Automatic feature point extraction for three-dimensional skull

作者：Feng Jun^*; Chen Yu; Tong Xin Long; He Xiao Wei; Zhou Ming Quan

来源：Optics and Precision Engineering, 2014, 22(5): 1388-1394.

DOI：10.3788/OPE.20142205.1388

• Benchmark models
• Feature point extraction
• Location accuracy
• Mapping relationships
• Similarity-matching
• Smooth regions
• Statistical deformable model
• Statistical modeling

A fully automatic skull feature point extraction method was proposed, which extracts the skull feature points by a partitioned statistical deformable model and a model similarity matching method. First, the statistic models of skull partition were constructed, and a benchmark model and a series of generated models were built by statistical model deformation. Then the mapping relationship between models was established and the model similarity was defined. Finally, the feature points of the model to be measured were indirectly obtained with the model similarity and the projection relationship. Experimental results indicate that the location average error of the feature points for an eye socket model is about 3.2325 pixels. When the distance threshold is 10 pixels (3% of the size of the model), the location accuracy for 90% of the feature points achieves 100%. The method proposed has higher accuracy and exaction for skull feature point extraction as compared with traditional methods, and can extract the feature points of smooth regions for skull models.

• 出版日期2014
• 单位西北大学