Removing Noise in Image

I'm trying to detect a semicircular region from an image to create a mask. Tried edge detection followed by filtering. Original image on the left and processed image on the right. How can I fill the dark region on the right and remove the extra features on left and right leaving me with just the semicircular region.

Answer 1

[VINAYAK LUHA]

Hi Kartikeya,

I understand that you want to postprocess the obtained mask to remove noise and accurately detect the semicircular object in your image.

Follow the below steps to achieve that -

1. Mirror your mask about the horizontal axis.
2. Stitch the original and mirrored image horizontally to obtain a complete circle.
3. Preprocess the image obtained from step 2.
4. Use the "Image Segmenter" app and detect and save the mask for the complete circle.
5. Horizontally crop the mask obtained from top till mid.

After step 3 -

After step 4 -

After step 5 -

Here is the code snippet for your reference.

image = imread('pathToMask');

%1 Mirror the mask about the horizontal axis.
flipped_image = flipud(image);

%2 Stitch the original and mirrored image horizontally to obtain a complete circle.
joined_image = [image;flipped_image];

%3 Preprocess the image obtained from step 2
gray_image = rgb2gray(joined_image);
blurred_image = imgaussfilt(gray_image, 5);

%4 Use the "Image Segmenter" app and detect and save the mask for the complete circle.
[fullmask,~] = segmentImage(blurred_image);

%Horizontally crop the mask obtained from top till mid.
halfmask = fullmask(1:int32(size(image,2)/2),:, :);
figure;
imshow(halfmask)
hold off;

%Function exported from Image Segmenter app to detect circle and obtain
%mask
function [BW,maskedImage] = segmentImage(X)
%segmentImage Segment image using auto-generated code from Image Segmenter app
%  [BW,MASKEDIMAGE] = segmentImage(X) segments image X using auto-generated
%  code from the Image Segmenter app. The final segmentation is returned in
%  BW, and a masked image is returned in MASKEDIMAGE.

% Auto-generated by imageSegmenter app on 01-Oct-2023
%----------------------------------------------------


% Adjust data to span data range.
X = imadjust(X);

% Find circles
[centers,radii,~] = imfindcircles(X,[400 600],'ObjectPolarity','bright','Sensitivity',0.90);
BW = false(size(X,1),size(X,2));
[Xgrid,Ygrid] = meshgrid(1:size(BW,2),1:size(BW,1));
BW = BW | (hypot(Xgrid-centers(1,1),Ygrid-centers(1,2)) <= radii(1));

% Create masked image.
maskedImage = X;
maskedImage(~BW) = 0;
end

Explore these links to the documentation of "flipud" and "imgaussfilt" for more details -

1. https://in.mathworks.com/help/matlab/ref/flipud.html
2. https://in.mathworks.com/help/images/ref/imgaussfilt.html

Hope this helps !

Regards,

Vinayak Luha

Answer 2

[Image Analyst]

What do we know about the circular/elliptical shape? Is it always partially outside the image? Do the streaks always emanate from the right and left side and never from the middle?

For this particular image, I'd probably find the top edge of your segmented image and fit those (x,y) to a circle. Then create a new mask based on the circular fit.