Erreur dans le code de FisherFace

**slaima15** · 27/11/2011, 12h56

Salut, j'essaye d'applique le code de FisherFace à ma base de visage, mon problème c'est que le taux de reconnaissance obtenu est très faible 2%. Donc, je pense qu'il ya un erreur au niveau du code.
Vous trouvez ci dessus les fonctions que j'utilise:

Code :

Sélectionner tout - Visualiser dans une fenêtre à part

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
function T = CreateDatabase(TrainDatabasePath)
% Align a set of face images (the training set T1, T2, ... , TM )
%
% Description: This function reshapes all 2D images of the training database
% into 1D column vectors. Then, it puts these 1D column vectors in a row to 
% construct 2D matrix 'T'. Each column of 'T' is a training image, which has been reshaped into a 1D vector.
% Also, P is the total number of MxN training images and C is the number of
% classes.
%  
% 
% Argument:     TrainDatabasePath      - Path of the training database
%
% Returns:      T                      - A 2D matrix, containing all 1D image vectors.
%                                        The length of 1D column vectors is MN and 'T' will be a MNxP 2D matrix.
%
% See also: STRCMP, STRCAT, RESHAPE
 
% Original version by Amir Hossein Omidvarnia, October 2007
%                     Email: aomidvar@ece.ut.ac.ir                  
 
%%%%%%%%%%%%%%%%%%%%%%%% File management
TrainFiles = dir(TrainDatabasePath);
Train_Number = 0;
 
for i = 1:size(TrainFiles,1)
    if not(strcmp(TrainFiles(i).name,'.')|strcmp(TrainFiles(i).name,'..')|strcmp(TrainFiles(i).name,'Thumbs.db'))
        Train_Number = Train_Number + 1; % Number of all images in the training database
    end
end
 
%%%%%%%%%%%%%%%%%%%%%%%% Construction of 2D matrix from 1D image vectors
T = [];
for i = 1 : Train_Number
 
    % I have chosen the name of each image in databases as a corresponding
    % number. However, it is not mandatory!
    str = int2str(i);
    str = strcat('\',str,'.jpg');
    str = strcat(TrainDatabasePath,str);
 
    img = imread(str);
    img = rgb2gray(img);
 
    [irow icol] = size(img);
 
    temp = reshape(img',irow*icol,1);   % Reshaping 2D images into 1D image vectors
    T = [T temp]; % 'T' grows after each turn                    
end
 
T = double(T);

Code :

Sélectionner tout - Visualiser dans une fenêtre à part

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
function [m_database V_PCA V_Fisher ProjectedImages_Fisher] = FisherfaceCore(T)
% Use Principle Component Analysis (PCA) and Fisher Linear Discriminant (FLD) to determine the most 
% discriminating features between images of faces.
%
% Description: This function gets a 2D matrix, containing all training image vectors
% and returns 4 outputs which are extracted from training database.
% Suppose Ti is a training image, which has been reshaped into a 1D vector.
% Also, P is the total number of MxN training images and C is the number of
% classes. At first, centered Ti is mapped onto a (P-C) linear subspace by V_PCA
% transfer matrix: Zi = V_PCA * (Ti - m_database).
% Then, Zi is converted to Yi by projecting onto a (C-1) linear subspace, so that 
% images of the same class (or person) move closer together and images of difference 
% classes move further apart: Yi = V_Fisher' * Zi = V_Fisher' * V_PCA' * (Ti - m_database)
%
% Argument:      T                      - (M*NxP) A 2D matrix, containing all 1D image vectors.
%                                         All of 1D column vectors have the same length of M*N 
%                                         and 'T' will be a MNxP 2D matrix.
% 
% Returns:       m_database             - (M*Nx1) Mean of the training database
%                V_PCA                  - (M*Nx(P-C)) Eigen vectors of the covariance matrix of the 
%                                         training database
%                V_Fisher               - ((P-C)x(C-1)) Largest (C-1) eigen vectors of matrix J = inv(Sw) * Sb
%                ProjectedImages_Fisher - ((C-1)xP) Training images, which are projected onto Fisher linear space
%
% See also: EIG
 
