-2-
FIELD OF THE INVENTION
The present invention relates to a presentation system for improving the appearance of a
caller in video conferencing for a better presentation at the receivers end. In particular the
invention relates to geographical and racial categorization of the callers skin type for applying
digital makeup which can run in real time.
BACKGROUND OF THE INVENTION
A videophone that can enhance the background environment of the user or the users
appearance, face, hairstyle and clothing has remained a dream for a long time. Document WO
99/57900 discloses a videophone with an enhanced user-friendly imaging system and an
information transfer device for communication between one or more viewers. The system
permits appearance, motion, sound and other characteristics of the users and their
environment to be changed, enhanced, filtered, deleted or modified'. The imaging system
obtains sensory information from users and their environments for transmission to the

-3-
viewer/s. This document provides a presentation system for creating a changed, enhanced,
filtered, deleted or modified presentation for viewers from the essential information on their
sensory appearances. A two-way communication circuit can be constructed from this one-way
communications for use in education, communication, entertainment and other multi-user
utilities.
However, the major hindrance in the system disclosed is that it does not provide any feature
for differentiating between different skin types and other characteristics of a user. Makeup
requirements for European or African skins, for example, can be different and similar make up
for two skin types may not be able to bring pleasing results.
There was thus a need for developing a system for improving the appearance of the caller in a
video conferencing for better presentation at the receivers end.
SUMMARY OF THE INVENTION
The main object of the present invention is to provide a system with digital toning for
application on the callers face, body and clothes for improving the overall looks perceived by
the call receiver in video conferencing, etc. The system functions at the callers end and can
run in real time during video conferencing.

-4-
This object is achieved by using colour clustering for clothes detection, skin and eye detection
and also uses light balancing techniques in order to improve the lighting variance on the face.
It also applies custom make up on the basis of skin colour. Smart body tracking is used and
digital make up process can be applied which can run in real time over a video conferencing
device such as digital television on mobile cell phones.
In a preferred embodiment, the present invention comprises: a presentation method for
improving the appearance of a caller in video conferencing for better presentation at the
receiver's end, said method comprising the steps of applying pre-processing operations on the
image of a captured raw stream; categorizing geographical / racial aspect of skin by detecting
skin type and locating eye-position; applying colour clustering over the captured image based
on the skin categorization; and applying post processing operators to the identified regions for
giving a better make up and resulting improvement in the perceived appearance of the caller.
BRIEF DESCRIPTION OF THE ACCOMPANYING DRAWINGS
The invention can now be described in detail with the help of the figures of the accompanying
drawings in which:

-5-
Figure 1 is a block diagram illustrating the user setting for the
system arrangement of the present invention.
Figure 2 illustrates in flow diagram form the operation of the present invention.
DETAILED DESCRIPTION
The system of the present invention operates in four stages. First, preprocessing operators
are applied on a stream of captured images. The skin type is then detected and eye position
located. Colour clustering over the images is applied and finally post processing filters are
used for removing visually displeasing noise from the callers face and body and improving the
overall looks of the caller.
In the first step preprocessing operators are applied on the image so as to decrease the
complexity of the system. The probable filters being colour quantization, low pass and down
scaling. These preprocessing filters help in reducing the computing requirements further in

-6-
the system of the invention because highly complex colour variance scenes take more CPU
cycles in colour clustering. This smaller sized and low complexity image is fed as input to the
skin detection and eye localization system. The light balancing techniques remove the lighting
invariance which is one of the biggest challenges in robust skin detection. Now depending on
the platform in which the system will be used, either of the two light balancing or low pass will
be used for simplification of the image.
The skin is detected in hue saturation variance (HSV) colour space due to its invariance to light
changes. Only the check is done on hue (H) and saturation (S) components. Red-green-blue
(RGB) to hue saturation variance (HSV) colour conversion is done before this skin pixel
detection. These operations being very CPU intensive and time consuming, they are carried
out on the first frame. Motion tracking is used for the subsequent image frames. Face and
eye detection provide vital information about the posture and location of body, for example, in
which direction the body is sitting. This step also categorizes the skin on the bases of
geographical / racial colour, as face toning requirements vary from race to race. For example
the make up requirements for lighter and darker skins are different. This categorization is
done by setting fixed quantifiers based on statistical learning done before hand using test
images of different races.

