Abstract: Vehicle re-identifcation aims to match and identifythe same vehicle crossing multiple surveillance cameras andobtain traffc information such as travel time. The AutomaticLicense Plate Recognition (ALPR) data are widely employed inurban surveillance.However, vehicle re-identifcation based onALPR data is challenging due to license plate recognition errorsand unrecognized issues, This paper proposes a vehicle matchingalgorithm designed to maximize the travel time probabilityusing ALPR data, while accounting for recognition errors andunrecognized issues. The proposed algorithm consists of severalmodules, including the estimation of travel time distribution.computation of travel time probability, calculation of travel timeconfdence intervals and matching time window size, restrictedfuzzy matching, and vehicle matching optimization. To evaluatethe effectiveness of the proposed algorithm across varying lightingand weather conditions, ALPR data was collected from a surveyroad in four scenarios: sunny day, sunny night, rainy day, andrainy night. The results indicate that when compared to a sunnyday scenario, severe lighting and adverse weather conditionslead to decreased matching accuracy and increased matchingaccuracy errors for all methods evaluated. However, our proposedmodel outperforms benchmark algorithms in both scenarios.demonstrating its superior performance.
IndexTerms: Vehicle reidentifcation, vehicle matchingalgorithm, automatic license plate recognition(ALPR)data, travel time distribution, travel time probability.
