An effective de-noising algorithm in CIE-lab color space using hybrid filtering

Bilateral filtering is effective for suppressing Gaussian noise but cannot filter out impulse noise. So we propose a de-noising algorithm mentioned in the title, which we believe is effective for suppressing both Gaussian noise and impulse noise. Sections 1 through 3 explain our de-noising algorithm and their core consists of: the bilateral filter has a good effect for suppressing the Gaussian noise but not for impulse noise; the recognized impulsive noise points are suppressed by the filtering based on the maximum of probability density and the remaining Gaussian noise points are still suppressed by bilateral filter; since the bilateral filter is nonlinear and only the recognized impulse noise points are filtered out differently, our algorithm can retain most of the edge features. The simulation results, presented in Fig. 1, and their analysis demonstrate the superiority of our method in suppressing the mixed noises, which consist of Gaussian noise and impulse noise, compared with the other de-noising methods of color image.