A Hybrid Approach for Characters Recognition in License plate Recognition System

Automatic license plate identification has many application in traffic system like electronic toll collection, custom check points. License plate recognition basically has three steps: license plate detection, segmentation of character and plate character recognition. In the first step of number plate identification system we use vertical projection method and edge detection method for number plate detection. In the next step of character segmentation part, we use vertical and horizontal projection and finally for character recognition we are proposing a hybrid approach. The Character Recognition is the key concept of automatic vehicle license plate detection and recognition system for accurate result. In the first method of hybrid approach, an image is divided into a sequence of horizontal scan lines by using raster scanning and each number appears in a matrix of size for 5x3. The presence of the pixel is represented by '1' while the absence is represented by '0' which would form a feature. This feature will be used to differentiate all characters and digits except similar looking characters. So, for these similar characters we used second method which is based on angle detection of the pixel of the non ambiguous part. We are taking first and seventh pixel of the matrix as a non ambiguous part and compare the angles of these pixels for recognizing the character.