Past

On New Variational Models for Selective Image Segmentation

Abstract

Image segmentation is one important problem in mathematical imaging research and computer vision applications. As the fast-growing technologies of imaging generate increasingly higher precision images, demand for fast and accurate solution techniques is equally high.

This talk will first discuss some models in variational image segmentation. Then it will focus on more recent works done at our Liverpool group to design robust models for segmentation of  images with weak contrast. Our work can be sued to help preparation of AI training data which is a crucial step in deep learning. The talk covers joint work with several colleagues inducing M. Roberts, L. Burrows, J. Spencer (Exeter) and J. M. Duan (Imperial).


Refs:

[1] J Spencer, Ke Chen and J M Duan (2018), ``Parameter-Free Selective Segmentation with Convex Variational Methods", to appear in  IEEE Transactions on Image Processing.

[2] M Roberts, Ke Chen and K Irion (2018), ``A Convex Geodesic Selective Model for Image Segmentation", to appear in Journal of Mathematical Imaging and Vision.