A heuristic algorithm for Chan-Vese model

• Published in: 2018 26th Signal Processing and Communications Applications Conference (SIU) pp. 1 - 4
• Main Authors: Cataloluk, Hatice, Celebi, Fatih Vehbi
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-05-2018
• Subjects: Active contour; Active contours; Computational modeling; contour initialization; gravitational search algorithm; Gravity; heuristic; Heuristic algorithms; Image edge detection; Image segmentation; Simulation
• DOI: 10.1109/SIU.2018.8404820

In this paper, a novel heuristic algorithm for Chan-Vese Model has been presented. In the conventional Chan-Vese model, proper initial contour selection is an important issue, since it causes to wrong segmentation due to local optima. Our model incorporates Gravitational Search Algorithm, which is used to keep contour evolution from getting stuck in local optimum points. A gravitational force based approach has introduced to make the contour reach the accurate boundaries. The proposed model was tested on various suitable images and simulation results has demonstrated that the proposed model can automatically and reliably segment images even in cases of inhomogeneity, cluttered images and not well-defined edges.