Wavelets and morphological filter (Alternating Sequential Filters) for robust multiscale image segmentation

Abstract

This paper proposes a new segmentation technique which combines multi-resolution wavelet decompositions with the morphological filter by applying the wavelet transform to the image‘s intensity. Gradient magnitudes of the approximation image at the coarsest resolution were computed, and an adaptive threshold was used to remove small gradient magnitudes. The alternate sequential filter (ASF) was then applied, and the segmented image was projected up to higher resolutions using inverse wavelet transforms. Typically, if a low resolution was chosen for the initial segmentation, large relevant objects will be captured; on the other hand, a higher initial resolution will lead to smaller (and more detailed) segmented objects. The low-pass filtering involved in the wavelet decomposition provides robust segmentation results for noisy images, even when the amount of noise is very large.