% Original version by Amir Hossein Omidvarnia, October 2007
%                     Email: aomidvar@ece.ut.ac.ir                  
 
 
Class_number = 48%( size(T,2) )/4 % Number of classes (or persons)
Class_population = 6%2; % Number of images in each class
P = 288%Class_population * Class_number; % Total number of training images
 
%%%%%%%%%%%%%%%%%%%%%%%% calculating the mean image 
m_database = mean(T,2); 
 
%%%%%%%%%%%%%%%%%%%%%%%% Calculating the deviation of each image from mean image
A = T - repmat(m_database,1,P);
 
%%%%%%%%%%%%%%%%%%%%%%%% Snapshot method of Eigenface algorithm
L = A'*A; % L is the surrogate of covariance matrix C=A*A'.
[V D] = eig(L); % Diagonal elements of D are the eigenvalues for both L=A'*A and C=A*A'.
 
%%%%%%%%%%%%%%%%%%%%%%%% Sorting and eliminating small eigenvalues
L_eig_vec = [];
for i = 1 : P-Class_number 
    L_eig_vec = [L_eig_vec V(:,i)];
end
 
%%%%%%%%%%%%%%%%%%%%%%%% Calculating the eigenvectors of covariance matrix 'C'
V_PCA = A * L_eig_vec; % A: centered image vectors
 
%%%%%%%%%%%%%%%%%%%%%%%% Projecting centered image vectors onto eigenspace
% Zi = V_PCA' * (Ti-m_database)
ProjectedImages_PCA = [];
for i = 1 : P
    temp = V_PCA'*A(:,i);
    ProjectedImages_PCA = [ProjectedImages_PCA temp]; 
end
 
%%%%%%%%%%%%%%%%%%%%%%%% Calculating the mean of each class in eigenspace
m_PCA = mean(ProjectedImages_PCA,2); % Total mean in eigenspace
m = zeros(P-Class_number,Class_number); 
Sw = zeros(P-Class_number,P-Class_number); % Initialization os Within Scatter Matrix
Sb = zeros(P-Class_number,P-Class_number); % Initialization of Between Scatter Matrix
 
for i = 1 : Class_number
    m(:,i) = mean( ( ProjectedImages_PCA(:,((i-1)*Class_population+1):i*Class_population) ), 2 )';    
 
    S  = zeros(P-Class_number,P-Class_number); 
    for j = ( (i-1)*Class_population+1 ) : ( i*Class_population )
        S = S + (ProjectedImages_PCA(:,j)-m(:,i))*(ProjectedImages_PCA(:,j)-m(:,i))';
    end
 
    Sw = Sw + S; % Within Scatter Matrix
    Sb = Sb + (m(:,i)-m_PCA) * (m(:,i)-m_PCA)'; % Between Scatter Matrix
end
 
%%%%%%%%%%%%%%%%%%%%%%%% Calculating Fisher discriminant basis's
% We want to maximise the Between Scatter Matrix, while minimising the
% Within Scatter Matrix. Thus, a cost function J is defined, so that this condition is satisfied.
[J_eig_vec, J_eig_val] = eig(Sb,Sw); % Cost function J = inv(Sw) * Sb
J_eig_vec = fliplr(J_eig_vec);
 
%%%%%%%%%%%%%%%%%%%%%%%% Eliminating zero eigens and sorting in descend order
for i = 1 : Class_number-1 
    V_Fisher(:,i) = J_eig_vec(:,i); % Largest (C-1) eigen vectors of matrix J
end
 
%%%%%%%%%%%%%%%%%%%%%%%% Projecting images onto Fisher linear space
% Yi = V_Fisher' * V_PCA' * (Ti - m_database) 
for i = 1 : Class_number*Class_population
    ProjectedImages_Fisher(:,i) = V_Fisher' * ProjectedImages_PCA(:,i);
end

