Speckle suppression and texture preservation in optical coherence tomography images using variational image decomposition

Filtering off speckle noise while preserving fine details for optical coherence tomography(OCT) is particularly challenging. In this paper, an efficient method based on variational image decomposition(VID) is proposed to suppress speckle from OCT retinal images. A new BL-G-BM3D model based on VID is proposed to decompose one OCT retinal image into the background part, the texture part and noise. Each part is described by suitable function space separately and processed individually. The proposed model is able to preserve structural information while sufficiently suppressing speckle noise. We test the proposed method on two raw OCT retinal images with low contrast and high noise level, and compare with four other related and widely used filtering methods in terms of both quantitative evaluation and visual quality.The experimental results have demonstrated the validity of the proposed method.