A New Localization Method for Malaysian License Plate

Abstract

The License Plate (LP) localization plays an important role in License Plate recognition system in some countries which have enormous variety of LPs such as Malaysia. In Malaysia, the LPs have extraordinary diversities in fonts, shapes and even using handwritten fonts in LPs and so on. Moreover, since this paper works on LP at the back of the cars, the presence of LP font on the body of car can also adds a major complication for LP recognition systems. In this paper, a new License Plate Localization algorithm is proposed for Malaysia LPs by Otsu method. The empirically results present a new threshold of Blob Analysis to extract LP location. Additionally, the new thresholds are introduced for Morphological Opening Operator and objects labeling. At the end, the LP localization is concluded by new boundaries limitations which demonstrate exactly LP location in image. These new thresholds cause LP localization system gain 90% accuracy while it is higher than any other existence methods that are showed real result for Malaysian LPs localization. These results were obtained from entirely experimental works on LPs different shape and fonts of LPs in natural environment conditions such as raining, light reflection and so on.