AUTOMATED STEREOSCOPIC IMAGE CONVERSION AND RECONSTRUCTION. DISPLAYING OBJECTS IN THEIR REAL DIMENSIONS(STEREOSCOPIC IMAGE CONVERSION)
Judit Z. Tövissy and Emiliyan G. Petkov
ABSTRACT - In this paper a software prototype that is able to convert a single image into a stereoscopic image pair is discussed. The prototype uses methods of image analysis and self-developed algorithms to aid the reconstruction of pixel information between 2D-3D conversions, the details of which are discussed in depth. The prototype obtains estimated depth values from a single still image and computes a most likely recreation of the original scene in the image based on regional clustering, lighting and camera focus information. New theoretical improvements to enhance accuracy during the depth map estimation process focusing on a Statistical Depth Map are discussed as well. The last aim of this paper is to give an approach for creating 3D images that contain objects which have to be displayed and appear on autostereoscopic screens in their real dimensions. For example, we have a picture or 3D model of a museum artefact, tool or whatever and we want to display it on an autostereoscopic screen in its real dimensions, so viewer may perceive the most of the real object.
Keywords: stereoscopy, image, conversion, real, dimensions, autostereoscopic, display