How do I measure a distance in real world with a picture in matlab?

first i have to convert pixels to cm , than i have the length between to points in cm . (in the picture) but in the real world , how many cm between the same point ? I do not know !!!

Answer 1

[Image Analyst]

Sara, you need to spatially calibrate. Run my attached demo and it will show you how.

[EDIT] Attached updated spatial_calibration_demo.m

[Image Analyst]

I don't know how, but somehow our conversation of 14 responses in an "Answer" posted by sara got deleted. I'm restoring it here. My and Matt J's comments are in bold, hers are not.

I run it , but it dose not show the real distance , for example 2 points in moon !!! in real

Worked perfectly fine for me. When you did the first part, what did you say the units were, and then what did you say the distance was? Then, when you selected the measure distance button, are you sure you followed the directions? If so, then run psr.exe (if you have windows) and upload it so I can see screenshot by screenshot exactly what you did.

when i run it , it ask me which demo image ? for example i choose moon !than it ask me to select my start and end point , so it ssk me enter real word unit and distance in dose unit ? it means I have measure between my 2 points by my self ? (by meter) in the case of cm ? and put in message box , ????

i want to know a distance in these unit in real !!! not in the picture !!!

is it possible to know with these codes ???

Matt J added: You have to know in advance the real-world physical distance (centimeters, meters, whatever...) between two sample points in the image in order to do what you are trying to do. Otherwise, there is no way to deduce the conversion formula relating distance in pixels to distance in centimeters.

I ADDED Do you think the program knows if the picture of a moon is the actual moon taken through a telescope, and so the units should be in km, or if it's a photo of a poster on my office wall and so the units should be in cm? No, it doesn't know that so you have to tell it.

how i can show u my screenshot ? tel me know plz run psr.exe what is this ? how i can find and run it ? have i download it ? Image Analyst i really confused , if u can help me with skype it is better :(((((((

matt and Image Analyst thank you , and what do u recommend for understand length of the each car at street , the car is moving and the camera and laser is fix

You must have something in the photo of known length, like a car that is 6 m long or whatever.

To run psr to do a screencam capture, if you have Windows, click the start button and type in psr.exe in the "Search programs and files" edit box. Then tell it to start recording, then do stuff, then stop the recorder and save it to a zip file. Then upload the zip file here with the paper clip icon.

i have windows vista , i do not have this exe file :( have i download it ? Image Analyst i really confused , if u can help me with skype it is better :(((((((

You left click the one end point, and right click the other point, like you did. Then it asks you for a number and some characters, like 3474 for the number and km for the unit. Then you click on a button to measure some distance so you do the same thing - left click and right click. Then it will tell you the length in km .

3474 ? i have to measure in picture and write it ? or i can put any numbers ?

No, sara. I knew, from my Google research, that the real diameter of the moon was 3474 km. So I typed that in. If it was a poster on my wall, I would have typed 30 inches into the dialog box when it asked me what the real world size was.

ok , but in the case of length of each car on the street , i can not be calculate , measure and write in the dialog box , the real size of each car !

so it dose not work for cares in the street , what do u think ?> I want a code that I can measure the picture of object , and the matlab code tell me the real length of each object

I gave you that. The demo does that.

[Image Analyst]

Another "Answer" from sara that somehow got deleted is reproduced below:

what do u recommend for understand length of the each car at street , the car is moving and the camera and laser is fix

    ANY OTHER RECOMMEND ? IF U HAVE CODES WRITE ME PLZ *

sara, please don't add answers. Add comments. I suggest you go out and stand in the road holding a yard stick or meter stick, or else measure the length of something in the scene, like a stop sign, so you'll know how many pixels correspond to a real world distance.

https://sites.google.com/site/todddanko/home/webcam_laser_ranger , can this page help me? if u have code let me know plz

we have camera at street , so how i can understand how many cm length of each car ? cars moving and the camera is fix also we can add laser too

*let me know the codes

every bodyyyyyyyyyyyyyyyyyyy help me as soon as possible|*

I wish were weren't answering in 5 different places. Anyway, if you already have some instrument that uses known geometry, then I'd ask the manufacturer.

