License plate recognition using topology structure features

License plate recognition (LPR) has been widely used for intelligent transportation systems. In this paper, we present a novel license plate recognition method using characters´ topology structure and feature-weighted template matching. As the topology skeleton feature is the most radical and intrinsic characteristics, our proposed algorithm is robust against license tilt and noise influence. First, detect license and correct skew using edge detection and Hough transformation. Then in the recognition step, segment characters using connected component detection method. We use improved template matching method to emphasize the importance of skeleton and contour. Our novel thinning algorithm extracts the topology features of characters effectively. Experiments show our proposed approach can achieve high recognition rate, and is robust against interference.