A License Plate Image Enhancement Method in Low Illumination Using BEMD

Recognizing vehicle license plate image captured in low illumination place is a difficult problem. To solve the problem, this paper proposes a new license plate image enhancement method using bidimensional empirical mode decomposition (BEMD) technique. BEMD is a 2D data-driven adaptive nonlinear signal decomposition approach derived from the 1D empirical mode decomposition (EMD). In the proposed algorithm, the main procedures are designed as the following: first, the license plate image is denoised by the use of alpha-trimmed mean filter and transformed from RGB color space into HSV color space, then, extract the V component to form intensity image for enhancement; second, with BEMD method, the intensity image is decomposed into a number of intrinsic mode functions (IMF) as well as a residual image; last, the brightness of residual image is adjusted using Retinex theory, and fused with the IMF images to achieve enhancement of license plate image. Experimental results show the proposed method provides superior performance over traditional schemes for license plate image enhancement in low illumination.