Find an object in an image by color

So I'm using a webcam to take a color picture of an area. In that area there is a red thing and a green thing among a bunch of other things. Basically I need to return the location of the center of the red thing and the green thing. The exact shape of the objects is not known in advance.

Basically I just need to tell matlab to find the color thing and have it give me coordinates from a color image

Answer 1

[Image Analyst]

See my attached demo where I track a green thing in the video. It's a simple matter to add a check for red things.

[Morpheuskibbe]

Ok lets see if im following. A lot of this is displaying stuff for your demo and hence unnecessary for me so I'm just parsing whats needed and whats not.

Given RGB image "img" I THINK this is all i need from your code.

hThresholds = whatever

sThresholds = whatever

vThresholds = whatever

hsv = rgb2hsv(double(img));

hue=hsv(:,:,1);

sat=hsv(:,:,2);

val=hsv(:,:,3);

binaryH = hue >= hThresholds(1) & hue <= hThresholds(2);

binaryS = sat >= sThresholds(1) & sat <= sThresholds(2);

binaryV = val >= vThresholds(1) & val <= vThresholds(2);

coloredMask = binaryH & binaryS & binaryV;

coloredMask = bwareaopen(coloredMask, 500);

coloredMask = imfill(coloredMask, 'holes');

[labeledImage, numberOfRegions] = bwlabel(coloredMask);

stats = regionprops(labeledImage, 'BoundingBox', 'Centroid');

And then stats(r).Centroid would be an array of coordinates for the centers of detected things of the desired color? Do I have this right?

[Image Analyst]

stats(r).Centroid would be a 1 by 2 array with the (x,y) coordinate of the r'th centroid. To get all centroids, for all blobs, do

centroids = [stats.Centroid];
xCentroids = centroids(1:2:end);
yCentroids = centroids(2:2:end);

[Morpheuskibbe]

Thanks. It works. I changed it a bit

the line: " hsv = rgb2hsv(double(img));" I just changed to "hsv = rgb2hsv(img);"

I noticed that if I did it that way then the V value was limited to the range 0-1 just like the others instead of big numbers like in your example