-7-
Once the face and eye location is detection via face localization of detected skin pixels, these
statistical details can be used in the next step of colour clustering for classifying colour data
sets which enables the access to the body location. Efficient colour clustering algorithms such
as N level K means are used, which are optimized utilization of K means algorithm. The
number of iterations that directly governs the accuracy of colour clustering is kept low, as
rough estimate of the clusters is required, for body location. Higher the number of iterations
higher is the CPU requirement hence this factor is also considered. Also to meet the real time
criteria the colour clustering will only be applied on area surrounding the face.
Now once the system has identified the body, post processing operators are applied on to the
identified regions in the form of hue balancing and brightness control, combined with noise
removal under eyes. Here colour toning is applied to the clothes / body identified. Now the
input specifications or the post processing operators are based on the skin colour
categorization done during an earlier step. While applying the hue balancing filter the system
also balances light on the face so as to give a better makeup. The new light balancing
technique for face is based on the skin categorization technique adopted.

-8-
The present invention provides caller polishing which considers important factors like
geographical / racial criteria for applying a makeup. Thus the invention provides correct digital
makeup due to skin colour categorization. The user looks good over the conference and the
system eliminates the need for daily make up. It makes video conferencing more user friendly
as users can adopt attractive features during video conferencing. The process can run in real
time in the current DTV / PC / Mobile video conferencing setup.

-9-
WE CLAIM
1. A presentation method for improving the appearance of a caller in video conferencing
for better presentation at the receiver's end, said method comprising the steps of:
- applying pre-processing operations on the image of a captured raw stream;
- categorizing geographical \ racial aspect of skin by detecting skin type and
locating eye-position;
- applying colour clustering over the captured image based on the skin
categorization; and
- applying post processing operators to the identified regions for giving a better
make up and resulting improvement in the perceived appearance of the caller.
2. The method as claimed in claim 1, wherein said preprocessing operations comprise
using colour quantization, low pass and down scaling filters for further reducing
computation requirements.

-10-
3. The method as claimed in claim 1, wherein before the step of skin pixel detection and
locating eye position, colour conversion is carried out from red-green-blue (RGB) to hue
saturation variance (HSV).
4. The method as claimed in claim 1, wherein said colour clustering process is applied on
the area surrounding the face using N level K means.
5. The method as claimed in claim 1, wherein said post processing operators are
applications of colour applied toning in the form of hue balancing and brightness control
and noise removal under the eye.
6. The method as claimed in the preceding claim 5, wherein said post processing
operators applied are based on the skin colour detected in the geographical / racial
categorization step.

-11-
7. A presentation method for improving the appearance of a caller in video conferencing
for better presentation at the receiver's end, substantially as herein described and
illustrated in the accompanying drawings.
Dated this 14th day of August 2007

The main object of the present invention is to provide a system with digital toning for
application on the callers face, body and clothes for improving the overall looks perceived by
the call receiver in video conferencing, etc. The system functions at the callers end and can
run in real time during video conferencing.
This object is achieved by using colour clustering for clothes detection, skin and eye detection
and also uses light balancing techniques in order to improve the lighting variance on the face.
It also applies custom make up on the basis of skin colour. Smart body tracking is used and
digital make up process can be applied which can run in real time over a video conferencing
device such as digital television on mobile cell phones.
In a preferred embodiment, the present invention comprises: a presentation method for
improving the appearance of a caller in video conferencing for better presentation at the
receiver's end, said method comprising the steps of applying pre-processing operations on the
image of a captured raw stream; categorizing geographical / racial aspect of skin by detecting
skin type and locating eye-position; applying colour clustering over the captured image based
on the skin categorization; and applying post processing operators to the identified regions for
giving a better make up and resulting improvement in the perceived appearance of the caller.