Providing a method for automatic recognition of plaques tarnished in the process of Intelligent Transport

Among the important challenges in reading the car number plate are inappropriate
conditions such as the presence of spots or dirt on the plaque, plaque crown and nail
pliers which makes it difficult to read the car number plate. binarization is one of the
most important steps in reading a car number plate and the quality of it can have
strong impact on the whole car number plate recognition process. Many of the errors
found in readings in car number plate are because of mistakes in binarization
process. Although, several threshold detection has been presented, all of them are
not feasible for binarization of car number plates with defacing. In this article, the
authors tried to present a compound procedure based on otsu and niblack which are proper ways for binarization to have high quality binary image. The
presented method has been tested on 100 photos which has been taken by a 3.15
Megapixel camera with 2048*1536-pixel resolution and Finally, the quality of the
images of the existing methods and the proposed method is evaluated qualitatively