image processing - Why multiple openings/closing with a same kernel does not have effect? -
i know closing , opening, there still 1 question me! according "digital image processing, 3rd edition", gonzales, multiple application of opening/closing doesn't have effect after first time apply it! couldn't figure out? can help?
this expected behavior since openings , closings idempotent operations. operation idempotent if, whenever applied twice value, gives same result if applied once: f(f(x)) = f(x). openings operators on lattice l idempotent, increasing, , anti-extensive while closings operators on l idempotent, increasing, , extensive. 1 can find discussion on idempotence here.
in more intuitive sense, opening on set x erosion followed dilation same structuring function. once first iteration done set x not change since erosion , dilation remove , add same '1's in set x. product of opening , closing idempotent operation - interesting. 1 other hand if @ each iteration 1 changes radius of structuring element openings/closings, 1 obtain alternated sequential filter, produces multiscale simplification of image, producing scale-space.
i refer book jean serra on mathematical morphology or better understanding.
Comments
Post a Comment