2D to 3D dynamic modeling of architectural plans in Augmented Reality

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 04 | Apr 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 2384
2D to 3D dynamic modeling of architectural plans in Augmented
Reality
Priyanka Shinde1, Abhijit Turate2, Kesha Mehta3, Renuka Nagpure4
1,2,3, Student, Dept. of Information Technology, Atharva College of Engineering, Maharashtra, India
4 Assistant Professor, Dept. of Information Technology, Atharva College of Engineering, Maharashtra, India
------------------------------------------------------------------------***-----------------------------------------------------------------------
Abstract- The research paper develops an augmented
reality application “AR-CHI-TECH” which proposes to
develop a dynamic 3D model from a 2D floorplan drawing.
It takes input from a 2D floorplan drawn by an architect
and dynamically generates a corresponding 3D model of
the same. Later this model is displayed in Augmented
Reality for a better visual understanding. This is achieved
using a 3D game engine Unity3D and an Augmented
Reality (AR) engine Vuforia. The experimental results
confirm that the system obtains accurate dynamic 3D
models using reliable 2D floor plans.
Keywords: Augmented Reality; floorplan; dynamic
3D modeling; 2D; image processing; AR;
I. Introduction
Augmented reality uses the existing real-world
environment and puts virtual information on top of it to
enhance the experience. It is often observed that
ordinary, non-technical people find it hard to follow the
complicated 2D floorplan drawings by an architect. As
these drawings include a lot of technical jargon and
symbols, the knowledge of which is unfamiliar to
ordinary people. These 2D images of a 3D space (like a
house floorplan) do not give a clear understanding of
how the finished product will look.
ARchiTech helps to solve all the above-
mentioned problems under a single roof. It takes input as
an image of a floorplan. That dynamically creates a 3D
model of the same at runtime. Lastly, displays the model
in AR.
This helps the user to see the floorplan
thoroughly which helps them to visualize how the
house/floorplan might look before its completion.
Mobile devices nowadays have become pretty
powerful and have the impressive processing power.
This fact can be leveraged to deploy Augmented Reality
applications on modern mobile devices. Also, the
increased network speeds help in reducing the delay
between the client and server communication required
for the app.
II. Proposed System
The floorplan that architect designs are hard to
distinguish by ordinary people. Also, 2D images of a 3D
space do not give a clear idea of how the final product
will look.
AR is basically used to digitally place an object
on the surface of a real-life existing body. AR lets us
interact with virtual stuff like it is present in front of us.
The process of augmented reality is preliminary.
The device supporting the AR feature uses a lens
to scan the object of which the digital impression is to be
created. The algorithm and the software are then used to
measure the distance of the object. AR is, in fact, an
enhancement version of the real physical world.
III. Literature Review
After doing thorough research we have noticed
that there are a few systems that exist under the same
domain. But these models have drawbacks and are not
very optimized. Several such products have some errors
like image processing, and inefficient image cleaning
while converting the 2D floorplan data to a 3D model.
This project mainly attacks these drawbacks and
works on building the model sustainable for a long time
and also very efficient. Thus, we put forward a project
for dynamically converting the floorplan to 3D. With the
optimization, we also offer the consumers an easy-
to-use UI as well as high portability, alongside the
mobile app.
IV. Methodology
Execution of the project is done in 3 phases,
specifically
 Image Processing,
 2D to 3D conversion
 Displaying in AR.

This is how the project architecture will look:
Figure 1: Project Architecture.
Initially, the image must be selected from the
mobile device that is to be sent to the server for
processing. This is achieved with the help of a plugin
called AndroidNativePlugin [9]. This plugin uses core Java
methods to provide basic Android functionalities like
toast messages, gallery access, etc.
A. Image Processing.
Once an image has been captured as shown in
figure 2 it is then sent to the server, which is then passed
through a series of OpenCV-python functions to extract
information. Firstly, a dilate operation is applied to erase
finer details like text. This dilate operation can be
performed multiple times to extract any unwanted
details. The components exceeding a minimum threshold
value in the size of pixels are stored.
Figure 2: Before Processing
Image Processing technology and OpenCV 4.5.4
library [7] in Python version 3.9[4] are used to detect
walls and clean noise for optimal visualization. As the
image is being processed as a grayscale the dilated floor
plan with minimum distortion is inverted (negated) with
the help of the ImageOps Module of Pillow Library [8]
available for python. Later, another round of denoising is
carried out resulting in a noise-free, well-processed floor
plan.
Figure 3: After Processing
B. 2D to 3D conversion.
The pre-processed image from the first phase is
used as an input here. The NumPy-STL Library [6] is
used to convert the input image to a 3D STL Model. This
is a simple library used to make working with STL files
(and 3D objects in general) fast and simple. Due to all
operations heavily relying on NumPy’s this is one of the
fastest STL editing libraries for Python available.
Figure 4: After converting into a 3D model
C. Augmented Reality (AR) display.
The final phase of implementation is the one
where the client receives the 3D model from the server.
The model is then displayed in Augmented Reality to
obtain a better idea of what the actual, finished product
would look like.
To achieve this, an Augmented Reality (AR)
engine is required. The AR engine we used in this project
is named Vuforia [5]. Vuforia uses computer vision
technology to recognize and track planar images and 3D
objects in real-time and requires two parameters to
complete this phase which is a 3D model which is
provided by the server and an “image target”, upon
which it overlays the 3D model.
Vuforia runs multiple scripts in the background
that work in conjunction and whenever the target image
is detected, it overlays the 3D model upon it and enables
the user to experience how the finished construction will
look.

