Technical Report: High-Quality Head Model Calibration Pipeline

Denis V. Ivanov, Victor S. Lempitsky, Andrey V. Selivanov and Yevgeniy P. Kuzmin
Mathematics and Mechanics Dept., Moscow State University
 Moscow, Russia

Abstract:

In this report we present a technique for generating a complete, high quality model of a human head from orthogonal images taken from front and profile directions. In order to utilize common knowledge about the head structure we consider a polygonal model of a generic head and calibrate it in accordance to the parameters estimated from images. The calibration process is designed as a pipeline of sequential stages that register available views in a framework associated with a head, adjust the geometry of a model in order to obtain precise matching with the images, and generate a consistent texture by merging the texture candidates obtained through inverse texture mapping. This novel approach in texture generation allows the elimination of visual differences between regions extracted from front and profile images while maintaining a level of detail comparable to the original resolution. Eye calibration is presented as a post-processing procedure, which may be required for appropriate animation. The visual quality of the obtained models depends on the spatial resolution of a generic mesh and the resolution of initial images, which allows the proposed calibration pipeline to be used in a wide range of applications including computer games, film production and video conferencing.

This report extends the paper published as a Short Paper at Eurographics 2002.

Available downloads (zip archives): (MS Word)
Other material may be available on request.
See other publication references here.