Propeller Cavitation 3d Reconstruction through Stereo-Vision Algorithms

In recent years an increasing interest towards innovative techniques to be applied in usual experimental facilities has grown. From cavitation tunnel point of view, attention has been given to measurement of dimensions of propeller cavitation bubbles. The need arises from the increase in ship performance and environmental/comfort requirements, leading to new standards in propeller design. In addition Computer Fluid Dynamics developed in recent years new tools to investigate these phenomena, but they still lack in validation with experimental data. Therefore the usual simple human observation seems to be no more adequate to determine propeller behaviour with respect to cavitation. Moreover it does not supply data directly and objectively comparable with numerical results. In this scenario computer vision seems to be a promising research tool, because of relative limited hardware cost and of the availability of many software libraries. Additionally, computer vision is continuously spreading in many fields providing new algorithms and techniques. In this paper the experimental setup and software tools currently under development in the cavitation tunnel of Genoa University are presented. In particular three standard video surveillance cameras are adopted, while the OpenCv computer vision library is used for software development. Although this library provides a good support to 3D reconstruction its application in this environment is not straightforward, mainly because the optical path is not standard. Currently a volume reconstruction method based on [1] and an active stereo technique, with source light produced by a triggered laser, are studied. In present paper both techniques are presented along with preliminary results, clearly outlining their merits and shortcomings.