Implementation of Dynamic Time Warping Method for the Vehicle Number License Recognition

Abstract

In the era of information technology vehicle numbers identification needs to be done by system automatically. Therefore, the accuracy of the data is well documented and work porses identification can be done quickly. Motor vehicle license recognition is a recognition system by comparing character feature in license plate with reference feature which exists in database. This system uses chain code method and template matching to extract character feature in license plate's image. Feature extraction with chain code method will result in an array of direction codes which stored in dynamic array, which stored in dynamic array. In this application test feature will be matched with feature stored in database using dynamic time warping method (DTW) to obtain a distance value between test feature and reference feature, the smaller the distance obtained shows that both the features are more similar. The result of this system is the recognition of each character in license plate's image. In this study, samples of license plate's images are tested with the number of research objects. From the study feature extraction is obtained with template matching method provides better success rate compared to feature extraction with chain code method, where the success rate of feature extraction with template matching method is at 78% whereas feature extraction with chain code method is at 68%.