[Image Analyst]

Yet another "Answer" from sara that somehow got deleted is reproduced below:

when i run your attach demo , i have these error too :((((

??? Error using ==> improfile>parse_inputs Requires an image in the current axis.

Error in ==> improfile at 88 [xa,ya,a,n,method,prof,getn,getprof] = parse_inputs(varargin{:});

Error in ==> spatial_calibration_demo>Calibrate at 110 [cx, cy, rgbValues, xi,yi] = improfile(1000);

Error in ==> spatial_calibration_demo at 85 Calibrate(); You right clicked at first. You were supposed to left click and then right click. It tells you the instructions in a message box, and also in the title above the image. Use the attached spatial_calibration_demo.m in blue text below. In there, I detect improper clicking and let you try again instead of throwing an error.

[sara]

??? Undefined function or variable 'ME'.

Error in ==> spatial_calibration_demo>Calibrate at 141 catch ME

Error in ==> spatial_calibration_demo at 85 success = Calibrate();

i did it now , and i have less error ! but these error

can u help me ????????????

[Image Analyst]

You must have change it some how. I just copied and pasted the demo and followed directions and it worked absolutely find with no error. The error you give is on the catch line:

catch ME

Now, the ME does not need to be defined in advance. I know that for a fact. It gets generated when you hit the catch line. Just to be sure I introduced an error into the program so that an error would be thrown and it would land on that line. It did not complain about ME, it complained about the error I created. So what did you change? Did you delete the try line or something? Why did you even hit the catch line at all??? Something is different.

Attached is a screen recording of what I did. What did you do differently from what I did?

[Josh Pham]

Thanks for the demo- it is much helpful! I was wondering, what is the algorithm used in the demo that allows you to measure distance? Is matlab simply taking a known length of pixels and real world length, and using proportionality to measure any other distance? I guess another way of saying this is, even though you know the diameter of the moon , there will be some error as matlab doesn't take into account the curvature of the moon?

[Image Analyst]

It just assumes a flat scene. If you have a tilted/perspective scene, you can use the camera calibration capability in the Computer Vision System Toolbox http://www.mathworks.com/products/computer-vision/features.html#camera-calibration If you have some crazy surface like a mountain range, you'll need some other kind of sensor to get the full topographic information.

[ahmed kat]

Hi image analyst, actually i need ur help i want a matlab program similar to catch ME but with tsukuba or Teddy image also plz if its possible i want the program with no instructions like (click the right or left) i hope u help. i want this so bad

[Walter Roberson]

ahmed kat: "catch" is a MATLAB command; see http://www.mathworks.com/help/matlab/ref/try.html

The construction

try
   Some Action
catch ME
   Some Other Action
end

means that the program is to attempt to execute "Some Action", and if at any point the execution of that results in an error, then the rest of the execution of that action is to be skipped and the code "Some Other Action" is to be executed after setting the variable named "ME" to the MException object that was generated by processing the error that came up.

"catch ME" is not a particular program!

[Image Analyst]

Here's a snippet you might find useful:

try
  % Some code that might generate an error.
catch ME
  errorMessage = sprintf('Error in function %s() at line %d.\n\nError Message:\n%s', ...
    ME.stack(1).name, ME.stack(1).line, ME.message);
  fprintf(1, '%s\n', errorMessage);
  uiwait(warndlg(errorMessage));
end

[Joachim Huet]

I would like to open this topic again, is there a way to do this calibration phase automatically ? Without needing the user to draw a line with known value !?

[Image Analyst]

Of course. For example I automatically calibrate some systems by (1) finding a 4 inch square filter paper, and (2) by using the known dimensions of the X-Rite Color Checker chart. All you need is a known real world distance of some thing , and some way to find that thing in the image and measure the distance in pixels (like by thresholding or whatever...).

Answer 2

[Caroline David]

Can someone show me the code for length measurements of an image in pixels. Most preferred in mm

[Image Analyst]

Did you overlook the demo I posted above? If not, then explain why that is not applicable to what you want to do. Like do you not know the length and need to find it automatically through image segmentation? If so, see my Images Segmentation Tutorial.

