Localization of vehicle number plate based on watershed transform and visualization technique

Vehicle number plate detection is a complex research problem in digital image processing. Many related issues are addressed in the research carried out while detecting vehicle number plates. The features considered in number plates for detection are boundary features, global image features, texture features, color features, character features etc. Majority of the works reported have used one of the above features to detect vehicle number plate. In this work, an attempt is made using two features to detect the number plate, for better efficiency than the single feature based methods. The proposed approach uses watershed algorithm(boundary feature) as well as visualization technique(color feature) for detection and also takes care of environmental conditions such as indoor/outdoor, day/night, rainy/sunny and less sensitive for plate variations in size, location, color and inclination. The implemented method is tested on the samples used in existing methods to study and compare the efficiencies.