Abstract: Classification of license plates is an important step in License Plate Recognition system (LPR).To perform the classification task effectively,a KNN (K-Nearest Neighbor) approach based on V-optimal histogram is proposed.To improve the efficiency of classification algorithm,the multi-class problem was firstly converted to be two 2-class problems based on the reverse color information.The skewness of color moments were used as the color features of license plates,which were extracted in the HSV (hue-saturation-value) color space.And then,the two binary classification problems were resolved by two KNN classifiers.Furthermore,V-optimal histogram approach was adopted to reduce the computation complexity of KNN and improve the efficiency of classification algorithm.Comparative experiments were conducted on real images.The experimental results show that the proposed algorithm achieves good classification performances.

Key words: V-optimal histogram, classification of license plates, K-Nearest Neighbor (KNN)