Adaptive filtering for image enhancement and noise reduction en computed tomography images

Computed Tomography (CT) is being increasingly utilised as a powerful tool for advanced Non-Destructive Testing and Evaluation (NDT&E) applications. CT images reconstructed from the attenuation datasets, particularly in the first and second generation CT systems, are prone to various noise and artefacts; hence, a need exists to explore post filtering options with an aim of image enhancement and noise reduction. However, this has to be achieved in such a way that the basic features and edges are not lost in the processing. In the present work, 2-D adaptive filtering has been applied and studied in comparison with conventional non-adaptive centred averaging filtering for various CT images. Adaptive filters have been implemented by computing the central pixel value using neighbourhood centred averaging and subsequently applying the adaptive correction on. each pixel using the local variance and overall noise variance estimates. The paper highlights the results of non-adaptive and adaptive filtering on various CT images. Adaptive filtering has been found to provide excellent results for noise reduction and image enhancement of CT images, without losing information of edges, boundary and feature details.