Boundary to medial axis transform

Question

Does anyone know an easy way, given a binary image, to construct a transform which for each point on the boundary (contour set pixel) returns the point on the medial axis that it is closest to?

Answer 1

I don't know which language you are using for your program, but you can use python.

with the scikit-image library it will be really easy.

First of all, you have to calcul all the component label (optional, depends on your binary image) thanks to the methods : skimage.morphology.label(image, return_num=True)

after that you can calculate each medial axis thanks to : skimage.morphology.medial_axis(img, return_distance=True)

so at this point you've got sets of points with contour pixel and sets of points with medial-axis pixel.

You just have to calculate for each component label a distance between pixel you want and all pixel in set of medial axis dist=sqrt((y2-y1)^2+(x2-x1)^2)

I hope it will help you.