How to use the find function against regionprops?

Hi there, I''m having trouble using the find function.. Basically, I have this image called binaryImage, I want to find specific areas of the image with a range of parameters. Here's my code:

Lh=bwlabel(binaryImage,8);
blobMeasurements = regionprops(Lh, I, 'all');

And a whole bunch of codes(not included here) that allow me to display in Matlab, a few parameters of the labeled areas(blobs):

Blob #      Mean Intensity  Area       Diameter      Eccentricity
# 1             196.6      4393.0       74.8            0.6             
# 2             234.1        55.0        8.4            0.9             
# 3             242.9         9.0        3.4            0.9             
# 4             216.2       316.0       20.1            1.0             
# 5             232.7        49.0        7.9            1.0            
.
.
.

para1 = [blobMeasurements.Eccentricity];
para2 = [blobMeasurements.MeanIntensity];

Keeper_Indexes = find(0.9<para1<=0.1 & 230<para2<250);
PossibleCracks = ismember(Lh, Keeper_Indexes);
figure, imshow(PossibleCracks, 'DisplayRange', []), title('Possible Cracks');

I thought it would show me the areas of blobs#2, #3, and #5, but figure 'Possible Cracks' doesn't show me that.. instead it shows me alot of unnessary other blob areas. What should i do? Please advice! Thank you!

Answer 1

[Walter Roberson]

When you have

A<B<C

then MATLAB interprets that as

((A<B)<C)

The result of (A<B) is a logical vector of 0's and 1's. If C is all 0's then the result would be false; if C is all greater than 1, then the result would be true; if C contains some 1's then it is possible to get a mix of true and false results out.

In the case of your 0.9<para1<=0.1 you are testing (0's and 1's) <= 0.1 which will be true for the 0's and false for the 1's, and so would be equivalent to just testing

0.9 >= para1

In the case of your 230<para2<250 you are testing (0's and 1's) < 250 which will always be true.

If you want to test a range, then use

(lowvalue < x) & (x < highvalue)

but adjusting the < to <= if appropriate.

Notice that in 0.9<para1<=0.1 that you are not going to be able to find any values that are less than 0.9 but simultaneously less than 0.1 . You should recheck what you want for that condition and rewrite it in the (lowvalue < x) & (x < highvalue) format.

[Kim Yu Jo]

Thank you very much sir!! If its possible could you help me with another problem? I have an 1024x1024 unit8 image, I want to convert it to a 1024x1024x3 unit8 image, which i believe represents a RGB image? How can i do this? Thank you!

[Sean de Wolski]

@Kim,

I3 = repmat(I,[1 1 3])

[Image Analyst]

Alternatively,

rgbImage = cat(3, grayImage, grayImage, grayImage);