Craniofacial stereo mapping: improving accuracy with natural points

Abstract

Stereo-photogrammetry has been reported in both photogrammetry and medical journals as the most suitable technique for craniofacial landmark mapping. Various size and shape of control frames were designed to provide three-dimensional object-space control for stereo images. A research was carried out to determine whether the accuracy could be further improved using natural points. In the research, control points and natural points were measured using photogrammetric digitising software and their coordinates computed using bundle adjustment technique. In addition, the six-camera stereo-photogrammetric system was electronically synchronized for optimum accuracy. Subsequently, a digital photogrammetric workstation was used for stereo orientation and stereo-digitising of the anthropometric landmarks. For the accuracy assessment of the 'natural point' technique, absolute orientation was performed on the stereo-model using control-frame control and natural points respectively. The results show that the developed technique achieves an accuracy of 0.6mm at one standard deviation. The best accuracy of the stereo-photogrammetric system was 0.25mm