Enhancement of annual rings on industrial CT images of logs

Many researchers have shown the ability of X-ray computer tomography (CT) of acquiring knowledge about the internal structure of logs. Contrarily to medical CT scanners, industrial CT scanners should satisfy more extreme constraints (in terms of speed, duty cycle, radiation safety, reconstruction circle and aperture size) in order to be used in a sawmill environment. All these factors lower the quality of industrial CT images while automatic methods for analyzing data are required. To ensure good performances and robustness, an automatic system should adapt to the quality of the input images. The paper focuses on an algorithm which can adaptively improve the contrast of the ridge and valley structure of annual rings in CT images of wooden logs. We also show that this algorithm does not only help in segmenting annual rings but also gives with little adaptation a good approximation of the localization of the pith.