Far East Journal of Experimental and Theoretical Artificial Intelligence
-->
Abstract: In this paper, a fast
algorithm of locating variable size license plate is proposed. The proposed
algorithm is composed of image preprocessing, locating the possible row
positions, determining the horizontal boundaries of license plates within
candidate rows and selecting the most possible location from the candidates. In
the first stage, a new de-interlacing algorithm which uses a median filter is
proposed. In the second and third stages, the run-length
encoding (RLE) scheme is usedto determine the plate
position of image. Finally, the last stage is calculating the local edge-segment
density of candidates using their RLE codes to determine the most possible
location. Since the algorithm represents and uses the input images by simple
text forms, the computation speed is fast and the system can be used in
real-time. In the simulations, the resolution of license plates of the test
images varies from to
and
the location rate is 96.69%.
Keywords and phrases: license plate location, de-interlace, run length encode, median filter, variable size license plate.
