Mixed filtering method based on Curvelet research and application

Adaptive median filtering method was mainly used for filtering Salt&Pepper (impulse) noise, while the method based on Curvelet transform was mainly used for filtering Gaussian noise. In the actual processing for sealed rubber ring image, which usually contains the mixture of Salt&Pepper noise and Gaussian noise. In order to filter out the mixed noisy image preferably, a new mixed filtering algorithm was proposed in this paper. First, pixel separation method was adopted to separate the pixels corrupted by Salt&Pepper noise from the pixels corrupted by Gaussian noise. Then for the former we utilize the adaptive median filtering method, for the later we utilize the improved threshold filtering algorithm based on the Curvelet transform. The simulated results showed that the mixed filtering algorithm was better for the mixed noisy image.