A pattern-based directional element extraction method for vehicle license plate detection in complex background

An efficient and fast method for vehicle license plate detection in complex background is proposed in this paper. It uses several convolution pattern filters to enhance character strokes while reducing irrelevant edges and background noise. Experimental results show that our proposed method has very high detection accuracy and low false positive rate compared with former methods based on edge detection and Gabor filters.