Structure-preserving multi-scale decompositions for skin and retinal OCT image

Optical coherence tomography (OCT) is an emerging technique for imaging skin and retina. However, severe noise and incompatible dynamic range between OCT image and display devices make OCT image analysis a challenging task. To address these problems, we present a structure-preserving multi-scale decomposition approach. Instead of outputting single image, it represents an OCT image by a set of images in multiple scales while preserved structures. Experimental results demonstrate that skin and retina structures can be well separated from noises and illuminations based on such decompositions.