International Journal of Computer Science Trends and Technology (IJCST) Volume 2 Issue 2, Mar-Apr 2014

ISSN: 2347-8578 www.ijcstjournal.org Page 49

Virtual Trial Room
Saurabh Botre
1
, Sushant Chaudhari
2
, Shamla Mantri
3
Research Student
1&2
, Assistant Professor
3
,
Department of Computer Engineering
MIT College of Engineering,
Pune-India

ABSTRACT
In 21
st
century, fashion is not just something we wear, it's a way of life. The amount and type of clothing worn is dependent on
physical stature, gender, as well as social and geographic considerations. Our aim is to build a compelling, interactive and
highly realistic virtual system, where users/customers can choose between many different types of garments designs and
proceed to simulate these garments on virtual users. Here in this paper, we have proposed a system which helps in coordination
of everyday fashion. The system Virtual Dressing Environment involves virtually trying out different cloth models which is
done by mining of the user image, alignment of models and skin color detection of image (clicked from a fix distance). Our goal
here is to save time of the users during trying out different attires while shopping in stores or online [15].
Keywords:-Virtual Dressing Environment, Haar Classifier, Size recommendation, 2D-model, Face detection, Haar Cascades.
I. INTRODUCTION
Trying different clothes in stores and finally selecting the right
one is a time consuming and tedious task. It may also happen
that the cloth ordered during online shopping may not fit the
customer and the recent studies have shown that the overall
online cloth shopping experience is not satisfactory. Fashion
can reflect the age, society, and lifestyle modifications with
reality. Virtual dressing environment is online equivalent of
the near-ubiquitous in-store changing room. It enables
shoppers to try on clothes one or more of size, fit or style, but
virtually rather than physically. It has been evident that when
customers are unable to try, most will not buy. Recently,
Virtual try-on of clothes has received much attention due to its
commercial potential. This virtual environment can be very
useful for online shopping or intelligent recommendation to
narrow down the selections to a few designs and sizes[15].
The entire process initiates from simply the alignment of the
user and the cloth models with accurate position, scale,
rotation and ordering. The detection of the user and the body
parts is one of the main steps in the creation of the system[6].
There are several approaches that are proposed for body parts
detection such as face and upper body detection, posture
estimation and skeletal tracking. Moreover the description of a
human body shape is a complex and application-dependent
task. Kjrside et al. [2] proposed a tag-based approach for
manual labeling of body parts in order to create an augmented
reality of the customer wearing a cloth simulation. One of the
methods for the body parts detection is the use of the shape
descriptors such as Histograms of Oriented
Gradients(HOG)[5]. But it may not be convenient for a

continuous body tracking system, the main reason is the use of
a limited number of skeletal posture states.
Different approaches that are available for face detection are:
Finding faces using images with controlled
background: This is the easy way out. We can use
images with a plain monocolour background and use
them with a predefined static background. By
removing the background, it will always give us the
face boundaries[6].
Finding faces by color: If color images are used, the
typical skin color to find face segments might be
used. The main disadvantage here is it is not very
robust under varying lighting conditions and doesn't
work with all kind of skin colors[9].
Finding faces by motion: If a real-time video is used,
the reality that a face is almost always moving can
help a lot. The face can be easily detected by
calculating the moving area. The main disadvantage
here could be,if there are moving objects in the
background[7].
Combining all the good techniques: The several good
approaches can be combined,which would yield an
even better result.
The various advantages our system would serve are as follows:
It helps to make a right choice for the size of clothes
without actually trying it [4].
International Journal of Computer Science Trends and Technology (IJCST) Volume 2 Issue 2, Mar-Apr 2014
ISSN: 2347-8578 www.ijcstjournal.org Page 50
Provides a platform for experimenting with clothing
style.
A wide range of other services like jewelry, glasses
and hair style recommendation can be done by using
the 2D model generated[11].
The precious time of consumers would be saved by
this system, as it reduces the human effort for trying
many clothes.
In this work, extraction of the user based on Face Detection
using Haar Classifiers is shown in section II. Image
Processing using OpenCV is explained in detail in section III.
Section IV gives us the background removal technique.
Section V gives results and conclusion and section VI lists the
references.
II. FACE DETECTION USING HAAR
CLASSIFIERS
For Haar classifier object detection the core basis is its Haar-
like features. Rather than using the intensity values of a pixel,
these features use the change in contrast values between
adjacent rectangular groups of pixels. Then the contrast
variances between the pixel groups are used to determine
relative light and dark areas in the image. The haar-like
feature is formed by two or three adjacent groups with a
relative contrast variance. The haar features can be easily
scaled and examined by increasing or decreasing the size of
the pixel group. Haar like features as shown below can be
used for face detection[1].



