Multiregion Multiscale Image Segmentation with Anisotropic Diffusion


Abstract:

We present a multiregion image segmentation approach which utilizes multiscale anisotropic diffusion based scale spaces. By combining powerful edge preserving anisotropic diffusion smoothing with isolevel set linking and merging, we obtain coherent segments which are tracked across multiple scales. A hierarchical tree representation of the given input image with progressively simplified regions is used with intra-scale splitting and inter-scale merging for obtaining multiregion segmentations. Experimental results on natural and medical images indicate that multiregion, multiscale image segmentation (MMIS) approach obtains coherent segmentation results.

Vectorial Multiscale Diffusion with Inter-Scale Linking [1]

(a) Scale space, T = 500 (b) Intra-scale splitting, shown at t = 300 (c) Inter-scale linking

BSDS500 Segmentation Examples [2]

Cardiac CT and Brain MRI Segmentation Examples [2]

References:

  1. V. B. S. Prasath. Color image segmentation based on vectorial multiscale diffusion with inter-scale linking. 3rd International Conference on Pattern Recognition and Machine Intelligence (PReMI), Delhi, India. Proc. in Springer LNCS 5909. (Eds.: S. Chaudhury, S. Mitra, C. A. Murthy, P. S. Sastry, Sankar K. Pal), pp. 339-344, December 2009. doi:10.1007/978-3-642-11164-8_55

  2. V. B. S. Prasath, D. N. H. Thanh, N. H. Hai, S. Dvoenko. Multiregion multiscale image segmentation with anisotropic diffusion. 7th Workshop Image Mining. Theory and Applications (IMTA VII), January 2021. Proc. Springer LNCS 12665, 129-140, February 2021. doi:10.1007/978-3-030-68821-9_13

  3. V. B. S. Prasath. Edge preserving diffusion flows based multiscale image segmentation. In preparation, 2022. Preliminary version at arXiv, 2022. doi:10.48550/arXiv.22xx.12345

For anisotropic diffusion of mono channel (gray-scale images), please visit this project page.