02-05-2011, 02:13 PM
Presented By
Archana H Sable
[attachment=13241]
Introduction
Face detection : Identify location and size human
faces in an image.
Challenges : Variations of image appearance
Pose (front, non-front).
Occlusion.
Presence or absence of structural components.
Image orientation.
illuminating condition.
facial expression.
Adaptive Bilateral Filter
The ABF retains the general form of a bilateral filter, but
contains two important modifications.
1. ζ[m0,n0] : Denotes, an offset which is introduced to the range filter in the ABF.
2. Second, both ζ[m0,n0] and σr the width of the range filter in the ABF are locally adaptive.
Steps to calculate ζ[m0,n0] and σr.
1] As before, let Ω m0,n0 denote the set of pixels in the window (2N+1)x(2N+1) of pixels centered at [m0,n0]
Let MIN, MAX, and MEAN denote the operations of taking the minimum, maximum, and average value of the data in Ω m0,n0 ,respectively.
Δm0,n0= g[m0,n0] − MEAN(Ωm0,n0).
demonstrate the effect of bilateral filtering with a fixed domain Gaussian filter σd and a range filter σr. shifted by the following choices for ζ.
1.No offset (conventional bilateral filter): ζ[m0,n0] = 0,
2. Shifting towards the MEAN: ζ[m0,n0] = − Δm0,n0,
3. Shifting away from the MEAN, to the MIN/MAX:
ζ[m0,n0] = MAX(Ωm0,n0)−g[m0,n0],if Δm0,n0 > 0
MIN(Ωm0,n0) −g[m0, n0],ifΔm0,n0 < 0,
0 ,if Δm0,n0 = 0.
2] σr : Is calculated for the set of pixels in the window (2N+1)x(2N+1) of pixels centered at [m0,n0] .
Literature Survey
In 1994, Yang and Huang used a hierarchical knowledge-based method to detect faces .Their system consists of three levels of rules. A multiresolution hierarchy of images is created by averaging and subsampling.
Important features: This method is that a coarse-to-fine or focus-of-attention strategy is used to reduce the required computation.
In 1996, Kjeldsen and Kender defined a color image in HSV color space to separate skin regions from background.
In 1998, Sirohey proposed a localization method to segment a face from a cluttered background for face identification, It uses an edge map, Robert cross edge detector so that only the face contour are preserved.
In 2002,Crowley and Coutaz used a histogram h(r,g) of (r, g) values in normalized RGB color space to obtain the probability of obtaining a particular RGB vector given that the pixel observes skin.
In 2002 ,S. Kannumuri and A.N. Rajagopalan, The algorithm uses color histogram for skin (in the HSV space) in conjunction with edge information to quickly locate faces in a given image.
In 2003,McKenna et al. presented an adaptive color mixture model to track faces under varying illumination conditions
In 2004,Inseong Kim, Joon Hyung Shim, and Jinkyu Yang presents a face detection technique mainly based on the color segmentation in YCbCr, image segmentation and template matching methods.
Conclusion
The Hybrid Methodology used for the detection of faces achieves a high rate of accuracy.
Skin Color segmentation
It shows actual distribution of skin color.
Image segmentation
Threshold : discriminate face edges from other edge lines effectively.
Edge Detection
Roberts cross operator : small-hole removal.