Tracking displacement of objects from a stationary point in an image

WARNING: I am VERY new to MATLAB so I apologise if this question is naive.

I currently have images (300+) such as the one attached with scattered 'black' markers. Throughout the series of images, the markers move in one direction and in fairly small amounts. The goal is to determine a stationary point (chosen by the user) then find the initial distance between the stationary marker and every other marker, then find how much each marker is displaced in reference to that stationary point.

I was able to loop through all the images and detect all the markers in each by finding their centroids. I then saved them each into cells using:

    cent{i} = centroids.

I also was able to let the user choose a stationary point by using:

    figure, imshow(firstImage);
    stationaryPoint = ginput(1);

Now I am trying to sort out how to find that initial distance then loop through all the images and find how much that distance changes by. Any advice would be more than helpful. Thanks!

Answer 1

[Image Analyst]

Try this

% Threshold the image
binaryImage = grayImage < 128;
% Label
[labeledImage, numRegions] = bwlabel(binaryImage);
% Find centroids.
props = regionprops(labeledImage, 'Centroid');
xyCentroids = [props.Centroid]; % [x1,y1,x2,y2,x3,y3,x4,y4,.....]
xCentroids = xyCentroids(1:2:end);
yCentroids = xyCentroids(2:2:end);
[x,y] = ginput(1);
distances = sqrt((xCentroids - x) .^ 2 + (yCentroids - y) .^ 2);

That gets the distance of every black blob's centroid to the point the user specifies with ginput().

[Kimberly Morales]

Hi ImageAnalyst,

Thank you so much for helping. The initial image does not show up to allow the user to manually select a stationary marker; instead it shows the attached image.

Also, how can I make sure it is tracking the same marker over time? The desired result is to divide the initial distance found in the first image by the difference between the "new" distance between the marker and the stationary point and the initial distance (displacement/initial). This should give me 100+ values (or however many markers there are) for each iteration which I want to store in spreadsheets.

This is the bottom portion of my code after a bunch of image enhancement (taken from your blobs demo) and the help you provided:

% Identify individual blobs by seeing which pixels are connected to each other.
% Each group of connected pixels will be given a label, a number, to identify it and distinguish it from the other blobs.
[labeledImage, numRegions] = bwlabel(im_BI, 8); 

% Find the coordinates for each marker in the image.
% bwboundaries() returns a cell array, where each cell contains the row/column coordinates for an object in the image.
boundaries = bwboundaries(binaryImage);
numberOfBoundaries = size(boundaries, 1);

% Apply random color labels to each marker
colorLabeled = label2rgb(labeledImage, 'hsv', 'k', 'shuffle');

% Get all the marker properties. 
markerMeasurements = regionprops(labeledImage,'centroid');
xyCentroids = [markerMeasurements.Centroid]; % [x1,y1,x2,y2,x3,y3,x4,y4,.....]
xCentroids = xyCentroids(1:2:end);
yCentroids = xyCentroids(2:2:end);
[x,y] = ginput(1);
distances = sqrt((xCentroids - x) .^ 2 + (yCentroids - y) .^ 2);

[Image Analyst]

Evidently you never displayed an image before you called ginput(). Try this demo:

grayImage = imread('coins.png');
binaryImage = grayImage > 128;
binaryImage = imfill(binaryImage, 'holes');
imshow(binaryImage);
% Enlarge figure to full screen.
set(gcf, 'Units', 'Normalized', 'Outerposition', [0, 0.05, 1, 0.95]);
axis on;

% Identify individual blobs by seeing which pixels are connected to each other.
% Each group of connected pixels will be given a label, a number, to identify it and distinguish it from the other blobs.
[labeledImage, numRegions] = bwlabel(binaryImage, 8); 

% Find the coordinates for each marker in the image.
% bwboundaries() returns a cell array, where each cell contains the row/column coordinates for an object in the image.
boundaries = bwboundaries(binaryImage);
numberOfBoundaries = size(boundaries, 1);

% Apply random color labels to each marker
colorLabeled = label2rgb(labeledImage, 'hsv', 'k', 'shuffle');

% Get all the marker properties. 
markerMeasurements = regionprops(labeledImage,'centroid');
xyCentroids = [markerMeasurements.Centroid]; % [x1,y1,x2,y2,x3,y3,x4,y4,.....]
xCentroids = xyCentroids(1:2:end);
yCentroids = xyCentroids(2:2:end);
uiwait(msgbox('Click on a point'));
[x,y] = ginput(1)
hold on;
plot(x, y, 'ro', 'MarkerSize', 15);
distances = sqrt((xCentroids - x) .^ 2 + (yCentroids - y) .^ 2)

% For fun, draw lines from each centroid to the clicked points.
for k = 1 : length(xCentroids)
  line([x, xCentroids(k)], [y, yCentroids(k)], ...
    'LineWidth', 2, 'Color', 'r');
end

[Kimberly Morales]

Ah yes, I forgot to display an image. I added that in and made sure I am only allowing the user to pick a point in the initial image. However, I am still lost on how to 'store' that initial distance value and how to calculate the distances for every following image.

Right now, I have done everything in a for loop that goes through all of the images. I think I am just getting very confused as to what I should leave in the loop and when I should exit it to perform any calculations. Is this calculating the distance from the stationary point to every marker for ever image? If so, how can I perform calculations on this and export the values into spreadsheets?

distances = sqrt((xCentroids - x) .^ 2 + (yCentroids - y) .^ 2)

You have been a tremendous help!

[Rajindu Goonewardena]

is it possible to run the above coins demo in real time ?