Cycle-spinning based contourlet denoising for multiple image

Cycle-spinning (CS) is useful to avoid the effect of Pseudo-Gibbs phenomena, and many variations of CS method have been proposed. However, the previous work aims to reduce the noise from a single image, which result will be limited. With the thought of cyclic shift, this paper builds a novel CS-based framework for multiple images by estimating state-space relationship among the inter-frames. The unique feature of the proposed algorithm is that it is effective for the contaminated images with various noise characteristics, even if the noise information are unknown for the imaging system. Experiments demonstrate the effectiveness of the proposed method and show the superiority to previous single image denoising methods in objective and subjective qualities.