Face Detection

1
Face Detection

• EE368 Final Project
• Spring 2003

- Group 6 - Anthony Guetta Michael Pare Sriram
Rajagopal
2
Overview

• Problem Identification
• Methods Adopted
• Color Segmentation
• Morphological Processing
• Template Matching
• EigenFaces
• Gender Classification

3
Color Segmentation

• Use the color information
• Two approaches
• Global threshold in HSV and YCbCr space using set
  of linear equations. Lot of overlap exists

(a)
(b)
Clustering in (a) YCbCr and (b) V vs. H space.
Red is non-face and blue is face data
4
Result of color segmentation using Global
thresholding
5
Overlap exists in RGB space also
Sample Blue vs Green plot for face (blue) and
non-face (red) data.

• Second approach involves RGB vector quantization
  (Linde, Buzo, Gray)
• Use RGB as a 3-D vector and quantize the RGB
  space for the face and non-face regions

6

• Results from initial quantization
• Common problems identified

7

• Better Code book developed
• Problem areas broken up

8

• Initial step of open and close performed to fill
  holes in faces
• Elongated objects removed by check on aspect
  ratio and small areas discarded

9
Morphological Processing

• Segmented and processed Image consists of all
  skin regions (face, arms and fists)
• Need to identify centers of all objects,
  including individual faces among connected faces
• Repeated EROSION is done with specific
  structuring element

10

• Previous state stored to identify new regions
  when split occurs

Superimposed mask image with eroded regions for
estimate of centroids
11
Template Matching

• Data set has 145 male and 19 female faces
• Need to identify region around estimated
  centroids as face or non-face
• Multi-resolution was attempted. But distortion
  from neighboring faces gives false values
• Smaller template has better result for all face
  shapes
• Template used is the mean face of 50x50 pixels

Mean Face used for template matching
12

• Illumination problem identified
• Top has low lighting, lower part is brighter
• Left and right edges of images do not have people
• 2-D weighting function for correlation values
  applied

2-D weighting function
Sample correlation result
13
Result from template matching and thresholding.
Rejected - Red x. Detected Faces Green x
14
EigenFace based detection

• Decompose faces into set of basis images
• Different methods of candidate face extraction
  from image

EigenFaces
(b)
(a)
Candidate face extraction (a) Conservative (b)
multi-resolution with side distortion
15
Sample result of eigenface. Red is from
morphological processing and green O is from
eigenfaces
16

• Minimum Distance between vector of coefficients
  to that of the face dataset was the metric.
• It depends very much on spatial similarity to
  trained dataset
• Slight changes give incorrect results
• Hence, only template matching was used

17
Gender classification

• Eigenfaces and template matching for specific
  face features do not yield good results
• Other features for specific females were used
  the headband
• Template matching was performed for it
• Conservative estimate was done to prevent falsely
  identifying males as a female

The headband template
18
Table of results for training images
Approx. 95 accuracy with about 75 seconds runtime
19
Training 1
20
Training 2
21
Training 3
22
Training 4
23
Training 5
24
Training 6
25
Training 7
26
Conclusion

• RGB Vector Quantization gave excellent
  segmentation
• Morphological processing gave good estimate of
  centroids
• Template matching with illumination correction
  gave near perfect results
• Specific female was identified with headband

27
Future Considerations

• Edge detection to better separate the connected
  faces
• Preprocess the image in HSV space before codebook
  comparison to improve runtime
• Improve rejection of highly correlated non-face
  objects

28
Thank You
Questions ?
29
(No Transcript)
30
(No Transcript)