Depth Estimation Using Collection of Models

  Karankumar Thakkar,  Viral Borisagar

Today images are easy to acquire, view, publish, and share however they lack critical depth information as the images usually are projected views of a 3-D scene. This makes severe restrictions for many image manipulation, editing, and retrieval tasks. Depth estimation is a technique that aims to retrieve depth information either based on depth cues such as texture, focus and shading or using collection of 3D models. With the recent considerable interest in 3D image analysis, estimating depth information has become a rapidly evolving topic in computer vision research. Also it finds applications in various imaging applications including depth-enhanced image editing, novel view generation etc. Hence, we have strong motivation to consider the problem of adding depth to an image of an object and provide a basis for 3D reconstruction. In this paper, we present an automatic method to find depth information of single image by employing collection of models of the same object class. The key advantage of this method is that even if the dataset does not contain the exact 3D model of the imaged object, it will characterize shape components and generate its depth map. We apply our method on various indoor objects like lamp, chair, cup and car and obtain plausible depth maps.

Keywords- Depth estimation,Collection of models,Pointcloud,depth map,3D Reconstruction

Unique Identification Number - IJEDR1502175
Page Number(s) - 1017-1021
Pubished in - Volume 3 | Issue 2 | May 2015
DOI (Digital Object Identifier) -   
Publisher - IJEDR (ISSN - 2321-9939)

  Karankumar Thakkar,  Viral Borisagar,   "Depth Estimation Using Collection of Models", International Journal of Engineering Development and Research (IJEDR), ISSN:2321-9939, Volume.3, Issue 2, pp.1017-1021, May 2015, Available at :http://www.ijedr.org/papers/IJEDR1502175.pdf