Volume morphology and its application in 3-D image speckle noise suppressing

Computed Tomography(CT) generates a three-dimensional image of the internals of an object from a large series of two-dimensional X-ray images taken around a single axis of rotation. Speckle noise in 3D images, such as medical images and industrial CT images, inevitably affects the analysis results of the interested objects contained in the volume data. This paper defines a group of volume morphology arithmetic operators, mainly including open and close, by extending area morphology into 3-D space. Using these operators, the light and dark objects of small size could be removed directly from the 3-D spaces of the target objects, while the connectivity of the main 3-D target objects in the volume data is still preserved. To demonstrate the validity of the volume morphology operators, they are applied to suppress speckle noises in 3-D images of coral and rat skull. Experimental results show that the algorithm proposed in this paper processes volume data as a whole, so that could protect the 3-D shapes of the target objects, especially the boundaries in the vertical direction. Comparing with the traditional process of treating 3-D images as 2-D image sequences, this method is more beneficial to volume segmentation and feature extraction for fine structures.