Abstract
In this thesis, we propose a new method for measuring the asymmetry measurement of cranium in infants. After obtaining the three-dimensional image of the head, we use the covariance analysis to filter the part of head with larger variation of curvature and a two-phase K-means classification to differentiate those parts. With the help of the geometric relationship of the cluster centers, we can locate the facial organs. Therefore, based on the locations of facial organs, we first rotate the three-dimensional head images to a common frontal orientation and then use the ICP algorithm for the fine registration. All the three-dimensional head images can be aligned to the same benchmark. Finally, we will find a sagittal plan to generate a mirrored head image. We use the head image before treatment compare with the mirrored image and after treatment, and estimating the asymmetric assessment. The experiment results, deformation parts displaying and volume calculating, demonstrate the feasibility of the proposed method.