CN-115690761-B - Method and system for improving license plate recognition rate of original fuel truck

CN115690761BCN 115690761 BCN115690761 BCN 115690761BCN-115690761-B

Abstract

The invention discloses a method and a system for improving the license plate recognition rate of an original fuel truck, which belong to the technical field of intelligent transportation, wherein the system comprises a bayonet unit and a vehicle management module, each bayonet unit comprises a main snapshot machine and 0 to a plurality of auxiliary snapshot machines, each snapshot machine comprises a main recognition algorithm and 0 to a plurality of auxiliary recognition algorithms, the vehicle management module is used for generating a dynamic license plate number library for each type of bayonet unit, the bayonet unit is used for acquiring license plate numbers recognized by the snapshot machines and comparing the license plate numbers with the dynamic license plate number library corresponding to the bayonet units to complete the recognition process, and the vehicle management module is also used for self-learning recognition of recognition results and is used for the next recognition process. The vehicle management module adopts a plurality of means such as multi-snapshot equipment, multi-recognition algorithm, fuzzy judgment of the vehicle management module and the like, so that the recognition rate is improved to the greatest extent, and the manual intervention is reduced.

Inventors

WANG ZHIJUN
ZHAO JING

Assignees

中冶南方工程技术有限公司

Dates

Publication Date: 20260505
Application Date: 20220905

Claims (5)

