with a target parking space according to the positioning information;
102, when the inspection vehicle runs to a target parking space, the industrial personal computer simultaneously obtains images shot by the front-looking camera and the license plate camera;
103, judging whether a vehicle is parked in the parking space according to the image shot by the front-view camera, identifying the license plate position of the vehicle when the vehicle is parked in the target parking space, generating an identification result aiming at the target parking space, and sending the identification result to the patrol cloud platform; when the parking space is judged to be not parked, the industrial personal computer acquires three-dimensional point cloud picture information generated by the laser radar, and secondarily judges whether the parking space is parked with the vehicle or not according to the three-dimensional point cloud picture information;
and 104, receiving the identification result by the inspection cloud platform, and managing the vehicles in the parking spaces according to the identification result.
9. The inspection tour identification method for in-road berth management according to claim 8, characterized in that: pre-calibrating berth calibration information, wherein the berth calibration information comprises longitude and latitude and routing inspection course angle information;
in the process of polling the polling car, the industrial personal computer receives the positioning information output by the combined navigation equipment in real time, compares the positioning information with the pre-calibrated parking position calibration information, judges whether the polling car enters a designated parking position acquisition area, and establishes the one-to-one correspondence between the shot evidence-obtaining picture and the parking position so as to complete the association between the polling car and the target parking position.
10. The inspection tour identification method for in-road berth management according to claim 9, characterized in that: the industrial personal computer judges that the inspection vehicle enters a designated parking acquisition area, and simultaneously acquires images shot by the forward-looking camera and the license plate camera; the image shot by the front-view camera is the latest front-view image taken out from the front-view camera image queue, namely the parking space image information at the same moment after the parking space is reached; and are
And analyzing the parking space state according to the parking space image shot by the front-view camera, wherein the analysis comprises the step of judging whether a vehicle stops by using a deep learning target classification recognition algorithm.
11. The inspection tour identification method for in-road berth management of claim 10, comprising the following steps: if the vehicle is judged to be parked in the target parking space, the industrial personal computer identifies the position of the license plate of the vehicle through an identification algorithm and judges whether shielding exists or not, and if the position of the license plate is identified and no shielding exists, the industrial personal computer identifies license plate information; and meanwhile, the industrial personal computer calls the image information collected by three recent license plate cameras, synchronously carries out license plate information recognition on the images, selects one image with the highest confidence coefficient of a license plate information recognition algorithm as a license plate recognition result of the image shot by the license plate camera, carries out similarity comparison on the license plate recognition information of the images shot by the front-view camera and the license plate camera, if the comparison result is similar, the license plate recognition information of the image shot by the license plate camera is used as the license plate recognition result of the parking vehicle in the parking lot, and if not, the license plate recognition information of the image shot by the front-view camera is used as the license plate recognition result of the parking vehicle in the parking lot.
12. The inspection tour identification method of in-road berth management according to claim 11, characterized in that: if the position of the license plate cannot be identified or the license plate is judged to be blocked, the industrial personal computer controls the middle camera and the rear-view camera to carry out supplementary shooting on the vehicle in the parking space along with the advancing of the inspection vehicle and positions the position of the license plate; and if the position of the license plate can be positioned, license plate recognition is carried out, and the license plate information recognition algorithm with higher confidence coefficient is taken as a recognition result.
13. The inspection tour identification method for in-road berth management of claim 8, characterized in that: when judging that there is not the vehicle that stops, the industrial computer obtains three-dimensional some cloud picture information for whether stop the vehicle on the parking stall and carry out the secondary and judge and specifically include: starting a point cloud analysis algorithm to obtain a current latest point cloud frame; performing point cloud filtering according to a preset range, only retaining point cloud data of a target area, filtering out ground points by using laser radar height information and a plane estimation algorithm, performing target segmentation according to point cloud clustering, and classifying by using a support vector machine model; and judging whether a vehicle point cloud target occupies a current parking space area or not according to the vehicle and the pre-collected parking space position, if the vehicle point cloud target is classified as the vehicle, judging that the vehicle exists in the parking space, and if the vehicle point cloud target does not exist, judging that the vehicle does not exist in the parking space.
