Application of denoising techniques to micro-tomographic images

Microcomputed tomography (microCT) is a particular version of computerized axial tomography commonly used by radiologists which reaches resolutions of the order of a few micrometers. In biology, this technique is especially useful for the study of hard tissues, such as calcified bone and dental matrices, because of their high linear attenuation coefficient. MicroCT images, though, are affected by a strong noise component, neither Gaussian nor white, caused by the characteristics of the acquisition system itself. In this paper we consider the problem of microCT image denoising, and compare the performance of two well-known denoising techniques and of BM3D, a recent technique based on the nonlocal approach. Although for the time being the performance analysis is mostly qualitative, results speak clearly in favor of BM3D.