1. The system for improving the identification rate of the license plate of the original fuel truck is characterized by comprising a bayonet unit and a vehicle management module; The bayonet unit comprises a factory entrance, a sampling bayonet, a weighing bayonet and a stock yard bayonet, wherein the factory entrance represents a bayonet through which all raw fuel trucks must pass, the factory entrance represents a bayonet through which trucks in the factory entrance can pass, the sampling bayonet represents a bayonet through which trucks entering the factory entrance and containing a sampling plan can pass, the weighing bayonet contains a truck of the weighing plan, the sampling plan is completed, or the truck entering the factory entrance and not containing the sampling plan can pass, the stock yard bayonet represents a bayonet through which trucks entering the factory entrance, no sampling plan or completed, and no weighing plan or completed weighing plan can pass, each bayonet unit comprises a main snapshot machine and 0 to a plurality of auxiliary snapshots, and each snapshot machine comprises a main identification algorithm and 0 to a plurality of auxiliary identification algorithms; The vehicle management module is used for generating a dynamic license plate number library for each type of bayonet unit, and the method for establishing the dynamic license plate number library comprises the following steps: the method comprises the steps of performing primary operation, wherein a dynamic license plate number library of each bayonet is empty; according to the corresponding management system of the original fuel purchase contract, acquiring information of the to-be-shipped vehicles, and adding the target license plate number of the to-be-shipped vehicles into a factory parking lot bayonet dynamic license plate number library; after the entrance of the factory parking lot is successfully identified, moving the target license plate number from a dynamic license plate number library of the entrance of the factory parking lot to a dynamic license plate number library of the entrance of the factory parking lot; After the identification of the entrance to the factory is successful, the target license plate number is removed from a dynamic license plate number library of the entrance to the factory, at the moment, if a vehicle corresponding to the target license plate number has a sampling plan, the target license plate number is added into the dynamic license plate number library of the sampling entrance, if the sampling plan is not available, but a weighing plan is available, the target license plate number is added into the dynamic license plate number library of the weighing entrance, and if the sampling plan is not available, the weighing plan is not available, the target license plate number is added into the dynamic license plate number library of the stock yard; after the sampling card is successfully identified, the target license plate number is removed from a sampling card dynamic license plate number library, at the moment, if a vehicle corresponding to the target license plate number has a weighing plan, the target license plate number is added into the weighing card dynamic license plate number library, and if the weighing plan does not exist, the target license plate number is added into the stock yard card dynamic license plate number library; after the weighing bayonet is successfully identified, the target license plate number is moved from a weighing bayonet dynamic license plate number library to a stock yard bayonet dynamic license plate number library; after the stock yard bayonet is successfully identified, removing the target license plate number from a stock yard bayonet dynamic license plate number library; the license plate identification unit is used for acquiring license plate numbers identified by a main identification algorithm of the main snapshot machine, if the license plate numbers are completely matched with a certain license plate in a dynamic license plate number library of a corresponding license plate, completing the identification process, otherwise, acquiring all license plate numbers identified by each identification algorithm of each snapshot machine corresponding to the corresponding license plate, completely matching each license plate number in the license plate number library corresponding to the corresponding license plate, if the successfully matched license plate numbers exist, selecting the successfully matched license plate numbers, and if all the license plate numbers cannot be matched, determining the final license plate numbers by fuzzy matching; the vehicle management module is also used for self-learning the recognition result and is used for the next recognition process.
2. The system of claim 1, wherein the main snapshot machine and the auxiliary snapshot machine are selected from different manufacturers or different models of products from the same manufacturer, the installation angles of the main snapshot machine and the auxiliary snapshot machine are different to improve the recognition range, the main path recognition algorithm of the snapshot machine adopts the manufacturer to carry, and the auxiliary path recognition algorithm adopts the self-defined recognition algorithm based on the deep neural network to reduce the leakage rate.
3. The system of claim 2, wherein the fuzzy matching includes a province fuzzy and a specified fuzzy, the province fuzzy refers to that the province is not considered, only other letters and numbers except the province are compared with the license plate, if all characters except the province are completely consistent, the matching is successful, and the specified fuzzy refers to that a certain character fuzzy is recognized as other characters according to the system preset.
4. A system according to claim 3, wherein the vehicle management module is configured to record a result of fuzzy recognition, record a fuzzy matching rule into the system when the number of fuzzy recognition meets a preset number threshold, and automatically map according to the recorded fuzzy matching rule when the fuzzy recognition is performed next time.
5. A method for improving the license plate recognition rate of an original fuel truck, comprising the following steps: (1) Generating a dynamic license plate number library for each type of bayonet, wherein the bayonet comprises a factory parking lot entering bayonet, a factory entering bayonet, a sampling bayonet, a weighing bayonet and a stock yard bayonet, the factory entering bayonet represents a bayonet through which all raw fuel trucks must pass, the factory entering bayonet represents a bayonet through which trucks in the factory entering bayonet pass, the sampling bayonet represents a bayonet through which trucks entering the factory entering bayonet and containing a sampling plan pass, the weighing bayonet contains a truck of the weighing plan, the sampling plan is completed, or the bayonet through which trucks entering the factory entering bayonet and not containing the sampling plan pass, and the stock yard bayonet represents a bayonet through which trucks entering the factory entering bayonet, no sampling plan or completed, and no weighing plan or completed weighing plan pass; (2) Acquiring license plate numbers identified by the snapshot machine, and comparing the license plate numbers with a dynamic license plate number library corresponding to the bayonets to finish the identification process; (3) Self-learning is carried out on the identification result for the next identification process; The step (1) comprises: (1.1) performing primary operation, wherein a dynamic license plate number library of each bayonet is empty; the method comprises the steps of (1.2) acquiring information of vehicles to be shipped according to a corresponding management system of a raw fuel purchase contract, and adding a target license plate number of the vehicles to be shipped into a factory parking lot bayonet dynamic license plate number library; (1.3) after the entrance of the factory parking lot is successfully identified, moving the target license plate number from the entrance dynamic license plate number library of the factory parking lot to the entrance dynamic license plate number library; (1.4) after the identification of the entrance of the factory is successful, removing the target license plate number from a dynamic license plate number library of the entrance of the factory, judging that if a vehicle corresponding to the target license plate number has a sampling plan, adding the target license plate number into the dynamic license plate number library of the sampling entrance, if the sampling plan is not available, but a weighing plan is available, adding the target license plate number into the dynamic license plate number library of the weighing entrance, and if the sampling plan is not available, and the weighing plan is not available, adding the target license plate number into the dynamic license plate number library of the stock yard entrance; (1.5) after the sampling port identification is successful, removing the target license plate number from a sampling port dynamic license plate number library, judging that if a vehicle corresponding to the target license plate number has a weighing plan, adding the target license plate number into the weighing port dynamic license plate number library, and if the weighing plan is not available, adding the target license plate number into the stock yard port dynamic license plate number library; (1.6) after the weighing bayonet identification is successful, moving the target license plate number from the weighing bayonet dynamic license plate number library to the stock yard bayonet dynamic license plate number library; (1.7) after the stock yard bayonet identification is successful, removing the target license plate number from a stock yard bayonet dynamic license plate number library; the step (2) comprises: (2.1) obtaining a license plate number identified by a main identification algorithm of a main snapshot machine, and if the license plate number is completely matched with a license plate in a dynamic license plate number library of a corresponding bayonet, completing an identification process; And (2.2) if the license plate number is not completely matched with a license plate in the dynamic license plate number library of the corresponding bayonet, acquiring all license plate numbers which are identified by the identification algorithms of the snap-shot machines corresponding to the corresponding bayonet, performing complete matching in the license plate number library corresponding to the corresponding bayonet aiming at each license plate number, if the successfully matched license plate number exists, selecting the successfully matched license plate number, and if all the license plate numbers cannot be matched, determining the final license plate number by fuzzy matching.

