Three-Dimensional Symmetry Identification Part II : General Algorithm and its Application to Medical Images

Implementation details of the algorithm for symmetry identification of an arbitrary three-dimensional object based on the theoretical points discussed in [9] are described in this part. The algorithm supports identification of a wide range of symmetry types represented by groups of proper and improper rotations and location of corresponding axes and/or planes of symmetry. It enables quantitative evaluation of identified symmetry types by means of a continuous measure of symmetry called the symmetry degree. In order to deal with rotational symmetries a novel method for two-dimensional symmetry identification based on complex moments is introduced, Execution times are also discussed. The algorithm application to symmetry identification and invariant positioning of three-dimensional medical entities reconstructed from their CT scans is described. It is found that in some fields of medicine the algorithm output may be useful in the clinical practice. Implementation details of the symmetry identification algorithm based on the theoretical issues discussed in [9] and its application to medical images are given in this part. Experimental results appear in [10].