Abstract: As an important means of the automatic traffic management, the technology of license
plate recognition plays a very important role in transportation surveillance and control. License plate
recognition can generally be divided into four parts: vehicle license plate location, license plate
correction, license plate character segmentation and character recognition. In license plate location, it
is often applied to simple background scene, but not suitable for complex background if only using
the texture or color features of the location. In character segmentation, the method of single line is
mature, but the method of double line license is not ideal. A location method is provided that combine
twice color calibration in HSV space and texture feature and a segmentation method of single or
double rows of characters. Because of the property of fixed colors in a plate, the picture is marked
twice, and then the projection method is used to locate the plates. The experimental result reaches
98% accuracy in the test of 200 different background pictures. Meanwhile, the improved method of
template matching reaches 95% accuracy.

Key words: license plate location, character segmentation, character recognition, SVM classifier