Description

Method and system for improving license plate recognition rate of original fuel truck Technical Field The invention belongs to the technical field of intelligent transportation, and particularly relates to a method and a system for improving the identification rate of a primary fuel truck license plate of a steel enterprise. Background Da Zong raw fuels for iron and steel enterprises are typically transported to the stockyard using trains or large trucks. If a truck is used, license plate recognition needs to be carried out on the vehicle, then the barrier gate can be operated, and the logistics information and the state information of the vehicle are managed. Taking da Zong raw fuels of a steel mill as an example, about 500-1000 trucks are transported each day, and 5-10 gates are required to pass each truck on average, which is equivalent to about 5000 identification operations. Because the license plate of the large truck is dirty, and part of road gate road surfaces are uneven, the recognition accuracy is affected, and the current vehicle recognition efficiency is generally about 95%, so that about 250 times of recognition errors occur every day, the vehicle passing efficiency is seriously affected, and the road is blocked. Meanwhile, the original planned unattended barrier gate needs a great deal of manual intervention, and the expected purpose cannot be achieved. Disclosure of Invention Aiming at the defects or improvement demands of the prior art, in order to infinitely increase the identification rate of the original fuel truck license plate of the iron and steel enterprise to 100%, the invention provides a method and a system for increasing the identification rate of the original fuel truck license plate, which adopt a plurality of means such as multi-snapshot equipment, multi-identification algorithm, fuzzy judgment of a vehicle management module and the like, so that the identification rate is increased to the greatest extent, and the manual intervention is reduced. In order to achieve the above object, according to one aspect of the present invention, there is provided a system for improving the license plate recognition rate of an original fuel truck, comprising a bayonet unit and a vehicle management module; each bayonet unit comprises a main snapshot machine and 0 to a plurality of auxiliary snapshot machines, and each snapshot machine comprises a main identification algorithm and 0 to a plurality of auxiliary identification algorithms; the vehicle management module is used for generating a dynamic license plate number library for each type of bayonet unit; the bayonet unit is used for acquiring the license plate number identified by the snapshot machine, comparing the license plate number with a dynamic license plate number library corresponding to the bayonet unit, and completing the identification process; the vehicle management module is also used for self-learning the recognition result and is used for the next recognition process. In some optional embodiments, the main snapshot machine and the auxiliary snapshot machine are selected from different manufacturers or different models of products from the same manufacturer, the installation angles of the main snapshot machine and the auxiliary snapshot machine are different, so that the identification range is improved, the main path identification algorithm of the snapshot machine is self-contained by the manufacturer, and the auxiliary path identification algorithm is self-defined identification algorithm based on a deep neural network, so that the leakage rate is reduced. In some alternative embodiments, the bayonet unit comprises a factory parking lot bayonet, a factory entering bayonet, a sampling bayonet, a weighing bayonet and a stock yard bayonet; the factory parking lot entrance represents an entrance through which all raw fuel trucks must pass, the factory entrance represents an entrance through which trucks in the factory parking lot can pass, the sampling entrance represents an entrance through which trucks including a sampling plan pass, the weighing entrance includes a truck of a weighing plan, the sampling plan is completed, or the entrance through which trucks including no sampling plan pass, and the raw material lot entrance represents an entrance through which trucks including no sampling plan or completed and no weighing plan or completed pass. In some optional embodiments, the method for establishing the dynamic license plate number library comprises the following steps: the method comprises the steps of performing primary operation, wherein a dynamic license plate number library of each bayonet is empty; according to the corresponding management system of the original fuel purchase contract, acquiring information of the to-be-shipped vehicles, and adding the target license plate number of the to-be-shipped vehicles into a factory parking lot bayonet dynamic license plate number library; after the entrance of the factory