how to display and store all the data that has been processed in the form of a matrix in neural network?

Hi,

I have a problem in my program below. there is an error in the matrix appearance. What should I do to keep all the data in a matrix form below so that the neural network can be processed?

   ??? Error using ==> minus
  Matrix dimensions must agree.

Error in ==> prestd at 90
  tn = (t-meant0*oneQ)./(stdt0*oneQ);

Error in ==> cobaaja at 71
[pn,meanp,stdp,tn,meant,stdt] = prestd(P,T);

my program:

clc;
clear all;

% You can customize and fix initial directory paths

DatabasePath = uigetdir('Pilih Database' );
prompt = {'Pilih image yang akan diproses ( 1 to 20):'};
dlg_title = 'cobacoba';
num_lines= 1;
def = {'1'};
cobacoba  = inputdlg(prompt,dlg_title,num_lines,def);
cobacoba = strcat(DatabasePath,'\',char(cobacoba),'.jpg');
I = imread(cobacoba); 
imshow (I);

%untuk mengubah gambar ke dalam bentuk nilai biner

gray=rgb2gray(I);
thresh=graythresh(gray);
imbw=im2bw(gray,thresh);

   %histogram

red=I(:,:,1); %memanggil matriks gambar yang hanya berisi piksel warna merah
green= I(:,:,2);% memanggil matriks gambar yang hanya berisi piksel warna hijau
blue=I(:,:,3); %memanggil matriks gambar yang hanya berisi piksel warna biru
merahgray2=0.3*red+0.5*green+0.2*blue ;
imhist(red);
imhist(green);
imhist(blue);
imhist(gray);

%menampilkan nilai RGB
R = I(:,:,1);
G = I(:,:,2);
B = I(:,:,3);

%menampilkan sum dari RGB

sumR = sum(sum(R));
sumG = sum(sum(G));
sumB = sum(sum(B));
n= (sumR+sumG+sumB);

% Normalisasi warna pada citra

NormalizedRed = sumR * sqrt ( sumR^2+ sumG^2 + sumB^2);
NormalizedGreen = sumG * sqrt ( sumR^2+ sumG^2 + sumB^2);
NormalizedBlue = sumB * sqrt ( sumR^2+ sumG^2 + sumB^2);

% Mencari mean 

MeanR = 1/n*sumR*1.0e+005
MeanG = 1/n*sumG*1.0e+005
MeanB = 1/n*sumB*1.0e+005

% Mencari standar deviasi

sR=(1/n*(sumR-MeanR))^1/2*1.0e+005
sG=(1/n*(sumG-MeanG))^1/2*1.0e+005
sB=(1/n*(sumB-MeanB))^1/2*1.0e+005

%Mencari kurtosis

kR=( 1/n*((sumR-MeanR)^4))/(1/n*((sumR-MeanR)^2))^2
kG=( 1/n*((sumG-MeanG)^4))/(1/n*((sumG-MeanG)^2))^2
kB=( 1/n*((sumB-MeanB)^4))/(1/n*((sumB-MeanB)^2))^2

% Mencari Skewness

SkR = ( 1/n*((sumR-MeanR)^3))/sqrt(1/n*((sumR-MeanR)^2))^3
SkG = ( 1/n*((sumG-MeanG)^3))/sqrt(1/n*((sumG-MeanG)^2))^3
SkB = ( 1/n*((sumB-MeanB)^3))/sqrt(1/n*((sumB-MeanB)^2))^3

% Data Input & Target

P=[MeanR MeanG MeanB sR sG sB kR kG kB SkR SkG SkB]
T=[-1 0 1]

% Preprocessing

[pn,meanp,stdp,tn,meant,stdt] = prestd(P,T);

% Membangun jaringan dengan propagasi umpan balik berbasis jaringan syaraf tiruan

net = newff(minmax(pn),[10 5 1],{'tansig' 'logsig' 'purelin'},'traingdm')

% melihat bobot awal jaringan

BobotAwal_Input = net.IW{1,1};
BobotAwal_Bias_Input = net.b{1,1};
BobotAwal_Lapisan1 = net.LW {2,1};
BobotAwal_Bias_Lapisan = net.b {2,1};

% Set parameter pelatihan

net.trainParam.epochs = 1000;
net.trainParam.goal = 1e-04;
net.trainParam.lr = 0.6;
net.trainParam.showCommandLine = true;
net.trainParam.show = 100;
net.trainParam.mc = 0.7;

