|
Siddharth Manay
ECE
Georgia Institute of Technology
Anti-Geometric Diffusion for Adaptive Thresholding and Fast Segmentation
In this presentation we utilize an anisotropic diffusion model,
which we call the anti-geometric heat flow, for adaptive thresholding of
bimodal images and for segmentation of more general greyscale images. In a
departure from most anisotropic diffusion techniques, we select the local
diffusion direction that smears edges in the image rather than seeking to
preserve them. In this manner, we are rapidly able to detect and
discriminate between entire image regions that lie nearby, but on opposite
sides, of a prominent edge. The detection of such regions occurs during the
diffusion process rather than afterward, thereby side-stepping the most
notorious problem associated with diffusion methods, namely, ``When should
you stop diffusing?'' We initially outline a procedure for adaptive
thresholding, but ultimately show how this model may be used in conjunction
with a fast region merging procedure for more general grayscale image
segmentation as a region splitting operator, and discuss a complementary
energy-based merging model. We present an iterative splitting and merging
framework for grayscale image segmentation, discuss the fast implementation
of these models, and demonstrate the model on medical, military, and scene
images in two and three dimensions. We conclude by presenting some recent
extensions of this model to color imagery.
|
