Three-dimensional reconstruction of fracture surfaces: Area matching algorithms for automatic parallax measurements

Abstract

This paper describes two solutions for systematic measurement of surface elevation that can be used for both profile and surface reconstructions for quantitative fractography case studies. The first one is developed under Khoros graphical interface environment. It consists of an adaption of the almost classical area matching algorithm, that is based on cross-correlation operations, to the well-known method of parallax measurements from stereo pairs. A normalization function was created to avoid false cross-correlation peaks, driving to the true window best matching solution at each region analyzed on both stereo projections. Some limitations to the use of scanning electron microscopy and the types of surface patterns are also discussed. The second algorithm is based on a spatial correlation function. This solution is implemented under the NIH Image macro programming, combining a good representation for low contrast regions and many improvements on overall user interface and performance. Its advantages and limitations are also presented.

This paper describes two solutions for systematic measurement of surface elevation that can be used for both profile and surface reconstructions for quantitative fractography case studies. The first one is developed under Khoros graphical interface environment. It consists of an adaption of the almost classical area matching algorithm, that is based on cross-correlation operations, to the well-known method of parallax measurements from stereo pairs. A normalization function was created to avoid false cross-correlation peaks, driving to the true window best matching solution at each region analyzed on both stereo projections. Some limitations to the use of scanning electron microscopy and the types of surface patterns are also discussed. The second algorithm is based on a spatial correlation function. This solution is implemented under the NIH Image macro programming, combining a good representation for low contrast regions and many improvements on overall user interface and performance. Its advantages and limitations are also presented.