mmdBeef
Detect the lean meat region in a beef steak image.

The gray-scale image of the beef steak is read.

Image a = mmReadgray("beef.tif");
mmShow(a);

a

The input image is simplified by the application of a a small closing. The dark area (inner lean part) is closed from the fat white area.

Image b = mmClose(a, mmSedisk(2));
mmShow(b);

b

The external marker is built from the detection of the complete beef region and the extraction of its internal edge. As the beef is dark, it is detected by a low value threshold. After this threshold, small residual regions are eliminated by the binary areaclose operator.

Image c = mmThreshad(a, 10);
Image d = mmAreaclose(c, 200);
mmShow(d);

d

The internal edge generated is 13 points thick. It is created by the residues of an erosion by a large structuring element.

Image e = mmGradm(d, mmSecross(1), mmSebox(13));
mmShow(e);

e

The internal marker is a severe erosion of the steak. Both markers are combined by union and displayed as overlay on the gradient image

Image f = mmEro(d, mmSecross(80));
Image g = mmUnion(e, f);
Image h = mmGradm(b);
mmShow(h, g);

h, g

Constrained watershed of the gradient of the smoothed image, restricted to the internal and external markers

Image i = mmCwatershed(h, g); 
mmShow(i);

i

Superposition of the dilated detected contour on the original image.

mmShow(a, mmDil(i));

a, mmDil(i)