How can i find the area of objects with RGB

Everythime I try find the area my valours is very wrong.This square has a 10000cm^2 and I don't know how this object has the area ocupped

My code:

close all; clear all; clc; %%

%pasta=('F:\LDSC\Eco\Processamento direto\Versao 3\P3-17032017WW');

pasta=('F:\LDSC\Eco\CAMPOS\2CAMPO\eefd\b\croped');

dire = dir(pasta); % nome do arquivo, DSC* abre todos os arquivos que começam com esse nome

tt = length (dire); % conta quantidade de arquivos semelhantes

w=1;

areareal=zeros(tt-2,100)

c=1;

   for i=3:tt

       arq = fullfile([dire(i).name]); % le 1 arquivo

      %Algoritmo pra aplicar o threeshold

y=imread(arq);

%Não esquece de chamar a imagem com o nome e extensão certa. %bw=rgb2gray(y);% FUNÇÃO TRANSFORMA IMAGEM EM CINZA %.90 foi o melhor numero para aplicar o filtro e nao perder qualidade imwrite(y,'foto.JPG') original = imread('foto.JPG'); pt=size(original);%pixelxpixel pixeltotal= pt(1,1)*pt(1,2);

imshow(original);

x = original(:,:,1) < 90;

imshow(x);

[B,L] = bwboundaries(x, 'noholes');

stats = regionprops(L, 'Area');

qtd_x = sum([stats.Area] > 1);

imshow(original);

hold on

for k = 1:length(B)

    area = stats(k).Area*10000/(pixeltotal-stats(1).Area);

    %*0.022469 Valor de pixel calculado 

    if area  > 1
        %Valor pra ser lido na AREAREAL
        areareal(w,c)=area;
        c=c+1;

    end
%Aprimoramento do código
%'Abrir Txt com valores de area '
%Imprimir Foto X - Valor de área

    if area > 1
        boundary = B{k};
        plot(boundary(:,2), boundary(:,1), 'black', 'LineWidth', 2);
        text(boundary(1,2), boundary(1,1), sprintf('%.3f',area),...
            'Color', 'black',...
            'FontSize', 12,...
            'FontWeight', 'bold',...
            'BackgroundColor', 'white');

    end

end

hold on

% % aqui vai a rotina de préprocessamento

 w=w+1;

 c=1;

   save

%

 end

Answer 1

[Image Analyst]

To convert from pixels to square cm you need to do a spatial calibration. Basically you have to somehow identify something in the image of known dimensions and then get a "cm per pixel" factor that you can multiply distances by to convert them from pixels to cm, and multiply by the square of that to convert an area in pixels to square cm.

See my attached demo.

[Werner Barros]

First thanks so muchhhhh!

This routine is perfect to me but how can I find another squares?Is there the way to automate? Because i have sometime 50,100 photos.

Thanks againnnn

[Image Analyst]

See the FAQ: http://matlab.wikia.com/wiki/FAQ#How_can_I_process_a_sequence_of_files.3F

or use MAGIC: http://www.mathworks.com/matlabcentral/fileexchange/24224

[Werner Barros]

Image Analyst, you are great guy, thanks for help me, but I have a big problem, my final result is not right, what can i do?

Sorry preach you, but I need this so much to my presentation

[Image Analyst]

I don't know what you did. What you should do is to threshold the image for dark things, then label and call regionprops. Something like

binaryImage = grayImage < 128;
binaryImage = imclearborder(binaryImage);
[labeledImage, numRegions] = bwlabel(binaryImage);
props = regionprops(labeledImage);

See my Image Segmentation Tutorial for a full demo. http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862&sort=downloads_desc