Reconstruction of three-dimensional flow patterns from lineof sight visualisations

Program Code: 

A/Prof Harald Kleine (

Description of Work: 


Density-sensitive flow visualisation techniques such as shadowgraphy, schlieren or interferometry have a so-called line-of sight character. This poses serious problems when the flow field is three-dimensional, as the visualisation record can only provide a projected view of the flow which will not reveal how the flow changes along the line of sight. A possible solution is to apply a tomographic approach, where several projected views, each taken from a different view angle, are combined to reconstruct the flow field in all its spatial dimensions. An example of such visualisations is given below. One piece of information that is very important for the design of vehicles travelling at supersonic speeds is the knowledge of the exact shape of shock fronts, contact surfaces and expansion zones. This information can be extracted from the aforementioned tomographic views.


The quantitative analysis of three-dimensional compressible flows remains one of the major challenges in fluid mechanics, and only few and specialised attempts have been made to tackle this problem. Typical visualisations of such flows have a line-ofsight character, that is, they integrate all information of the flow along the depth of the flow field. It is the aim of this project to develop robust and efficient algorithms that can be used to extract the spatial distribution of the flow field from adequate visualisations, with special emphasis on sequences of images obtained from timeresolved recordings. The high number of such image records – typically more than 100 in a single experiment – poses a particular challenge as the algorithm for image processing has to be highly efficient and fast. The final goal of the project is to create a tool that allows one to reconstruct how three-dimensional features observed in a high-speed compressible flow evolve with time.

Description of work:

  • It is expected that this project will deliver a fast and efficient algorithm that allows one to reconstruct a three-dimensional, transient flow field from a sequence of line-of-sight visualisations, obtained with techniques such as schlieren or interferometry. This project will deal with the general case of a fully three-dimensional flow and the main objective is to determine the threedimensional shape of the observed flow features. The principal techniques for image acquisition are well established and a large amount of expertise and high-class equipment for this purpose is available in the School. There already exists a large amount of image data that is suitable for the proposed September 2014 30 analysis. Image registration algorithms that have been developed in the School for atmospheric imaging can be adapted and expanded to this application.