A CCD-based 3-D imaging system for the generation of facial surface models

Three dimensional information about the human face is of importance, not only for medical diagnostic purposes, but also as input to facial recognition systems and, more recently, for the entertainment industry. The orthodontic surgeon in particular needs quantitative information about the average sizes and relations of the constituent parts of the human face as a whole or parts of it. A number of methods including a matrix of mechanical probes, lasers, holography. Moire fringe patterns and stereophotogrammetry have been investigated as possible ways in which three dimensional records of human heads could be made. Each method has its own merits and demerits ranging from accuracy requirements, safety factor to the subject, complexity and cost of analysis. The aim of this research was to develop components of a non-contact stereophotogrammetric system to acquire, process, display and replicate the surface of the human face. In contrast to previous work in photogrammetry that used two film based camera stations, this system's key requirements were all-round coverage of the complex facial surface, accuracy and surface quality and a data acquisition time of less than 2 seconds (approximate time for which a person can remain expressionless and motionless). The resulting design is unique in its combination of the complex three dimensional surface coverage, accuracy and speed. The data are acquired using four Pulnix T526 CCD cameras mounted around a semi-circular steel rig. Using an area based stereomatcher, dense disparity models are generated automatically. With the use of control points imaged in the scene, the three resulting models from the stereomatcher can be combined through transformations to form one complete facial model which can then be manipulated to yield profiles, distances between features of interest, and angles in a graphics visualisation suite. Planimetric accuracy is 0.1 mm and accuracy in height is 0.3 mm