[jue xi]

hi how can i adjust this coding so that i can measure the bounding box?

endfolder=('C:\Users\user\MATLAB');
  baseFileName=('img11.jpg');
  fullFileName=fullfile(folder,baseFileName);
  format long g;
  format compact;
  fontSize = 20;

    %IMAGE FILTERING
    img = imread('img11.jpg');
    h=ones(5,5)/25;
    img1=imfilter(img,h);

    %IMAGE SEGMENTATION
    img2=rgb2ycbcr(img1);
    for i=1:size(img2,1)
        for j= 1:size(img2,2)
            cb = img2(i,j,2);
            cr = img2(i,j,3);
            if(~(cr > 132 && cr < 173 && cb > 76 && cb < 126))
                img2(i,j,1)=235;
                img2(i,j,2)=128;
                img2(i,j,3)=128;
            end
        end
    end
    img2=ycbcr2rgb(img2);
    subplot(2,2,1);
    image1=imshow(img2);
    axis on;
    title('Skin Segmentation', 'FontSize', fontSize);
    set(gcf, 'Units', 'Normalized', 'OuterPosition', [0 0 1 1]);
    %SEGMENTED IMAGE TO GRAYIMAGE
    grayImage=rgb2gray(img2);
    subplot(2,2,2);
    image2=imshow(grayImage);
    axis on;
    title('Original Grayscale Image', 'FontSize', fontSize);
    set(gcf, 'Units', 'Normalized', 'OuterPosition', [0 0 1 1]);
    %GRAY TO BINARY IMAGE
    binaryImage = grayImage < 245;
    subplot(2, 2, 3);
    axis on;
    axis image ;
    image3=imshow(binaryImage, []);
    title('Binary Image', 'FontSize', fontSize);

    % Label the image 
    labeledImage = bwlabel(binaryImage);  % label the connected components in an image and assigning each one a unique label

    measurements = regionprops(labeledImage, 'BoundingBox', 'Area');

    for k = 1 : length(measurements)
      thisBB = measurements(k).BoundingBox;
      rectangle('Position', [thisBB(1),thisBB(2),thisBB(3),thisBB(4)],...
      'EdgeColor','r','LineWidth',2 )
    end

[Image Analyst]

It looks like you ARE measuring it.

measurements(k).BoundingBox has everything you need to know about it. What do you think you're not getting???

[jue xi]

tq sir, but

this is what i get from that coding

1.i need to measure length of hand by getting the width from that bounding box. 2. i will first remove the unwanted noise from that shirt 3. then i have only two bounding box from both of the hand 4. how to get the measurement of two bounding box. 5. i want it to display the width as below coding.

if true
  % code
clc;    % Clear the command window.
close all;  % Close all figures (except those of imtool.)
clear;  % Erase all existing variables. Or clearvars if you want.
workspace;  % Make sure the workspace panel is showing.
format short g;
format compact;
fontSize = 25;
%===============================================================================
% Get the name of the image the user wants to use.
baseFileName = 'img1.jpg';
% baseFileName = '1k_100kph_5x20mm.JPG';
% Get the full filename, with path prepended.
folder = []; % Determine where demo folder is (works with all versions).
fullFileName = fullfile(folder, baseFileName);
%===============================================================================
% Read in a demo image.
grayImage = imread(fullFileName);
% Get the dimensions of the image.
% numberOfColorChannels should be = 1 for a gray scale image, and 3 for an RGB color image.
[rows, columns, numberOfColorChannels] = size(grayImage)
if numberOfColorChannels > 1
	% It's not really gray scale like we expected - it's color.
	% Use weighted sum of ALL channels to create a gray scale image.
	grayImage = rgb2gray(grayImage);
	% ALTERNATE METHOD: Convert it to gray scale by taking only the green channel,
	% which in a typical snapshot will be the least noisy channel.
	% grayImage = grayImage(:, :, 2); % Take green channel.
end
% Display the image.
subplot(2, 2, 1);
imshow(grayImage, []);
axis on;
axis image;
caption = sprintf('Original Gray Scale Image');
title(caption, 'FontSize', fontSize, 'Interpreter', 'None');
drawnow;
hp = impixelinfo();
% Set up figure properties:
% Enlarge figure to full screen.
set(gcf, 'Units', 'Normalized', 'OuterPosition', [0, 0, 1, 1]);
% Get rid of tool bar and pulldown menus that are along top of figure.
% set(gcf, 'Toolbar', 'none', 'Menu', 'none');
% Give a name to the title bar.
set(gcf, 'Name', 'Demo by ImageAnalyst', 'NumberTitle', 'Off')
drawnow;
% binarize the image.
binaryImage = imbinarize(grayImage); 
% Make sure there is only one blob
binaryImage = bwareafilt(binaryImage, 1);
% Display the image.
subplot(2, 2, 2);
imshow(binaryImage, []);
axis on;
axis image;
caption = sprintf('Binary Image');
title(caption, 'FontSize', fontSize, 'Interpreter', 'None');
drawnow;
hp = impixelinfo();
% Label the image
labeledImage = bwlabel(binaryImage);

