Abstract
Digital inline holography (DIH), as a three-dimensional (3D) measurement technique, is widely used in characterizations of the particles, droplets or bubbles under different multi-phase flow circumstances. By analyzing the phase information carried by the interference pattern, the reconstruction of shape and the location of a test target is then achieved. However, such reconstruction mechanism produces different levels of uncertainty between the in-plane (the plane parallel to the hologram plane) direction and out-of-plane (the plane normal to the hologram plane) direction, and the uncertainty of the latter is larger than the former. Also, the reconstruction algorithm fails when the interference patterns of some sections of the target are overlapped on the hologram since the overlapped patterns are merged into a pure shadow which doesn’t carry any phase information. This paper tested a method, the Multi-view Digital Inline Holography (MvDIH), that combines the holograms recorded from multiple views to overcome the addressed defects of the single view DIH. This technique uses the similar setup as the DIH but applies a different post-process method to implement the reconstruction. As the DIH is applied to each view, one can not only acquire the cross-section of the target in the hologram plane but also the outline of such cross-section in the space. Then, two reconstruction methods with different ideologies are developed as, the one based on the outline and the one based on the cross-section. A post-process algorithm is developed to realize these two reconstruction methods with the holograms recorded from different views. To evaluate the performance of the Multi-view DIH, a test model which imitates the droplet and liquid ligament structure is 3D printed and measured by the proposed method. The results demonstrate that, with only three view, both method provides limited reconstruction result. When comparing to the true test model, for the outline based method, some parts of the reconstructed model are missing and some details are merged into one piece with simple geometry. Yet, for the cross-section based method, the reconstructed model contains redundant parts which also make such result unsatisfied. As the used holograms are increased to six views, the reconstructed result for cross-section based method is approaching to the true model, but still some sections are reconstructed with certain level of ambiguity.
On the Danger of Applying Statistical Reconstruction Methods in the Case of Missing Phase Information
