Fuzzified Gabor filter for license plate detection

摘要

This paper presents a new algorithm for detecting and extracting license plates from a complex image by using a fuzzy two-dimensional Gabor filter. The filter parameters are fuzzified to optimize the Gabor filter. In particular, the orientation and wavelengths of the Gabor filter are fuzzified. Fuzzification of the wavelength results in greater selectivity. These parameters dominate the filtering results. Bell's function and triangular membership function were proven to be the most efficient method for selecting the filter parameters in our experiments and were used for fuzzification. The use of these filters provided satisfactory results; the components of interest were efficiently extracted, and the procedure was found to be very noise-resistant.