An Automatic Color Image Restoration Filter
Mieng Quoc Phu, Nathan Faggian, Peter Eric Tischer · 2006
In this paper, a novel approach to impulsive noise detection with automatic parameter tuning is proposed for colour image restoration. First, a simple noise estimator is used to estimate the amount of noise in the given image, then a global adaptive region growing scheme (ARG) is used to separate uncorrupted clusters of pixels from the corrupted clusters of pixels. ARG has automatic parameter tuning so it is image independent. Based on the noise proportion measured and the classification of noisy pixels, a switch based vector filter is proposed for the reconstruction of corrupted pixels. Various performance analysis show the detection scheme is more robust for a wide range of impulse noise than some of the state-of-the-art detectors. Our proposed filter outperforms other filters in both objective and subjective image assessments