INTERACTIVE IMAGE-BASED 3D MODELING

ADM BABIKER, MAl BABIKER (2012) INTERACTIVE IMAGE-BASED 3D MODELING. PhD. thesis, Universiti Teknologi PETRONAS.

[thumbnail of 2012 -COMPUTER & INFORMATION SCIENCES - INTERACTIVE IMAGE-BASED 3D MODELING - MAI BABIKER ADM BABIKER.pdf] PDF
2012 -COMPUTER & INFORMATION SCIENCES - INTERACTIVE IMAGE-BASED 3D MODELING - MAI BABIKER ADM BABIKER.pdf
Restricted to Registered users only

Download (3MB)

Abstract

The Shape-From-Shading (SFS) problem is a fundamental and classical problem
m computer vision. It is the process of computing three dimensional shape of a
surface using its shading information from a single image. SFS deals with the
recovery of surface orientation and surface shape (height) from the gradual variation
of shading in the image. SFS is implemented by firstly modeling the image brightness
as a function of surface geometry and then reconstructing a surface which, under the
given imaging model, generates an image close to the input image.
In this work, a new interactive method for reconstructing shape from shading is
presented. Most of the existing works exhibit difficulties in computing 3D shape for
t1at areas as well as area with high brightness (highlights). Interactive shape from
shading uses human knowledge to compute the 30 shape for areas with zero gradients
on the surface. The previous techniques use either fast marching smoothing after
segmenting the image into smaller parts or mass spring fitting to provide smooth
transition between hills and valleys on the image. This work, unlike similar previous
techniques, focuses more on enhancing the problem areas in traditional shape from
shading which makes it efficient and robust.
In this work, the user interactively correct the generated 30 shape of objects from
the 20 image especially in highlight and dark areas in the image. This method uses
human knowledge for extracting as well as correcting the problem areas. After the
user has identified the rough location of the problem area. the exact area is selected
using edge detection process. Sometimes there are no clear edges in the Image.
Therefore, these areas have to be selected interactively using polygonal fitting.

Item Type: Thesis (PhD.)
Subjects: Q Science > Q Science (General)
Departments / MOR / COE: Sciences and Information Technology > Computer and Information Sciences
Depositing User: Mr Ahmad Suhairi Mohamed Lazim
Date Deposited: 23 Sep 2021 09:59
Last Modified: 23 Sep 2021 09:59
URI: http://utpedia.utp.edu.my/id/eprint/21620

Actions (login required)

View Item
View Item