principal components analysis
DESCRIPTION
As applied to face recognition. Principal Components Analysis. video. Face Recognition. Detection vs. Recognition. Face Recognition. Identification vs. Verification. Face Recognition. Components: Face Detection Face Alignment Feature Extraction Matching. Face Recognition. Components: - PowerPoint PPT PresentationTRANSCRIPT
As applied to face recognition
Detection vs. Recognition
Identification vs. Verification
Components: Face Detection Face Alignment Feature Extraction Matching
Components: Face Detection Face Alignment Feature Extraction Matching
Dimensionality Reduction
“Eigenface” analysis
Unordered Observations
LightTemp.
2.5 2.4
0.5 0.7
2.2 2.9
1.9 2.2
3.1 3
2.3 2.7
2 1.6
1 1.1
1.5 1.6
1.1 0.9
Turns 4096 dimensions -> 40 or less dimensions
1.81 1.91
2.5 2.4
0.5 0.7
2.2 2.9
1.9 2.2
3.1 3
2.3 2.7
2 1.6
1 1.1
1.5 1.6
1.1 0.9
1.81 1.91
2.5 2.4
0.5 0.7
2.2 2.9
1.9 2.2
3.1 3
2.3 2.7
2 1.6
1 1.1
1.5 1.6
1.1 0.9
0.69 0.49
-1.31 -1.21
0.39 0.99
0.09 0.29
1.29 1.09
0.49 0.79
0.19 -0.31
-0.81 -0.81
-0.31 -0.31
-0.71 -1.01
0.69 0.49
-1.31 -1.21
0.39 0.99
0.09 0.29
1.29 1.09
0.49 0.79
0.19 -0.31
-0.81 -0.81
-0.31 -0.31
-0.71 -1.01
.69 -1.31
.39 .09 1.29
.49 .19 -.81 -.31 -.71
.49 -1.21
.99 .29 1.09
.79 -.31 -.81 -.31 -1.01
.69 -1.31
.39 .09 1.29
.49 .19 -.81 -.31 -.71
.49 -1.21
.99 .29 1.09
.79 -.31 -.81 -.31 -1.01
0.61655556 0.61544444
0.61544444 0.71655556
0.0490834 1.28402771
-.73517866 -0.6778734
0.6778734 -0.73517866
EigenvaluesEigenvector 1 Eigenvector 2
“Characteristic”
“Characteristic”Vector characterizing a feature of
the matrix
“Characteristic”Vector characterizing a feature of
the matrixEigenvalue = strength
-.73517866 -0.6778734
0.6778734 -0.73517866
Eigenvalues
Eigenvector 1 Eigenvector 2
0.0490834 1.28402771
-.73517866 -0.6778734
0.6778734 -0.73517866
-.73517866 0.6778734
-0.6778734 -0.73517866
.69 -1.31
.39 .09 1.29
.49 .19 -.81 -.31 -.71
.49 -1.21
.99 .29 1.09
.79 -.31 -.81 -.31 -1.01
-.828
1.78
-.992
-.27
-1.67
-.912
.099
1.144
.438
1.22
2.5 2.4
0.5 0.7
2.2 2.9
1.9 2.2
3.1 3
2.3 2.7
2 1.6
1 1.1
1.5 1.6
1.1 0.9
[0,0,0,127, 55, 234, 255, 123, 98… n] n = width * height
Image1
Image2
Image3
Image4
0 0 0 127
55 234
255
123
98 65
23 15 67 125
76 209
132
64 92 22
76 234
200
98 11o 85 145
97 44 32
209
53 99 198
39 201
38 220
77 92
Average
0 0 0 127
55 234
255
123
98 65
23 15 67 125
76 209
132
64 92 22
76 234
200
98 11o
85 145
97 44 32
209
53 99 198
39 201
38 220
77 92
-77 -75.5 -91.5 -10 -1.67 51.75 112.5 -3 20.25 12.25
-54 -60.5 -24.5 -12 19.3 26.75 -10.5 -62 14.25 -30.75
-1 158.5 108.5 -39 53.3 -97.25 2.5 -29 -33.75 -20.75
132 -22.5 7.5 61 -17.67 18.75 -104.5 94 -0.75 39.25
77 75.5 91.5 137 56.67 182.3 142.5 126 77.75 52.75
Eigenvalues
Eigenvectors
.000064 50.97 84.828 173.8 213.018
-.24 -.05 -.17 .13 .33
-.24 -.001 -.034 .462 .317
-.24 -.367 -.1 .006 .134
-.24 -.222 .412 .082 -.308
-.24 .0008 .048 -.057 .192
Principal component
Animation of reconstruction
.5 .2 .1
.03 .005
Demo