Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 22

Dissertation I Seminar

On

A Methodology for Extracting


Standing Human
Bodies From Images
By

Miss. Zubaida khan (M.E. COMP)

Under the Guidance of


Prof. Vaishali Bodade

Agenda
1.
2.
3.
4.
5.
6.
7.
8.

Introduction
Literature Survey
Problem Statement
Existing System
Proposed System
Applications
Conclusion
References

Introduction

Sr.
No

Literature Survey

Title of the paper &


Author

Observations/Remark

Hand Gesture Recognition


for Human Computer
Interaction, by Meenakshi
Panwar and Pawan Singh
Mehra.

This paper describes a novel approach of hand


gesture recognition based on detection of some
shape based features. The overall algorithm is
divided into four main steps, which includes
segmentation,
orientation
detection,
feature
extraction and classification. The weakness of this
method is that it does not follow any systematic
approach to define certain parameters for gesture
recognition, and all the threshold values taken in
this approach are based on assumption made after
testing on some number of images. It heavily
depends on some constraints like location of hand
in image for orientation detection, proper gap
between each finger, etc. if any user fails to follow
these constraints then results may degrade.

Hand Gesture based


Interface for Aiding visually
Impaired, by Meenakshi
Panwar.

This proposed approach uses some pre-processing


steps for removal of background noise and
employs K-means clustering for segmentation of
hand object . It calculates some essential features
required for hand pattern detection and for
generating the 7 bit sequence for 36 different hand
shapes. The weakness of this method is that it
defines certain parameters and threshold values
experimentally since it does not follow any

Literature Survey Contd..


Sr.
No

Title of the paper &


Author

A Hand Gesture
Recognition using Feature
Extraction- Ashis Pradhana,
M.K.Ghosea and Mohan
Pradhana-ISSN 2277

Observations/Remark

This paper presents a very simple and efficient


approach for recognizing the hand gesture that
represents numbers from zero to nine. The design
of hand gesture recognition system is broadly
divided into two phase. The first phase is the
preprocessing phase and the second phase is the
classification phase. The weakness of this
approach is that the designed system is able to
classify only the static images
Vision Based Real Time
In this paper, a real time finger counter is proposed
4
Finger Counter for Hand
for counting the number of fingers (1-5) shown in a
Gesture Recognitionby S.
hand gesture. In the first phase, the gesture is
Nagarajan, Dr. T. S.
captured using a camera in real time. The skin color
Subashini, Dr. V. Ramalingam, of hand region is detected using HSV color space
- CPMR-IJT
and morphological operations are performed in the
third phase. In the feature extraction phase, the
convex hull method is used to detect the boundary
points of the segmented binary hand image. Finally,
the recognition phase recognizes the gesture
number based on the extracted features. The main
Dissertation seminar on Hand Gesture Recognition
for
drawbacks
of this proposed work are hand pose
5
HCI
orientation.

Problem Statement

Extracting the human bodies in cluttered images is very


difficult due to unrestricted position, high dimensional
pose.

The existing works require computationally complex


training and template matching processes.

We propose a methodology for automatic extraction of


human bodies from images, in the case of almost upright
poses in cluttered environments.

Existing System

Proposed System

The idea is to provide user with a natural interface that


allows controlling the computer with hand movements.

Input-> Hand Gesture


Output-> Assigned Action Executed

The proposed system consists of four phases,

1.

Image Acquisition

2.

Image Segmentation and Pre-processing

3.

Feature Extraction

4.

Gesture Recognition

Dissertation seminar on Hand Gesture Recognition


for HCI

A Methodology for Extracting Standing Human


Bodies From Images
Segmentation of human bodies in images is a challenging
task that can facilitate numerous applications, like scene
understanding and activity recognition.
Extracting of the human body in unconstrained still

in

images due to several factors, including shading, image


noise, occlusions, background clutter, the high degree of
human body deformability, and the unrestricted positions
due to in and out of the image plane rotations.

A Methodology for Extracting Standing Human


Bodies From Images
We propose a novel framework for automatic segmentation of
human bodies in images.
Our main aim is to extract the standing

human

from

images(single and multiple) by using skin detection algorithm.


We combine information gathered from different levels of image
segmentation, which allows efficient and robust computations
upon groups of pixels that are perceptually correlated.
Without making any assumptions about the foreground
and background.