Fig 1. Edge features

Fig 2. Line features

Fig 3. Four Square Features
OpenCV comes with a trainer as well as detector. It can be
trained with your own classifier for any objects like cars,
cricket bat etc. We have used OpenCV to create one such
trained classifier.Here we will deal with face detection.
OpenCV already contains many pre-trained classifiers for
face, eyes, smile etc. Those XML files can be stored at our
desired path. We have developed face detector with OpenCV
using haarcascades. Initially we need to load the required
XML classifiers in our system. We have used the
haarcascade_frontalface_default.xml as the base xml file.
Later on the input image (or video) in grayscale mode which
is captured by the camera of the system is loaded.Now we find
the faces in the image. If faces are found, it returns the
positions of detected faces as Rect(x,y,w,h). We can create a
ROI for the face detected, when this locations are detected[8].

Fig 4. Result image of Face Detection
III. OPEN COMPUTER VISION (OpenCV)
It is an open source software. OpenCV has C, C++, Python
and Java interfaces which supports Windows, Linux and
Android. It was basically designed for efficient computation.
The OpenCV library can take advantage of multi-core
processing as it is written in optimized C/C++. The usage
ranges from interactive art, to the inspection of mines and also
stitching maps.[13]
International Journal of Computer Science Trends and Technology (IJCST) Volume 2 Issue 2, Mar-Apr 2014
ISSN: 2347-8578 www.ijcstjournal.org Page 51
The following modules have been used in our Virtual
Dressing Environment:
core:
Core is the basic module. The basic data structures
(e.g.- Mat data structure) and basic image processing
functions are included in the core. Other modules
like highgui, etc also make use of this module.
highgui:
Highgui provides simple user interface capabilities,
video codecs, image and video capturing capabilities,
handling track bars, manipulating image windows
and mouse events and etc. If we use User Interface
frameworks like Qt, WinForms, etc we need more
advanced User Interface capabilities.
imgproc:
The imgproc module includes the basic image
processing algorithms like image filtering, image
transformations and color space conversions etc.
video:
It is a video analysis module which has object
tracking algorithms and background subtraction
algorithms etc.
objdetect:
The objdetect includes object detection and
recognition algorithms for standard objects.
IV. BACKGROUND REOMVAL

Fig 5. Background Removal. Image with white
pixels(left),transparent image(right)
The cloth models downloaded from web usually have a white
background. To view the cloth models on the user, we need to
avoid the occlussions of the cloth models with the users body
parts. For this purpose, we have used a technique called
Background Removal. In this technique we make the pixels
with RGB values ranging between 230 and 255 as transparent
by anding the rgb value with 0x00FFFFFF[14].
V. CONCLUSION
After an introduction, the related work was presented; starting
with cloth selection and virtual dressing environment, cloth
size recommendation system is available. Further a broader
look on the technologies and frameworks that were used for
the implementation, like Haar classifier algorithm, of the
Virtual Dresssing Environment was implemented. After this
the different aspects of the design process up to the
construction of the garment models were highlighted. This is
followed by the screenshots of implementation of Virtual
Dressing Environment.

