Scribd
Upload
38 views

Face Recognition

This document describes a face recognition system created by Taral Oza. It uses 3D facial recognition to identify individuals from images and video. The system detects, aligns, and measures facial features in 3D to create a unique numerical code for each face. It then matches this code to those in a database to identify or verify people. This provides more accurate recognition compared to 2D systems by accounting for variations in lighting, orientation, and expression.

Uploaded by

gevijahok
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPS, PDF, TXT or read online on Scribd
38 views

Face Recognition

This document describes a face recognition system created by Taral Oza. It uses 3D facial recognition to identify individuals from images and video. The system detects, aligns, and measures facial features in 3D to create a unique numerical code for each face. It then matches this code to those in a database to identify or verify people. This provides more accurate recognition compared to 2D systems by accounting for variations in lighting, orientation, and expression.

Uploaded by

gevijahok
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPS, PDF, TXT or read online on Scribd
You are on page 1/ 29

Face

Recognition
System
Made By :
Taral Oza (2044135)
What exactly it is?
 A face recognition system is a computer-driven
application for automatically identifying or verifying a
person from a digital still or video image.

 It does that by comparing selected facial features in the


live image and a facial database

 typically used for security systems

 A new trend, claimed to achieve previously unseen


accuracies, is three dimension face recognition
Problems Definition ….
 Identify and locate human faces in Image
regardless
1) Position
2) Scale
3) Orientation
4) Illumination
Software measured for...
 Distance between the eyes
 Width of the nose
 Depth of the eye sockets
 The shape of the cheekbones
 The length of the jaw line
Terms…..
 Each human face has approximately 80 nodal points
 These nodal points are measured creating a
numerical code, called a faceprint, representing the
face in the database.
Something about FR….
 In the past, facial recognition software has relied on a
2D image to compare or identify another 2D image
from the database
 To be effective and accurate, the image captured
needed to be of a face that was looking almost directly
at the camera, with little variance of light or facial
expression from the image in the database. This
created quite a problem
 Even the smallest changes in light or orientation could
reduce the effectiveness of the system, so they
couldn't be matched to any face in the database,
leading to a high rate of failure.
3-D Face Recognition
 Capturing a real-time 3D image of a person's facial
surface
 3D facial recognition uses distinctive features of the
face
curves of the eye socket, nose and chin to
identify the subject.
 Using depth and an axis of measurement that is not
affected by lighting
 3D facial recognition can even be used in darkness
 ability to recognize a subject at different view angles
up to 90 degree
Steps by Software to recognize the face
1) DETECTION
digitally scanning an existing photograph (2D) or by
using a video image to acquire a live picture of a
subject (3D).

2) ALIGNMENT
the system determines the head's position, size
and pose.
As stated earlier, the subject has the potential to be
recognized up to 90 degrees, while with 2D, the
head must be turned at least 35 degrees toward
the camera.
Steps by Software to recognize the face

3) MEASUREMENT
measures the curves of the face on a sub-millimeter
(or microwave) scale and creates a template.
4) REPRESENTATION
system translates the template into a unique code
This coding gives each template a set of numbers to
represent the features on a subject's face
Steps by Software to recognize the face
5) MATCH
If the image is 3D and the database contains 3D
images, then matching will take place without any
changes being made to the image.

there is a challenge currently facing databases that


are still in 2D images

3D provides a live, moving variable subject being


compared to a flat, stable image. New technology is
addressing this challenge.

Cont……
Steps by Software to recognize the face
5) MATCH
For example, the outside of the eye, the inside of the
eye and the tip of the nose will be pulled out and
measured.

Once those measurements are in place, an


algorithm (a step-by-step procedure) will be applied
to the image to convert it to a 2D image.

After conversion, the software will then compare the


image with the 2D images in the database to find a
potential match
Steps by Software to recognize the face

6) Verification or Identification
If identification is the goal, then the image is
compared to all images in the database resulting in a
score for each potential match (1:N).
Small Algorithm For DVR….
Small Algorithm For DVR….

 First, RGB image is captured from the PTZ camera


 system judges whether captured image is included in
face candidate area or not.

 To do this…..
 Haar-Like method :- Haar-like features can be
constructed in many shapes and computed in
different ways.
Small Algorithm For DVR….
Haar-Like method using…….

The used Haar-like oriented edge filter having a


block structure can be evaluated very fast.
Small Algorithm For DVR….

 With DVR, tracks faces at high-speed and extracts


facial features by moving the PTZ camera up and
down, right and left, and zoom in and out etc.
 face image is modeled and processed by the
proposed Normalized-Intensity/Discrete-Cosine
Transform/Hidden
Small Algorithm For DVR….

Normalized Intensity:-

 Input image is affected by outside illumination

 image is normalized within a certain range of


intensity.

 Excessive changes of intensity affect frequency


domain
(a) (b)
 (b) is brighter than (a), and it displays the DCT result
that adds 100 to the pixel value of original image.

 This resulted in a great change of frequency


coefficient, and it accordingly affected detection of
facial features and decreased face recognition rates.
HMM Modeling and Training……..
 1D-HMM is running with input image
 The width and height of image is divided by
overlapping block that has the same width and height
(L)
 Overlapping size (P) of consecutive block is same as
or smaller than L-1.
 And then 2D-HMM is running by dividing forehead
into 3 areas, eye into 5, nose into 5, mouth into 5,
and chin into 3.
HMM Modeling and Training……..
The proposed face recognition algorithm
 Step 1 : using the viterbi’s back tracking algorithm

 Step 2: The recognized face corresponds to I

 Step 3: repeat step 1 and 2 for all the stored data in


database.

 Step 4: determine the candidate by comparing the


probability in Step 2 with the probabilities
stored in database using maximum likelihood
 Step 5: repeat Step 4 after removing the best
candidate in database and select more 2
candidates.
 Step 6 : if the probability is generated different from
the probabilities of database during Step 4-5,
the system judges the unidentified person to
be an invader, stores his/her image newly,
and then alarm the appearance of invader.
 Step 7 : if the probability value is located in a certain
range in step 4-5, recognition is considered
to be ‘success’ and then user name is
displayed.

You might also like