Research on Segmentation

Feature descriptors evaluation
Feature descriptors for Mitochondria and Synapse Segmentation. Full understanding of the architecture of the brain is a long term goal of neuroscience. To achieve it, advanced image processing tools are required, that automate the the analysis and reconstruction of brain structures. Synapses and mitochondria are two prominent structures with neurological interest for which various automated image segmentation approaches have been recently proposed. In this work we present a comparative study of several image feature descriptors used for the segmentation of synapses and mitochondria in stacks of electron microscopy images.
Related Publications: ICPR'2014
Non-parametric Higher-Order Random Fields
Non-parametric Higher-Order Random Fields for Image Segmentation. Models defined using higher-order potentials are becoming increasingly popular in computer vision. However, the exact representation of a general higher-order potential defined over many variables is computationally unfeasible. This has led prior works to adopt parametric potentials that can be compactly represented. In this work we proposed a non-parametric higher-order model for image labeling problems that uses a patch-based representation of its potentials. We use a known transformation scheme to convert the higher-order potentials to a pair-wise form that can be handled using traditional inference algorithms. This representation is able to capture structure, geometrical and topological information of labels from training data and to provide more precise segmentations. Other tasks such as image denoising and reconstruction are also possible. We evaluate our method on denoising and segmentation problems with synthetic and real images.
Related Publications: ECCV'2014
Morphological Snakes
A morphological approach to curvature-based evolution of curves and surfaces. We introduce new results connecting differential and morphological operators that provide a formal and theoretically grounded approach for stable and fast contour evolution. Contour evolution algorithms have been extensively used for boundary detection and tracking in computer vision. The standard solution based on partial differential equations and level-sets requires the use of numerical methods of integration that are costly computationally and may have stability issues. We present a morphological approach to contour evolution based on a new curvature morphological operator valid for surfaces of any dimension. We approximate the numerical solution of the curve evolution PDE by the successive application of a set of morphological operators defined on a binary level-set and with equivalent infinitesimal behavior. These operators are very fast, do not suffer numerical stability issues and do not degrade the level set function, so there is no need of re-initializing it. Moreover, their implementation is much easier since they do not require the use of sophisticated numerical algorithms. We validate the approach providing a morphological implementation of the Geodesic Active Contours, the Active Contours Without Borders and Turpopixels. In the experiments conducted the morphological implementations converge to solutions equivalent to those achieved by traditional numerical solutions, but with significant gains in simplicity, speed and stability
Related Publications: CVPR'2010, PAMI'2014