Figure 5: Displaying 3D model in AR
V. Result and Analysis
The end result of this project is a 3D model,
which is obtained from a drawing of a 2D architectural
floorplan, after undergoing image processing and
dynamic 3D modeling. The main thought behind this
project was to create an app that provides multiple
functionalities like image processing, dynamic 3D
modeling, and displaying in AR all under one roof.
This app uses a server running Python
scripts, so as to relieve the client’s device from all the
heavy processing and at the same time acquire the
benefits of Python’s extensive image processing libraries.
The Vuforia SDK uses a tracker-based method which also
brings a little delay in the AR display phase.
VI. Conclusion
In this paper, we developed a tool to extract
values from a 2D image of a blueprint and created a 3D
model in real-time. The major outcome of this research
was to create 3D models dynamically and offer
customers the ability to customize their prefabricated
home interiors while keeping the design of the exterior
structure. Augmented Reality offers a real visualization
of the final product for customers, enabling clients to
create their dream homes.
REFERENCES
[1] Developing augmented reality-based
application (Accessed November 2021);
Link:
https://www.researchgate.net/publication/337
466471_
[2] Pillow Library (Accessed November 2021);
Link: https://pillow.readthedocs.io/en/stable/
[3] Dynamic 3D model construction using
architectural house plans (Accessed January
2019); Link:
https://ieeexplore.ieee.org/document/7872850
[4] Vuforia Augmented Reality SDK (Accessed
September 2021); Link: -
https://en.wikipedia.org/wiki/Vuforia_Augmen
ted_Reality_SDK
[5] Numpy-stl Library (Accessed September
2021) Link:
https://numpy-stl.readthedocs.io/en/latest/
[6] OpenCV Library (Accessed September 2021);
Link: https://en.wikipedia.org/wiki/OpenCV
[7] Stack Overflow (Accessed August 2021);
Link:-
https://stackoverflow.com/questions/5427461
0/separate-rooms-in-a-floor-plan-using-opencv
[8] Python (Accessed 2020/21); Link:
https://www.python.org/
[9] Parsing floor plan images (May 2017).
DOI:10.23919/MVA.2017.7986875. Samuel
Dodge University Jiu Xu, Bjorn Stenger.
Conference: 2017 Fifteenth IAPR International
Conference on Machine Vision Applications
(MVA).
[10] 3- Phase Recognition Approach to
Pseudo 3D building generation from 2D
floorplan (June 2011). Raj Kishen Moloo,
Muhammad Ajmal Sheik Dawood, Abu Salmaan
Auleear. International Journal of Computer
Graphics & Animation (IJCGA).
[11] Generating 3D Models from 2D Floor
Plan Images Using Ensemble Methods
(December 2020).
Sungsoo Park 1,2 and Hyeoncheol Kim,
Department of Computer Science and
Engineering, Korea University, Seoul, Korea
(MDPI).
[12] Floorplanning for 3-D VLSI Design
(July 2005), Lei Cheng, Liang Deng, and M. D. F.
Wong. Asia and South Pacific Design Automation
Conference, 2005 (IEEE).

2D to 3D dynamic modeling of architectural plans in Augmented Reality

More Related Content

Similar to 2D to 3D dynamic modeling of architectural plans in Augmented Reality (20)

More from IRJET Journal (20)

Recently uploaded (20)

2D to 3D dynamic modeling of architectural plans in Augmented Reality