Code :

Sélectionner tout - Visualiser dans une fenêtre à part

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
function OutputName = Recognition(TestImage, m_database, V_PCA, V_Fisher, ProjectedImages_Fisher)
% Recognizing step....
%
% Description: This function compares two faces by projecting the images into facespace and 
% measuring the Euclidean distance between them.
%
% Argument:      TestImage              - Path of the input test image
%
%                m_database             - (M*Nx1) Mean of the training database
%                                         database, which is output of 'EigenfaceCore' function.
%
%                V_PCA                  - (M*Nx(P-1)) Eigen vectors of the covariance matrix of 
%                                         the training database
 
%                V_Fisher               - ((P-1)x(C-1)) Largest (C-1) eigen vectors of matrix J = inv(Sw) * Sb
 
%                ProjectedImages_Fisher - ((C-1)xP) Training images, which
%                                         are projected onto Fisher linear space
% 
% Returns:       OutputName             - Name of the recognized image in the training database.
%
% See also: RESHAPE, STRCAT
 
% Original version by Amir Hossein Omidvarnia, October 2007
%                     Email: aomidvar@ece.ut.ac.ir                  
 
Train_Number = size(ProjectedImages_Fisher,2)
%%%%%%%%%%%%%%%%%%%%%%%% Extracting the FLD features from test image
%InputImage = imread(TestImage);
temp = TestImage(:,:,1);
%tic;
 
[irow icol] = size(temp);
InImage = reshape(temp',irow*icol,1);
Difference = double(InImage)-m_database; % Centered test image
ProjectedTestImage = V_Fisher' * V_PCA' * Difference; % Test image feature vector
 
%%%%%%%%%%%%%%%%%%%%%%%% Calculating Euclidean distances 
% Euclidean distances between the projected test image and the projection
% of all centered training images are calculated. Test image is
% supposed to have minimum distance with its corresponding image in the
% training database.
Euc_dist = [];
for i = 1 : Train_Number
    q = ProjectedImages_Fisher(:,i);
    temp = ( norm( ProjectedTestImage - q ) )^2;
    Euc_dist = [Euc_dist temp];
end
 
[Euc_dist_min , Recognized_index] = min(Euc_dist);
%toc
OutputName = Recognized_index;
%OutputName = strcat(int2str(Recognized_index),'.bmp');

et voila la fonction que j'utilise pour faire l'appel de ces fonctions:

Code :

Sélectionner tout - Visualiser dans une fenêtre à part

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
 
 
clear all
clc
close all
 
TrainDatabasePath = uigetdir(strcat(matlabroot,'\work'), 'Select training database path' );
TestDatabasePath = uigetdir(strcat(matlabroot,'\work'), 'Select test database path');
T = CreateDatabase(TrainDatabasePath);
[m V_PCA V_Fisher ProjectedImages_Fisher] = FisherfaceCore(T);
T=[];
ix1=[]
froot = './RenommerImageTest';
list = dir(sprintf('%s\\*.jpg', froot));
frootProbe = './ImageTest';
listProbe = dir(sprintf('%s\\*.jpg', frootProbe));
frootGalerie = './ImageGalerie';
listGalerie = dir(sprintf('%s\\*.jpg', frootGalerie));
for index = 1:length(list)
   chemain_image=sprintf('%s\\%s', froot, list(index).name);
   TestImage = imread(chemain_image);
 
 
OutputName = Recognition(TestImage, m, V_PCA, V_Fisher, ProjectedImages_Fisher);
 
