An Effective Algorithm of Character Segmentation for Chinese License Plates with Different Colors

This paper presents a character segmentation method of vehicle license plate using modified vertical projection. After analyzing the reason of failure of traditional segmentation methods, the paper gives some methods to deal with character adhesion and segmentation mistakes, which improves the success probability of segmentation obviously. The most valuable feature of this new algorithm is that it is useful to license plate with different colors. The experimental result shows that the success probability can reach 94%.