License plate localization using MSERs and vehicle frontal mask localization using visual saliency for vehicle recognition

This paper presents a new method for the vehicle license plate and the frontal mask localization. The proposed license plate localization initializes candidate regions based on maximally stable extremal regions (MSERs). Then, the candidate regions are categorized into three classes of license plate character components, plate background components and the other components by using intensity, size, aspect ratio, and orientation of those candidate regions as features. Finally, a rule-based decision is applied to verify the candidate regions. For the frontal mask localization, we develop the method that does not refer to license plate location. Visual saliency, edge projection, symmetrical property, and Pyramid Histogram Oriented Gradients (PHOG) are applied in our proposed the frontal mask localization process. The experiments show that the proposed method can provide a precision of 95.32% and a recall of 98.07% for license plate localization process, and a precision of 97.00 % and a recall of 97.98% for the frontal mask localization process.