�ݺ�ߣ

Guided by,
Mrs. Alpha Vijayan
Associative. Professor
CSE Departement
Presented by,
Sam.V.Varghese (11010012)
Sudheesh K.S (11010022)
Tibin Thomas (11010016)
Gokul G (11010021)
19 October 2014 Dept of CSE 1

• Reduce rise of immoral digital visual contents in
broadcasts.
• Motivate the research in fast and reliable obscene and
immoral material filtering.
• Use the power of computer vision for socially
admissible projects.
• Contribute to open source OpenCV library function.

• Manual censoring and grading in tv broadcasts and
films.
• IP address based filtering of internet sites.
• Computer vision based censoring.(limited)

• Difficulty in Skin and Non-skin colour separable.
• Illumination changes over time.
• Skin tones vary dramatically within and across
individuals.
• Different cameras have different output for the
identical image.
• Movement of objects cause blurring of colours.
• Ambient light, shadows change the apparent colour of
the image.

Skin colour based detection using combination of
HSV & RGB colour space
 Allows fast processing.
 Robust to geometric variations of the skin patterns.
 Robust to resolution changes.
 Experience suggests that human skin has a
characteristic colour, which is easily recognized by
humans.

• Process of finding skin-colored pixels and regions in an
image or video.
• Skin Detector transforms a given pixel into an
appropriate color space .
• Then use a skin classifier to label the pixel whether it is
a skin or non-skin pixel.
• The color needs to be represented in a color space
• RGB color space ,HSV,YCrCb are common color space
used to represent images.

• Pixel-Based Methods
• Classify each pixel as skin or non-skin individually,
independently from its neighbors.
• Color Based Methods fall in this category
• Region Based Methods
• Try to take the spatial arrangement of skin pixels into
account during the detection stage to enhance the methods
performance.

Automatic Detection of Human
Body
 Detects whether there are human immoral scenes
present in an Image.
 The system marks skin-like pixels using combined
color and texture properties. .
 These skin regions are then fed to a specialized
grouper, which attempts to group a human figure.

Automatic Detection of Human
Body (cont..)

• The algorithm successfully extracts 43% of the test images
• This system is not as accurate as some recent object
recognition
• It is detecting jointed objects of highly variable shape
diverse range of poses, seen from many different camera
positions.
• Both lighting and background are uncontrolled previous
object recognition experiments.
• It might detect non humanbody whose color similar to
human

A Novel Scheme for Intelligent
Recognition of immoral Images
• Features extracted from skin region and whole image.
• Fourier descriptors and signature of boundary of skin region
is used as shape descriptor features.
• Self Organizing Feature Maps (SOFM).
• Finally, the selected features are fed to parallel classifiers
• The output of each classifier is sent to decision making
component.
• experimental results are convincing but the results can
be improved by extracting high .tp:86.7% fp:6.64%

• Large number of pictures taken under special lighting
• A feasible solution for the problem is to adapt the adopted skin
chroma distribution
• The skin tone is formed by the interaction between skin and
light
• Model based on Cb and Cr values can provide good coverage
of all human races.
• The popular skin colors of immoral images, including various
lighting conditions, into several categories

 every sample skin image is first trans-formed from
the RGB color space to the YCbCr space
YCrCb is an encoded nonlinear RGB
signal, commonly used by European
television studios and for image
compression work.
Y – Luminance component, C –
Chorminance
The detection rates of the immoral images
and the ordinary images are 71.7% and
84.2%,

HSV based skin detection
• HSV skin detection : -
• Image is converted from RGB colour space to HSV
colour space.
• The converted image is probed for the skin colours
according to the H channel discription for skin colours.
• The noisy image is passed through morphological
filters to produce the desired image.

HSV based skin detection

HSV based skin detection(cont..)

 The objective of this paper is to show that for every color
space there exists an optimum skin detector scheme
 such that the performance of all these skin detectors schemes
is the same.
 Used 4 color space – RGB, YCrCb, HSV, Cr Cb
 Proved mathematically for the existence of optimum skin
color detector using Neyman-Pearson Test

• Hsv size gave the best performance, superior to the other
model with less false detection
• Hsv model gives slightly better performance as compared to
Gaussian mixture.
• It is possible that color spaces other than RGB could result in
improved detection performance.

comparison of 5 different color spaces HSV,
HS,Normalized RGB and YCrCb
HSV, HS gave the best results
Normalized rg is not far
behind
YCrCb gave poor results

The project will involve building -:
 Create a skin color/texture database
 Extract frames from the video to get images that
can be used for analysis
 Search for skin color in the image
 Extract the image part with the skin color.
 Shape identification
 If match stop video for some minutes repeat,this
after sometime

1. Get the
video/image
Pass the video/image
Break the video into
images
Mark objectionable
Detect
skin color
Extract the part with
skin color in it
Shape
identification
No Yes
No end

 Create a skin color/texture database
 Extract frames from the video to get images that
can be used for analysis
 Search for skin color in the image
 Extract the image part with the skin color.
 Shape identification
 Post on imposition
 We use opencv for implementing modules

 With consistency increase growth in the immoral scenece
market we need to stop the people to have illegal use of the it
 This tool will help them to detect and block the supply this
things to the people who are not supposed to see them.
 This can be also in the censer board to detect any kind of porn.
 This can also be used by the parents to stop their children from
watching unwanted thing without their permission or till
he/she is not up to certain age limit.

THANK YOU

�ݺ�ߣ

immoral scene sensoring

More Related Content

immoral scene sensoring