Automatic 2D to 3D Stereoscopic Conversion

With increasing amount of 3D content there is huge amount of growth in past many years, the availability of 3D content is still below the normal size by that of its 2D corresponding closely to one another. To remove this distance many 2D to 3D image conversion techniques has been proposed. It is important to estimate relative depth map in a single view image fo r 2D to 3D image conversion techniques. Semiautomatic image conversion method which makes use of human operators has been most successful but it takes many t imes to proceed and costly too. Hence automat ic image conversion methods come in result which reduces complexity. But have not still achieved same level of quality. The stereoscopic image provides information on detail s of each object in the picture in the three dimensions and helps us to feel the picture realistic one. The main step of 2D to 3D image conversion process consists of depth estimation fo r given 2D image. So this paper is focused is depth recovery. In this paper proposed 2D to 3D image conversion method is based on globally estimat ing the entire depth map of query image d irectly from the repository of 3D images which contains images and their depth pairs using nearest neighbour regression type idea .