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.