A Denoising Method Based On Image Reconstruction From Partial Spectrum Data

In this thesis, we proposed a novel denoising approach based on reconstruction-average mechanism. First, different parts of the original complete spectrum are chosen. Each partial spectrum is then used to reconstruct an image based on singularity spectrum analysis (SSA) model. We achieve denosing by averaging these reconstructed images since they have the same noise-free image but different stochastic noise. In this thesis, we apply this algorithm on both one-dimensional (1-D) and two-dimensional (2-D) partial spectrum reconstruction. The experimental results on both simulated and real images demonstrated that proposed method is able to obtain efficiently denoising effect while maintaining high image quality, it presents significant advantages over conventional denoising methods. The proposed method maintains the advantages of both averaging multi-signal denoising method and single signal denoising method, we call it denoising by image reconstruction from partial spectrum data.