Image Processing Reference
In-Depth Information
Fig. 4.16. An example of a weighted mask for calculation of distance using the
5 × 5 × 5 neighborhood.
Fig. 4.17. Example of 3D distance transformation. (The 26-neighbor distance was
employed. Figure shows only a part of a large 3D image.)
image are kept unchanged. In this transformation, distance values are
calculated at all 1-voxels (Fig. 4.17). The use of Euclidean distance is most
preferable, but computation load is heavy. Squared Euclidean distance is
often employed. Various types of algorithms have been developed as will
be presented in Chapter 5.
Remark 4.20. Many reports have been published concerning a distance
function and a distance metric. Most of them, however, have treated a 2D
image, and have discussed properties of metric theoretically [Borgefors84,
Borgefors86a, Borgefors86b, Klette98, Okabe83a, Okabe83b, Ragnemalm90,
Toriwaki92, Verwer91]. Discussion about their effectiveness in practical image
processing is limited. Still some of them contain useful lists of literatures for
related research. Furthermore, recent progress in technology such as decrease
 
Search WWH ::




Custom Search