Abstract
One of the objectives for a smart city is by applying the fast car retrieval system, which requires the tracking of a vehicle LP (License Plate) using a detection sensor system. ALPR (automatic license plate recognition) system is important through different walks of life, such as: security, access control, and law enforcement in traffic violation systems. In a previous work [1], multiple closed-set patterns belonging to the Saudi LP are extracted and classified using classification and clustering techniques. A reasonable delay in time is accumulated while extracting the image features before being categorized to a certain cluster. A problem arises that we are not able to process the amount of information in real-time, therefore it is crucial to reduce processing time. The solution is by applying the parallelization processes that enables the online processing of images by an embedded system. In this paper the parallelization of the testing phase of the system is studied using three different scenarios for the purpose to scale the relative speedup using general-purpose CPUs. The parallel computing processes make the system more efficient and reliable.