Speckle Noise Reduction Mechanism Based on Dual-Density Dual-Tree Complex Wavelet in Optical Coherence Tomography

Image quality is an important parameter characterizing the performances of an optical coherence tomography (OCT) system. Low image quality not only deteriorates the image analysis and interpretations, but also impacts on the clinical applications of OCT systems, leading to misdiagnosis. Speckle noise is always present in OCT signals, and thus inevitably affects the OCT image quality. This paper studies the speckle noise reduction problem in OCT systems, and tries to compare a variety of the wavelet transform based methods. Specifically, we give the logical flow diagram of the dual-density dual-tree complex wavelet method first, and then combine it with the local variance estimation based bivariate contraction model for speckle noise reduction. By performing experiments on OCT images of human retina, swine eye and human dental, we compare the speckle noise reduction effects of the dual-density method, dual-density dual-tree real wavelet method (R2D) and dual-density dual-tree complex wavelet (C2D) method. Results show that the C2D method can effectively eliminate the speckle noise while retaining the important edge detail information of the OCT images.