Find distance from centroids to a specific point and remove objects that are having greater distance than the threshold

Remove objects having computed centroid distance greated than the threshold value. ECR point which is a point that crossing the 100th row from the bottom of the image and centre of the image. Find distance from each centroid to ECR and remove if any one object is having greater distance than the threshold.

Step1: Find threshold of the image using OSTU method

Step2: Find and locate the centroids that are present in each object and an ECR point which is a point that crossing the 100th row from the bottom of the image and centre of the image.

Step3: Draw a line from each centroid to ECR

Step4: Find a distance from each centroid to ECR

Step5: Remove objects having computed centroid distance greater than the threshold value

step6: Save the final image

function CentroidThresholding()
% input binary image file
I = 'D:\PRIVATE\Medical Image Processing\MedicalImageProcessing\Resources\RPE\SmallObjectsRemovedImages\SmallObjectsRemoved_GreenChannel_Rescaled_Cropped_NSB001.tif';
BW = imread(I);
%OTSU Threshold 
T = graythresh(uint8(BW));
[r, c] = size(BW);
%Find centroids of objects
stat = regionprops(BW,'centroid');
centroids = cat(1,stat.Centroid);
% ECR is the point crossing the 100th row from the bottom and centre of the image
ECR = [r-100, c/2];
imshow(I); 
hold on;
% mark centroids
plot(centroids(:, 1), centroids(:, 2), 'r.','LineWidth',2, 'MarkerSize', 15); 
% mark ECR point
plot((r-100)/2, c, 'y.','MarkerSize', 15);
%Draw a line from each centroid to ECR point
% Find centroid distance to ECR point
CentroidDistances=pdist2(centroids,ECR,'euclidean');
%Remove objects having computed centroid distance greater than the threshold value
hold off;
end

Problem to solve

1. I want to draw a line from each centroid to ECR point.

2. I want to remove objects having computed centroid distance greater than the threshold value

Answer 1

[Matt J]

stat = regionprops(BW,'centroid','PixelIdxList');

...

CentroidDistances=pdist2(centroids,ECR);

stat=stat(CentroidDistances > threshold);

for i=1:numel(stat)
    
    BW(stat(i).PixelIdxList)=0;
    
end