% Make measurements of bounding box props = regionprops(labeledImage, 'BoundingBox'); width = props.BoundingBox(3); height = props.BoundingBox(4); % Display it with the box overlaid on it. subplot(2, 2, 3); imshow(grayImage, []); axis on; axis image; caption = sprintf('Gray Scale Image with box overlaid'); title(caption, 'FontSize', fontSize, 'Interpreter', 'None'); drawnow; hp = impixelinfo(); hold on; rectangle('Position', props.BoundingBox, 'EdgeColor', 'r', 'LineWidth', 2); message = sprintf('The width = %f.\nThe height = %f', width, height); uiwait(helpdlg(message)); end

[jue xi]

then i try this coding, i get what i want, only the hands and i can try to combine the coding from before to display the measurement. but i need both arm. how can i do it sir?

[Image Analyst]

Try using bwareafilt(binaryImage, 2) to extract the two largest blobs. Or you can use it to extract blobs in a specified size range only.

Please read this link

[jue xi]

if true
  % code
folder=('C:\Users\user\MATLAB');
  baseFileName=('img11.jpg');
  fullFileName=fullfile(folder,baseFileName);
  format long g;
  format compact;
  fontSize = 20;
  %IMAGE SEGMENTATION
  img=imread(fullFileName);
  img=rgb2ycbcr(img);
  for i=1:size(img,1)
      for j= 1:size(img,2)
          cb = img(i,j,2);
          cr = img(i,j,3);
          if(~(cr > 132 && cr < 173 && cb > 76 && cb < 126))
              img(i,j,1)=235;
              img(i,j,2)=128;
              img(i,j,3)=128;
          end
      end
  end
  img=ycbcr2rgb(img);
  subplot(2,2,1);
  image1=imshow(img);
  axis on;
  title('Skin Segmentation', 'FontSize', fontSize);
  set(gcf, 'Units', 'Normalized', 'OuterPosition', [0 0 1 1]);
  %SEGMENTED IMAGE TO GRAYIMAGE
  grayImage=rgb2gray(img);
  subplot(2,2,2);
  image2=imshow(grayImage);
  axis on;
  title('Original Grayscale Image', 'FontSize', fontSize);
  set(gcf, 'Units', 'Normalized', 'OuterPosition', [0 0 1 1]);
  %GRAY TO BINARY IMAGE
  binaryImage = grayImage < 245;
  subplot(2, 2, 3);
  axis on;
  image3=imshow(binaryImage, []);
  title('Binary Image', 'FontSize', fontSize);
  % Label the image 
  labeledImage = bwlabel(binaryImage);  % label the connected components in an image and assigning each one a unique label
  measurements = regionprops(labeledImage, 'BoundingBox', 'Area');
  for k = 1 : length(measurements)
    thisBB = measurements(k).BoundingBox;
    rectangle('Position', [thisBB(1),thisBB(2),thisBB(3),thisBB(4)],...
    'EdgeColor','r','LineWidth',2 )
  end
  % Let's extract the second biggest blob - that will be the hand.
  allAreas = [measurements.Area];
  [sortedAreas, sortingIndexes] = sort(allAreas, 'descend');
  handIndex = sortingIndexes(2); % The hand is the second biggest, face is biggest.
  % Use ismember() to extact the hand from the labeled image.
  handImage = ismember(labeledImage, handIndex);
  % Now binarize
  handImage = handImage > 0;
  % Display the image.
  subplot(2, 2, 4);
  image4=imshow(handImage, []);
  title('Hand Image', 'FontSize', fontSize);
end

[Image Analyst]

Start your own question for this, and attach 'img11.jpg'.

