Site Loader

The
design and reconstruction of 3D models has now become a significant part of
computer graphics and of computer vision applications and also in areas such as
architecture, visual and special effects in entertainment, augmented and
virtual environments, games, engineering and education. 3D reconstruction1-3,
12 can be achieved by two approaches called as active and passive. 3D Max and
Blender are the examples of the traditional modeling system that uses 3D meshes
for the construction of 3D models. However, reconstruction from these 3D meshes
was challenging and therefore, resulted in need for a better approach. A major
progress has been seen in the last decade to resolve this issue. Photogrammetry,
Structure from motion(SFM) and Image-Based Modelling(IBM) are the  widely used modern approaches for building 3D
models. There are few alternative technologies such as laser scanner, lidar,
structured light, camera calibration, etc. used for the same purpose. 3D laser
scanner is accurate but quite expensive and hence not adapted easily. Though
structure from motion and photogrammetry fail to reach high performance, but it
is an inexpensive approach and is more accessible when compared with laser and
lidar3, 12.

Structure
from motion(SFM) approach comes from the roots of photogrammetry3 and
computer vision. Photogrammetry is the technology of extracting information from
images. It measures and processes features such as length, depth, angular
position, etc in images(photographs). This technology has been successfully
employed in a wide range of industries. The main application of photogrammetry
is to generate 3D models out of the images taken from an object, and also used for
many different purposes. Processing complicated shapes of small objects, uneven
surfaces, reflective surfaces and objects, extracting features of transparent
object surfaces are the challenges required to be improved3.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

SFM
allows to achieve 3D structure of object from images as well as the camera
motion which is done by bundle adjustment. The extracted features from 2D
images that include pixel positions, edges and corners of objects, lines and
curves across the corners and edges, angular positions of object with respect
to 3D coordinates, light intensity and other measurements such as length, depth,
resolution is given as input to the SFM algorithm. These images are taken at
different angles from a single camera by either moving the camera position or
by moving the object. Luo Jianxin et al in 2 has explained the basic principles
of SFM photogrammetry. The implementation methodologies and their research work
has been reviewed in 2.

The sfm
photogrammetry starts with processing of images and removing redundancy from
overlapping images. The feature extraction involves edge and corner detection
of the object from the images. Harris corner detection algorithm has been
proposed by Harris4. Then comes the feature matching process that is carried
out by  SIFT5 and BRISK6. The image
features are quantized using K-Means algorithm7 and Brute-Force algorithm to
compare features of images. The RANSAC8, 9 is used for the estimation of the
fundamental or projection matrix between two images and validation of the
putative matches from the images(closest descriptor of feature from two images).

To
determine the correlation between the world and image coordinates, camera
calibration is the method used10. The next step is to minimize the reprojection
errors and perform techniques to recover the structure. Bundle Adjustment  (BA)9, 11 is that step which performs
refinement of all elements after processing of images and matching the
features. Soulaiman El hazzat1 have proposed a method for 3D object reconstruction using structure
from motion approach, by global and  local bundle adjustment.

In this paper, Section II gives the idea
about the technology used and a brief introduction about the terms and
intermediary processes involved in structure from motion. Section III shows the
application areas of SFM and Section IV holds the proposed work for improving
the user’s experience in viewing the online products by providing a complete 3D
view of product, and further the complete process of 3D modelling has been
explained. The expected outcome is given Section V.

Post Author: admin

x

Hi!
I'm Erica!

Would you like to get a custom essay? How about receiving a customized one?

Check it out