10

Proposed System
The idea is to provide user can easily extract the human
bodies from any images.
The proposed system consists of two phases,

Multilevel Image Segmentation

Skin detection algorithm

11

PRAPOSED SYSTEM

INPUT
IMAGES

MULTI LEVEL
SEGMENTATI
ON
SKIN
DETECTION
ALGORITH
M

ROI 1

MERGIN
G

ROI 2

Figure 1: Block diagram of Proposed


System

EXTRAC
T
HUMAN
BODIES
12

Proposed System Methods


1. Skin Detection Algorithm
2. Multilevel Image Segmentation

13

Proposed System Methods


Skin Detection Algorithm :
Skin detection is an important step vision system , like gesture
recognition,hand tracking, video indexing,face detection ,
person tracking etc.
Skin detection challenging task due to various factors such as
illumination, camera characteristics, ethnicity, and added facial
components.
Basically, skin detection can be considered as classification of
skin pixels and non-skin pixels.
Colorspaces used for skin modelling [RGB, HSV, HSI, HSL, and
TSL]
14

For each text image

Apply color correction

Convert image to desired color


spaces

Calculat pixels

If P >T ,classify P as skin

Fig 2 :Steps for skin

15

Proposed System Methods (contd)


Skin Detection Algorithm Steps :
1. Read RGB images.
2. Find the image pixels by using two methods:
. Pixel based skin detection:- Pixel-based skin detection uses
the pixel color information for detecting skin pixels.
. Regional-based skin detection:- Regional-based methods
use the relationship between pixels to detect the human
skin.

16

Proposed System Methods (contd)


Regional-based skin detection steps:
1. Read RGB images
2. Find pixel in skin locas region
3. If pixel is locas at %> threshold then
Select pixel mark as white else
Mark as black
4.

Remove disconnected single pixels.

5.

Result binary map of skin region and non skin region .

6.

Stop
17

APPLICATION
1. We can easily Count of people in single images as well as multiple
images.
2. We can also use to manage the heavy croud.
3. Skin detection was used to detect anchors in TV news videos for
the sake of video automatic annotation, archival, and retrieval,
4. We can also easily extract the human bodies without making
any assumption of pose, and foreground and background .

18

CONCLUSION

The proposed system provides an simple color correction


algorithms for improving skin detection accuracy.

The proposed system provides a novel methodology for extracting


human bodies
from images.

Soft anthropometric constraints guide an efficient search for the


most
visible body parts, namely the upper and lower body, avoiding
the need for strong prior knowledge, such as the pose of the body.

The advantage of our methodology over those tested is that it


can automatically localize and segment the human body.
19

REFERENCES
[1] A Methodology for Extracting Standing Human Bodies From Single Images Athanasios
Tsitsoulis,Member, IEEE, and Nikolaos G. Bourbakis,Fellow, IEEE.
[2] M. Everingham, L. Van Gool, C. K.Williams, J.Winn, and A. Zisserman,The pascal visual
object classes (VOC) challenge, Int. J. Comput. Vis.,vol. 88, no. 2, pp. 303338, 2010.
[3] OJO, J.A. and S.A. Adeniran, Colour face image database for skin segmentation, face
detection, recognition and tracking of Black faces under real-life situations.
International Journal of Image Processing (IJIP), 2011. 4(6): p. 600 .

[4] Nadian, A. and A. Talebpour, A New Skin Detection Approach for Adult Image
Identification. Research Journal of Applied Sciences, Engineering and
Technology, 2012. 4(21): p. 4535-4545.
[5] Soriano, M., et al., Adaptive skin color modeling using the skin locus for
selecting training pixels. Pattern Recognition, 2003. 36(3): p. 681-690.

20

REFERENCES
[6] Brett Allen1. And Brian Curless. And Zoran Popovi. And Aaron
Hertzmann.2006." Learning a correlated model of identity and
posedependent body shape variation for real-time synthesis,"
Eurographics/ ACM SIGGRAPH Symposium on Computer Animation 2006.
[7] G. Mori and J. Malik." Estimating Human Body Configurations using
Shape Context Matching". ECCV 2002, pp 666-680.
[8] C. Barron, I. A. Kakadiaris." Estimating anthropometry and pose from
single image", CVPR 2000, vol.1, pp. 669-676.
[9] C.J. Taylor. "Reconstruction of articulated objects from point
correspondences in a single uncalibrated image". CVIU 80(3): 34963,
December 2000.

21

Thank You!!

22

You might also like