License plate recognition system based on SIFT features

cale invariant feature transform (SIFT) describing local features is a robust and reliable method for many pattern recognition purposes and can be applied to a wide range of problems in which local features are critical and helpful, like recognizing characters of license car plates. This work is based on using SIFT for license plate recognition (LPR) considering the capabilities and flaws of using the method. Some cases of failure or bad recognition are improved with various kinds of image preprocessing, however some kind of failures of car plate detection are essential and need more investigation and substitute techniques. Thus, applying a method based on distribution of vertical edges is employed to detect the car plate position. Numerical rate of success employing the proposed method has been given for our database versus pure SIFT for comparison