% pelatihan

net = train(net,pn,tn);pause

% melihat bobot akhir jaringan

BobotAkhir_Input = net.IW{1,1};
BobotAkhir_Bias_Input = net.b{1,1};
BobotAkhir_Lapisan1 = net.LW {2,1};
BobotAkhir_Bias_Lapisan = net.b {2,1};

%simulasi terhadap data pelatihan

an = sim(net,pn);
a = poststd (an,meant,stdt);
H = [(1:size(P,2))' T' a' (T'-a')];
sprintf('%2d %9.2f %7.2f %5.2f\n',H')

%Evaluasi output jaringan (data pelatihan dengan target)

[m1,a1,r1] = postreg (a,T);
pause
plot (1:size(P,2),T,'bo',1:size(P,2),a','r*');
title('Hasil pengujian dengan data pelatihan Target (o), Output (*)');
xlabel('Data ke-');
ylabel('Target/Output');
pause

In addition, please guide me, sincerely Yours Een Qurota Aini

[Greg Heath]

Reformat and resubmit. Search using

tutorial format

[Greg Heath]

%Error in ==> prestd at 90 tn = (t-meant0*oneQ)./(stdt0*oneQ);

%Error in ==> cobaaja at 71 [pn,meanp,stdp,tn,meant,stdt] = prestd(P,T);

Dimension check

whos t meant0 oneQ stdt0 P T

[Een Qurota Aini]

thanks for the help. well I'll check back. I want to ask if the matlab source code for storing the results like? suppose we have a program at the top, once in run only one outcome, and how to store the results are then combined with the results of other data into one matrix?

[Greg Heath]

Save the output of each run into a separate matrix.

Write a program that loads the saved matrices and concatenates them

[Een Qurota Aini]

Thanks so much mr. Greg. What should I do in my thesis research, I classify here where the leaf is damaged, severely damaged leaves and healthy leaves using a neural network. My initial first process uses image processing with 4 input features no mean, standard deviation, skewness, kurtosis. well as entry into the neural network mw I'm confused. do i have to create the database or how, please help me? in this case the process is the same as the finger print right? but for programs like what yah? program made always error

[Greg Heath]

I don't understand. Please rewrite carefully with more detail.

[Image Analyst]

My guess is that you do not have the correct features unless you have some overly simplistic situations like you have the leaves laid out flat with uniform lighting where any observer could instantly tell a healthy leaf from a diseased one. In a real world situation like you're driving past the orchard or crop and taking video of the leaves, these features most likely won't be enough. If the features don't adequately describe then a neural network won't rescue you. For example if you want to predict a person's gender, and you're putting in height, weight, and hair color, then you're out of luck. Garbage in, garbage out. However if you have a completely artificial and controlled situation like the leaves laid out flat on a brightly lit table, then those histogram measures might possibly be able to discriminate the amount of disease.

Were you going to invent and add anything unique to improve the classification, or were you simply going to apply already discovered and known methods to the images?

[Een Qurota Aini]

thank you very much for the reply. I've tried several programs. Well now I am confused on the database for the value of the feature on the leaves. As in the fingerprint program with neural network, the study was also no image database containing finger. Well, I am confused. 1. When the program start to figure out the histogram data leaves no feature value in the command window 2. When data from the value of the features I put my gui uitable program confused what should I enter or source code to connect the data values to be entered into uitable feature in gui? I was confused about entering data values in the command window feature leaves that go into uitable and can be saved in the uitable as inputs for training and testing the neural network Please help me for source code add selected image to database or input data fitur daun to uitable gui :'(

[Greg Heath]

What is

size(P)

size(T)

oneQ

[Een Qurota Aini]

P=[MeanR MeanG MeanB sR sG sB kR kG kB SkR SkG SkB];

T=[-1 0 1];

sR=Deviation Standart from RGB Value R

http://id.scribd.com/doc/110731967/How-to-Display-Image-Data-Into-Table

sG=Deviation Standart from RGB Value G

sB=Deviation Standart from RGB Value B

kR=Kurtosis from RGB value R

kG=Kurtosis from RGB value G

kB=Kurtosis from RGB value B

SkR=Skewness from RGB value R

SkG=Skewness from RGB value G

SkB-Skewness from RGB value B

What is the mean oneQ??

Dislay image data into table