Segmentation of Facial Bones from Skull Point Clouds Based on Smoothed Deviation Angle

Abstract

The human skull was the subject of study in various fields. Segmentation could be a basic tool for better understanding the skull. One of the most challenging tasks was facial bone segmentation. Our previous study had succeeded in segmenting facial bones from skull point clouds, however the quality of the results needed to be improved. In this paper, we proposed a new method to improve the results of facial bone segmentation from skull point clouds. The method consists of three stages: deviation angle extraction, smoothing, and thresholding. Each point in the point cloud was assigned a value based on the deviation angle. These values then went through a smoothing process to clarify the differences between the facial bone region and other regions. Next, thresholding was performed to divide the skull into two regions, namely facial bone and non-facial bone. The proposed method had succeeded in improving the quality of the segmentation results by achieving precision=0.931, recall=0.9854, and F=0.9573.