SVM approach to number plate recognition and classification system

Automatic maintenance of the bus arrival time requires numberplate recognition. One way to achieve the same is by making use of support vector machine (SVM) approach. SVM involves in finding the classification boundaries of the multi class problem. The features from the numberplate are extracted directly from the spatial domain and wavelet domain to get two sets of vectors. 75% of the collected vectors are used for training the vector for constructing the two SVM classifier (spatial domain and wavelet domain). 25% of the collected vectors are used for testing. 55.65 % success is achieved in case of spatial domain method. 77.78% success is achieved in case of frequency domain approach.