Estimating the 3D Motion and Morphology of Line-like Structures from Orthogonal Projections

In image processing, the tracking of visible objects through time is a very common task. This task normally requires the definition of the object boundaries in order to describe its motion. However, that approach not always gives sufficient information about the position of specific points located on the surface of the object, especially in case the shape of the object also changes. Using additional information and making some simplifying assumptions can solve this general and mathematically ill-posed problem. In this research, the problem is restricted to the assessment of the motion and morphology of line-like 3D-structures (lines or tubes) by tracking the positions of small segments of such structures in two orthogonal projections. The method is based upon several assumptions regarding the physical properties of the real-world object. A typical (biomedical) application is the tracking of arteries for subtraction angiography and densitometric measurements. Keywords— motion estimation; object recognition; image processing