[jue xi]

i used this coding, now i want to convert the width in cm.. and i have read and try out the calibration demo and still got no clue on how to convert it to cm. </matlabcentral/answers/uploaded_files/114879/Capture.PNG>

if true
  % code
folder=('C:\Users\user\MATLAB');
  baseFileName=('img2copy.jpg');
  fullFileName=fullfile(folder,baseFileName);
  format long g;
  format compact;
  fontSize = 20;
  %IMAGE SEGMENTATION
  img=imread(fullFileName);
  img=rgb2ycbcr(img);
  for i=1:size(img,1)
      for j= 1:size(img,2)
          cb = img(i,j,2);
          cr = img(i,j,3);
          if(~(cr > 132 && cr < 173 && cb > 76 && cb < 126))
              img(i,j,1)=235;
              img(i,j,2)=128;
              img(i,j,3)=128;
          end
      end
  end
  img=ycbcr2rgb(img);
  subplot(2,2,1);
  image1=imshow(img);
  axis on;
  title('Skin Segmentation', 'FontSize', fontSize);
  set(gcf, 'Units', 'Normalized', 'OuterPosition', [0 0 1 1]);
  %SEGMENTED IMAGE TO GRAYIMAGE
  grayImage=rgb2gray(img);
  subplot(2,2,2);
  image2=imshow(grayImage);
  axis on;
  title('Original Grayscale Image', 'FontSize', fontSize);
  set(gcf, 'Units', 'Normalized', 'OuterPosition', [0 0 1 1]);
 %GRAY TO BINARY IMAGE
  binaryImage = grayImage < 245;
  subplot(2, 2, 3);
  axis on;
  image3=imshow(binaryImage, []);
  title('Binary Image', 'FontSize', fontSize);
  % Label the image 
  labeledImage = bwlabel(binaryImage);  % label the connected components in an image and assigning each one a unique label
  measurements = regionprops(labeledImage, 'BoundingBox', 'Area');
  for k = 1 : length(measurements)
    thisBB = measurements(k).BoundingBox;
    rectangle('Position', [thisBB(1),thisBB(2),thisBB(3),thisBB(4)],...
    'EdgeColor','r','LineWidth',2 )
  end
  % Let's extract the second biggest blob - that will be the hand.
  allAreas = [measurements.Area];
  [sortedAreas, sortingIndexes] = sort(allAreas, 'descend');
  handIndex = sortingIndexes(2); % The hand is the second biggest, face is biggest.
  % Use ismember() to extact the hand from the labeled image.
  handImage = ismember(labeledImage, handIndex);
  % Now binarize
  handImage = handImage > 0;
  % Display the image.
  subplot(2, 2, 4);
  image4=imshow(handImage, []);
  axis on;
  axis image;
  title('Hand Image', 'FontSize', fontSize);
   % Label the image 
  labeledImage = bwlabel(handImage);  % label the connected components in an image and assigning each one a unique label
  measurements = regionprops(labeledImage, 'BoundingBox', 'Area');
  for k = 1 : length(measurements)
    thisBB = measurements(k).BoundingBox;
    rectangle('Position', [thisBB(1),thisBB(2),thisBB(3),thisBB(4)],...
    'EdgeColor','r','LineWidth',2 )
  end
  % Make measurements of bounding box
props = regionprops(labeledImage, 'BoundingBox');
width = props.BoundingBox(3);
height = props.BoundingBox(4);
hold on;
rectangle('Position', props.BoundingBox, 'EdgeColor', 'r', 'LineWidth', 2);
message = sprintf('The width = %f.\nThe height = %f', width, height);
uiwait(helpdlg(message));