how can i use a Gaussian gradient routine for remove the out of focused section in image processing

first of all i remove the background and using a threshold for detecting bubble.

but i can't remove only out of focused bubble cuz that threshold is similar with focused bubble

i read that Gaussian gradient routine can remove that part. but i didn't use that . help me~!

Answer 1

[Constantino Carlos Reyes-Aldasoro]

You could refine your processing using bwlabel to assign unique labels to the regions you have already detected. Then use regionprops to detect the size (with Area) and the average intensity, assuming your second image is called image_thresholded, you could do the following:

image_labelled      = bwlabel(image_thresholded);
image_properties    = regionprops(image_thresholded,original,'Area','MeanIntensity');

the properties would look like this

image_properties

image_properties = 

  75×1 struct array with fields:

    Area
    MeanIntensity

 [image_properties.Area]

ans =

  Columns 1 through 4

      269962          34           2           2

  Columns 5 through 8

         122         116          25          65

etc.

 [image_properties.MeanIntensity]

ans =

  Columns 1 through 5

   27.7029   87.3529   58.5000   60.0000  124.5984

  Columns 6 through 10

  122.6983   87.0800   79.6923   71.2500   72.3750

etc.

Then you can select only those that comply with a combination of area (to remove the ring) and intensity (to remove the blurred) using find and ismember, e.g. if you only want those that are below 100 of intensity:

  spots = ismember(image_labelled,find([image_properties.MeanIntensity]<100));

Hope this has answered your question