Image De-noising Based on Nonparametric Adaptive Estimation in the Finite Ridgelet Domain

Ridgelet is a new signal analysis method which is especially suitable for describing the 2-D signals which have linear or super-plane singularities. Recently, an orthonormal version of ridgelet for discrete and finite-size images has been presented, named finite ridgelet transform (FRIT). In this paper, we propose a new image de-noising method by FRIT with the threshold based on nonparametric adaptive estimation. Experiments show that this method represents the better characteristic than the traditional de-noising methods in wavelet domain and the de-noising methods based on donoho strategy