Student seminars

Przemysław Kopycki #StudentSeminar: Reconstruction of integral images and examining integral imaging monitors

Europe/Madrid
Universe

Universe

Description

Lightfield imaging is a technique that allows the capture and display of 3D information from 3D scenes obtained either from computer simulations or from the real world. Apart from the display of 3D scenes, lightfield imaging has many useful applications such as, for example, obtaining depth maps, performing digital refocusing, removing occlusions, or making object recognition.

Recently, many computational volumetric reconstruction techniques are reported; computational volumetric reconstruction, 3D profilometry, pixel of elemental images rearrangement technique (PERT), PERT with projected empty space (PERTS), and so on. In general, computational volumetric reconstruction is used because of its simplicity. In this method, each elemental image is back-projected through the virtual pinhole array on the reconstruction plane. Since typical computational volumetric reconstruction uses the pixilated process and the number of shifting pixels for reconstruction is the fixed integer value and uniformly distributed, the depth resolution of the reconstructed 3D image is not sufficient. Further, the reconstructed 3D images have poor visual quality when elemental images with low resolution are used. To solve these problems, a new reconstruction method is required. I will present new computational volumetric reconstruction method with improvement of depth resolution and visual quality for the reconstructed 3D images considering continuously non uniform shifting pixels when elemental images with low resolution are used. In proposed method, we use non-uniformly distributed and integer valued shifting pixels for reconstruction. Thus, the depth resolution can be improved compared with the computational volumetric reconstruction method. Also, the visual quality of the reconstructed 3D images can be even enhanced than the conventional reconstruction.

Integral imaging monitor consist high resolution screen and pinhole array. We can ask if is there any optimum value for the pinhole size that allows the observers to have a sufficient 3D experience? The question comes out from the fact that to provide an accurate 3D reconstruction, the pinholes should be as small as possible. However, small pinholes give rise to dotted-like display aspect, and to images with very low brightness. Although these features are known, it is highly surprising that, at the best of our knowledge, no study has been reported aiming to link the pinhole aperture with the 3D capability, and consequently to find the optimum configuration in terms of 3D effect and brightness. During the presentation I will present designed in optics laboratory a psychophysical test specifically adapted for lightfield monitors, based on stereoacuity measurements using modified binary search (MOBS), to examine the 3D capacity of pinhole-type monitor.

Organized by

Pablo Martínez Reviriego