2017/02/08 by Jan Henrik Fitschen, Fitschen, Jan Henrik, Katharina Losch +3
Computer Science · Engineering · Medicine · #Advanced X-ray and CT Imaging #FOS: Mathematics #Medical Image Segmentation Techniques #Medical Imaging Techniques and Applications #Numerical Analysis (math.NA) #Sparse and Compressive Sensing Techniques
paper · pdf · doi:10.48550/arxiv.1702.02300
openalex publication_date 2017/02/08 · openalex created_date 2022/10/01 · openalex updated_date 2026/07/28
Intensity inhomogeneities in images constitute a considerable challenge in\nimage segmentation. In this paper we propose a novel biconvex variational model\nto tackle this task. We combine a total variation approach for multi class\nsegmentation with a multiplicative model to handle the inhomogeneities. Our\nmethod assumes that the image intensity is the product of a smoothly varying\npart and a component which resembles important image structures such as edges.\nTherefore, we penalize in addition to the total variation of the label\nassignment matrix a quadratic difference term to cope with the smoothly varying\nfactor. A critical point of our biconvex functional is computed by a modified\nproximal alternating linearized minimization method (PALM). We show that the\nassumptions for the convergence of the algorithm are fulfilled by our model.\nVarious numerical examples demonstrate the very good performance of our method.\nParticular attention is paid to the segmentation of 3D FIB tomographical images\nwhich was indeed the motivation of our work.\n