Research on number-plate recognition based on neural networks

The paper presents a technique of number-plate recognition based on neural networks. Each number-plate in a color image can be correctly located by analyzing number-plate colors which are classified by neural networks. An integrated neural network and template matching method is used to recognize characters of a number-plate. Experimental results show that the correct rate of number-plate locations is close to 100%, and the number-plate locating time is less than 3 seconds using the proposed method. Moreover, recognition rate of characters is improved due to the known number-plate types and novel method of character recognition. It is also observed that our algorithm is not sensitive to variations of weather, illumination and vehicle speed. Another advantage of the proposed method is that the size of number-plate is not required to be known a prior.