TY - GEN A1 - Baum, Daniel A1 - Titschack, Jürgen T1 - Cavity and Pore Segmentation in 3D Images with Ambient Occlusion N2 - Many natural objects contain pores and cavities that are filled with the same material that also surrounds the object. When such objects are imaged using, for example, computed tomography, the pores and cavities cannot be distinguished from the surrounding material by considering gray values and texture properties of the image. In this case, morphological operations are often used to fill the inner region. This is efficient, if the pore and cavity structures are small compared to the overall size of the object and if the object’s shape is mainly convex. If this is not the case, the segmentation might be very difficult and may result in a lot of noise. We propose the usage of ambient occlusion for the segmentation of pores and cavities. One nice property of ambient occlusion is that it generates smooth scalar fields. Due to this smoothness property, a segmentation based on those fields will result in smooth boundaries at the pore and cavity openings. This is often desired, particularly when dealing with natural objects. T3 - ZIB-Report - 16-17 Y1 - 2016 UR - https://opus4.kobv.de/opus4-zib/frontdoor/index/index/docId/5915 UR - https://nbn-resolving.org/urn:nbn:de:0297-zib-59151 SN - 1438-0064 ER -