Fig 6. Image of cloth model placed on user
In the last section the tests were executed, also discussing the
output, the appearance and the interaction with the Virtual
Dressing Environment. Overall, the presented Virtual
Dressing Environment seems to be a good solution for a quick,
easy and accurate try-on of garment. In this system compared
International Journal of Computer Science Trends and Technology (IJCST) Volume 2 Issue 2, Mar-Apr 2014
ISSN: 2347-8578 www.ijcstjournal.org Page 52
to other technologies like augmented reality markers or real-
time motion capturing techniques no expensive configurations
and time-consuming build-ups are required. From this point of
view it is an optimal addition for a cloth store and online
shopping.
REFERENCES
[1] P. Ian Wilson and Dr. J. Fernandez Facial feature detection using Haar
classifiers, JCSC 21, 4 (April 2006).
[2] F. Isikdogan and G. Kara A Real Time Virtual Dressing Room
Application using Kinect, Bogazi ci University, Istanbul, Turkey 2012.
[3] J. Shotton, A. Fitzgibbon, M. Cook, T. Sharp, M. Finocchio, R. Moore, A.
Kipman, and A. Blake, Real-Time Human Pose Recognition in Parts from
Single Depth Images, in Proceedings of IEEE Conference on Computer
Vision and Pattern Recognition,2011.
[4] F. Cordier, W. Lee, H. Seo, N. Magnenat-Thalmann Virtual-Try-On on
the Web, in proceedings of International Conference on Virtual Reality,
Laval Virtual, University of Geneva 2, May 16-18, MIRALab, 2001.
[5] Philipp Presle A Virtual Dressing Room based on Depth Data, Vienna
University of Technology, Klosterneuburg.
[6] K. Onishi, T. Takiguchi, and Y. Ariki, 3D Human Posture Estimation
using the HOG Features from Monocular Image, in proceedings of 19th
International Conference on Pattern Recognition, 2008.
[7] D. Chai, and K. N. Ngan, Face Segmentation using Skin-Color Map in
Videophone Applications, IEEE Transactions on Circuits and Systems for
Video Technology, vol. 9, no. 4, June 1999.
[8] J. Young Choi, Y. Man Ro and Konstantinos N. Plataniotis, Color Local
Texture Features for Color Face Recognition in proceedings of IEEE
Conference, 2011.
[9] P. J. Phillips, H. Moon, S. A. Rizvi, and P. J. Rauss, The FERET
evaluation methodology for face recognition algorithms, IEEE Trans Pattern
Anal. Mach. Intell., vol. 22, no. 10, pp. 10901104, Oct. 2000.
[10]. D. Chai, and K. N. Ngan, Face Segmentation using Skin-Color Map in
Videophone Applications, IEEE Transactions on Circuits and Systems for
Video Technology, vol. 9, no. 4, June 1999.

[11]. J. Young Choi, Y. Man Ro and Konstantinos N. Plataniotis, Color
Local Texture Features for Color Face Recognition in proceedings of IEEE
Conference, 2011.

[12] P. J. Phillips, H. Moon, S. A. Rizvi, and P. J. Rauss, The FERET
evaluation methodology for face recognition algorithms, IEEE Trans Pattern
Anal. Mach. Intell., vol. 22, no. 10, pp. 10901104, Oct. 2000.

[13] Euratex (2000), Bulletin 2000/5, The European Textile/Clothing
Industry on the eve of the New Millennium, Brussels.

[14] Blanz V., Vetter T. (1999), A morphable model for the synthesis of 3D
faces, in Computer Graphics, (Proc.SIGGRAPH99, Los Angeles California,
USA), ACM Press New York, pp. 187-194.3.

[15] Volino P., Magnenat-Thalmann N. (2000) Virtual Clothing-Theory and
Practice, Springer, Berlin Heidelberg.