how to remove bifurcation points which were pixels with more than two eight-way neighbours in an retinal black and white image?

i have an black and white image which is skeletonaized. For that image i have to find out the pixel which has more than two eight way connective components(all neighbours which are ones and not zeros). i have used bwareaopen(image,p,conn),where conn is 8 connectivity and removes the points which have less than p pixels.it is helpful when we need only one eight neighbour hood but i wanted two eight way neighbourhood.

thank you.

Answer 1

[Image Analyst]

To remove bifurcation pixels, use bwmorph(binaryImage, 'branchpoints') and then set those pixels to false. Untested code:

endPointsImage = bwmorph(binaryImage, 'branchpoints');
binaryImage(endPointsImage) = false;

You can dilate the endPointsImage with imdilate() if you want to enlarge the area around the branchpoints being erased.