SortieReconnaissance=OutputName
 
   chemain_imageProbe=sprintf('%s\\%s', frootProbe, listProbe(index).name);
   ProbeImage = imread(chemain_imageProbe); 
   NomProbeImage= listProbe(index).name
 
fg1=[];
for ml=1:4
fg1=[fg1 NomProbeImage(ml)];
end
fg1=fg1
   chemain_imageGaleie=sprintf('%s\\%s', frootGalerie, listGalerie(OutputName).name);
   GalerieImage = imread(chemain_imageGaleie);
   NomGalerieImage=listGalerie(OutputName).name
fp1=[];
for ml=1:4
fp1=[fp1 NomGalerieImage(ml)];
end
fp1=fp1
 if(strcmp(fp1,fg1))
      c=[1 0]
  else
      c=[0 1]
  end;
  T=[T;c];
end
Taux_False_Classification= False_Erreur_Rate(T)
Taux_recognitionGauche=1-Taux_False_Classification

Pouvez-vous m'aidez à corriger ce code car je me sens que le résultat obtenu n'est pas logique??

**Jean Dumoncel** · 27/11/2011, 15h09

Bonjour,

Première remarque : pense à indenter ton code, dans l'éditeur de MATLAB, sélectionne tout ton code (ctrl a) et effectue une indentation (ctrl i). Cela facilitera la lecture des autres membres qui liront ton code.

peux-tu expliquer comment tu calcules le taux de reconnaissance? Notamment cette partie du code :

Code :

Sélectionner tout - Visualiser dans une fenêtre à part

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
   SortieReconnaissance=OutputName
 
  chemain_imageProbe=sprintf('%s\\%s', frootProbe, listProbe(index).name);
   ProbeImage = imread(chemain_imageProbe); 
   NomProbeImage= listProbe(index).name
 
fg1=[];
for ml=1:4
fg1=[fg1 NomProbeImage(ml)];
end
fg1=fg1
   chemain_imageGaleie=sprintf('%s\\%s', frootGalerie, listGalerie(OutputName).name);
   GalerieImage = imread(chemain_imageGaleie);
   NomGalerieImage=listGalerie(OutputName).name
fp1=[];
for ml=1:4
fp1=[fp1 NomGalerieImage(ml)];
end
fp1=fp1
 if(strcmp(fp1,fg1))
      c=[1 0]
  else
      c=[0 1]
  end;
  T=[T;c];
end
Taux_False_Classification= False_Erreur_Rate(T)
Taux_recognitionGauche=1-Taux_False_Classification

**slaima15** · 27/11/2011, 18h25

merci pour la première remarque.

Pour cette partie de code:
Pour chaque image de test

Code :

Sélectionner tout - Visualiser dans une fenêtre à part

for index = 1:length(list)

je calcule la plus proche image qui possède la plus petite distance euclidienne

Code :

Sélectionner tout - Visualiser dans une fenêtre à part

OutputName = Recognition(TestImage, m, V_PCA, V_Fisher, ProjectedImages_Fisher);

les images sont identifiées par des numéros, donc avec cet OutputName je récupère le numéro de l'image dans la galerie. Cet numéro nous sert à récupérer le nom réel de la personne dans la base ImageGalerie en récupérant les 4 premières caractères. De même avec index on peut récupérer le nom réel de la personne dans la base ImageTest en récupérant les 4 premières caractères.
Si ces deux nom sont identique sont c'est la même personne sinon c'est un autre personne

Code :

Sélectionner tout - Visualiser dans une fenêtre à part

1
2
3
4
5
6
 if(strcmp(fp1,fg1))
      c=[1 0]
  else
      c=[0 1]
  end;
  T=[T;c];

On conserve les résultats des personnes identifiés dans le tableau T:, puis je calcule le taux de fausse identification:

Code :

Sélectionner tout - Visualiser dans une fenêtre à part

1
2
3
function taux=False_Erreur_Rate(T)
nbre=sum(T(:,2))
taux=(nbre/length(T));

ensuite :

Code :

Sélectionner tout - Visualiser dans une fenêtre à part

Taux_recognitionGauche=1-Taux_False_Classification

**slaima15** · 28/11/2011, 18h16

ya-il quelqu'un qui peut m'aider, je suis vraiment bloqué??

Erreur dans le code de FisherFace

MATLAB

Discussions similaires

Partager

Partager