Low contrast image, ablated area extraction

Hi,

With a laser I ablated a certain area from an ink sample. Not the entire layer is ablated. Therefore, the contrast of the hole with the surrounding area is low.

With image processing I would like to determine the area of the ablated hole. This information would be very valuable for my research.

Below you find the script with my attempts to solve this problem. please help.

Kind regards,

Hidde

clear all; close all; clc;

I = imread('400_2_300.jpg');

I = rgb2gray(I);

imh = imadjust(I);

J = medfilt2(imh);

se = strel('disk',25);

thfilter = imtophat(J,se);

adjust = imadjust(thfilter);

bin = im2bw(adjust,0.35);

fill = imfill(bin,'holes')

filt_bin = medfilt2(bin);

fill1 = imfill(filt_bin,'holes');

img_close = imclose(fill1 , strel('disk', 1));

%take complement to adhere to convention

imgbw = imcomplement(fill1) ;

%make bridges across small "leaks"

img_close = imclose(imgbw , strel('disk',5));

%find holes in resulting image

holes = ~img_close & imfill(img_close, 'holes');

%add back holes to original image

img_filled = imfill((imgbw | holes) , 'holes');

%optional additional filling

img_filled_closed = imclose(img_filled, strel('disk', 5)) ;

figure(1), clf(1), hold on

subplot(3,3,1)

imshow(I)

subplot(3,3,2)

imshow(imh)

subplot(3,3,3)

imshow(J)

subplot(3,3,4)

imshow(thfilter)

subplot(3,3,5)

imshow(adjust)

subplot(3,3,6)

imshow(bin)

subplot(3,3,7)

imshow(fill)

subplot(3,3,8)

imshow(fill1)

subplot(3,3,9)

imshow(img_filled_closed)

hold off

[Hidde Pik]

I already used a cropped image through the imtool.

See the attachment

[Image Analyst]

Is the spot always in the same place in the field of view so that you can crop out the same place? Or are you willing to manually crop it each time?

[Hidde Pik]

Unfortunately the spot is not always in the same place. I am willing to do it manually. It will be some work, since I have 92 images. However cropping takes about 5 seconds to do.

[Hidde Pik]

@Image AnalystI managed to crop all the images. However I didn't manage to extract the area of the ablated hole. I attached the Binary Image that I end up with. As you can see he still takes a lot of the surrounding as well.