Problems of objects identification in three-dimensional X-ray tomography

X-ray computed tomography is a unique non-destructive technique which is suitable for dimensional measurements and quality testing. X-ray tomography is applied widely not only in the traditional medical area, but in natural science and engineering fields. The result of tomography is a volumetric map of attenuation coefficients which has, in fact, a limited practical sense. Usually, segmentation that converts the map into binary objects by means of threshold value is applied. In this paper we discuss the some key features of segmentation applied to a test object which is gravel composed of unconsolidated rock fragments. Gravel appeared to be a challenging object for segmentation and image recognition. One of effective approach for image processing of tomographic data, among many possible ones, is presented in this paper.