Denoising optical coherence tomography using second order total generalized variation decomposition

In this paper, we apply image decomposition for image denoising by considering the speckle noise in the (OCT) image as texture or oscillatory patterns. A novel second order total generalised variation (TGV) decomposition model is proposed to remove noise (texture) from the OCT image. The incorporation of the TGV regularisation in the proposed model can eliminate the staircase side effect in the resulting denoised image (structure). By introducing auxiliary splitting variables and Bregman iterative parameters, a fast Fourier transform based split Bregman algorithm is developed to solve the proposed model explicitly and efficiently. Extensive experiments are conducted on both synthetic and real OCT images to demonstrate that the proposed model outperforms state-of-the-art speckle noise reduction methods.