An Efficient way of Number Plate Alphabets and Numbers Extraction for Security Purpose

Abstract

Extraction of number plate plays a key role in the recognition of vehicles. With the advancement in technology and increase in number of vehicles it is the need of the day to recognize registered and unregistered vehicles. For this purpose, in this paper we propose a system to automatically extract digits and alphabets from the number plate of vehicle and search it in the database for its registration. The proposed approach involves four different processes that include Image smoothing, Edge detection, Image segmentation and data extraction. The result shows that the proposed approach can easily detect and extract data from the number plates. Some of the number plates and its extracted data are shown in this paper. The results can further be implemented on automatic extraction of data from shields, sign etc. These two steps play an important role in the License Plate Recognition of a vehicle. There are many methods/algorithms developed for License Plate recognition. (8) and (7) extract the License Plate based on vertical edges using Hough Transform. (9) extract the colour using Neural Network and discriminate character using template matching. (10) Extracted the number plate based on genetic algorithm segmentation.(11) uses optical character recognition techniques for number plate recognition.(12) identified automatic vehicle using template matching and knowledge-guided boundary following. (13) Proposed a technique for extracting characters without preceding information of their location and range in the image. (14) Proposed a method for extracting Korean license plate on the basis of the colour of the plate. (6) Proposed feature based approach for the localization of number plate and its characters. Since number plate standards changes from place to place (3-6) therefore in this paper we propose a fresh based approach for accurately extracting data of same pixel value from the number plate. The proposed approach is independent of the digits and character variation but depends upon the analysis of the number of connected components. This paper is based on the analysis of connected components, containing four major steps that include, RGB to gray-scale conversion, Image smoothing, Edge detection, Image segmentation and data extraction. As RGB to gray-scale conversion is the key step in every image processing algorithm because it is easier to deal with 1 pixel value compare to 3 pixel values at one point and at the end it give us the same result.