14. A computer medium having stored thereon a computer program which, when executed by a processor, implements a patrol identification method for in-road berth management according to any one of claims 8 to 13.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211119839.8A CN115641724B (en) | 2022-09-15 | 2022-09-15 | Inspection identification method, system and computer medium for managing berths in roads |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211119839.8A CN115641724B (en) | 2022-09-15 | 2022-09-15 | Inspection identification method, system and computer medium for managing berths in roads |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115641724A true CN115641724A (en) | 2023-01-24 |
| CN115641724B CN115641724B (en) | 2023-10-27 |
Family
ID=84940987
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211119839.8A Active CN115641724B (en) | 2022-09-15 | 2022-09-15 | Inspection identification method, system and computer medium for managing berths in roads |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115641724B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130147954A1 (en) * | 2011-12-13 | 2013-06-13 | Electronics And Telecommunications Research Institute | Parking lot management system in working cooperation with intelligent cameras |
| CN109767626A (en) * | 2018-12-20 | 2019-05-17 | 北京筑梦园科技有限公司 | A kind of curb parking method for inspecting, system, cruiser and server |
| CN112949530A (en) * | 2021-03-12 | 2021-06-11 | 新疆爱华盈通信息技术有限公司 | Inspection method and system for parking lot inspection vehicle and inspection vehicle |
| CN113436361A (en) * | 2021-06-03 | 2021-09-24 | 超级视线科技有限公司 | Roadside berth management system based on unmanned inspection vehicle |
| CN113542678A (en) * | 2021-06-17 | 2021-10-22 | 超级视线科技有限公司 | Parking management removes system of patrolling and examining |
| CN114446059A (en) * | 2021-12-29 | 2022-05-06 | 北京智联云海科技有限公司 | System and method for vehicle-mounted monitoring of roadside parking vehicles |
-
2022
- 2022-09-15 CN CN202211119839.8A patent/CN115641724B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130147954A1 (en) * | 2011-12-13 | 2013-06-13 | Electronics And Telecommunications Research Institute | Parking lot management system in working cooperation with intelligent cameras |
| CN109767626A (en) * | 2018-12-20 | 2019-05-17 | 北京筑梦园科技有限公司 | A kind of curb parking method for inspecting, system, cruiser and server |
| CN112949530A (en) * | 2021-03-12 | 2021-06-11 | 新疆爱华盈通信息技术有限公司 | Inspection method and system for parking lot inspection vehicle and inspection vehicle |
| CN113436361A (en) * | 2021-06-03 | 2021-09-24 | 超级视线科技有限公司 | Roadside berth management system based on unmanned inspection vehicle |
| CN113542678A (en) * | 2021-06-17 | 2021-10-22 | 超级视线科技有限公司 | Parking management removes system of patrolling and examining |
| CN114446059A (en) * | 2021-12-29 | 2022-05-06 | 北京智联云海科技有限公司 | System and method for vehicle-mounted monitoring of roadside parking vehicles |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115641724B (en) | 2023-10-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110487562B (en) | Driveway keeping capacity detection system and method for unmanned driving | |
| JP7218535B2 (en) | Traffic violation vehicle identification system and server | |
| CN107305627B (en) | Vehicle video monitoring method, server and system | |
| CN102765365B (en) | Pedestrian detection method based on machine vision and pedestrian anti-collision warning system based on machine vision | |
| US7366325B2 (en) | Moving object detection using low illumination depth capable computer vision | |
| KR101647370B1 (en) | road traffic information management system for g using camera and radar | |
| JP5297078B2 (en) | Method for detecting moving object in blind spot of vehicle, and blind spot detection device | |
| KR101891460B1 (en) | Method and apparatus for detecting and assessing road reflections | |
| CN103824037B (en) | Vehicle anti-tracking alarm device | |
| CN110738150B (en) | Camera linkage snapshot method and device and computer storage medium | |
| WO2019223655A1 (en) | Detection of non-motor vehicle carrying passenger | |
| CN111727437A (en) | Multispectral system providing pre-crash warning | |
| CN108966145A (en) | Hit-and-run criminal is tracked using V2X communication | |
| CN111915883A (en) | Road traffic condition detection method based on vehicle-mounted camera shooting | |
| CN111368612A (en) | Overman detection system, personnel detection method and electronic equipment | |
| CN105303826A (en) | Violating side parking evidence obtaining device and method | |
| CN115240148A (en) | Vehicle behavior detection method and device, storage medium and electronic device | |
| CN115909223A (en) | Method and system for matching WIM system information with monitoring video data | |
| CN109874099B (en) | Networking vehicle-mounted equipment flow control system | |
| CN105303827A (en) | Traffic violation image obtaining device and method | |
| CN112435479B (en) | Target object violation detection method and device, computer equipment and system | |
| CN113591643A (en) | Underground vehicle station entering and exiting detection system and method based on computer vision | |
| CN111862621B (en) | Intelligent snapshot system of multi-type adaptive black cigarette vehicle | |
| CN105303825A (en) | Violating inclined side parking evidence obtaining device and method | |
| CN115641724B (en) | Inspection identification method, system and computer medium for managing berths in roads |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |