Image Segmentation by Improved Level Set Evolution Algorithm

This paper investigated a novel variational level set evolution mathematical model for image segmentation. Considering some drawbacks of existed level set approaches, such as the re-initialization time-consuming process and the weighted parameters for the area region term. By virtue of recent progress in level set evolution algorithms, we in this paper present two strategies that may be treated as our contributions towards variational level set method. Two scenarios are considered, namely, the fitting energy term in the new model is to weak the sensitives of heavy noise and blur effect. Second, the edge detector can detect the weak boundaries by revising two parameters to offer some suitable convergence and satisfy the necessaries for different type of image edge detection. Simulation studies are presented to verify our method for natural and synthetic degraded images, and to evaluate and compare the new proposed algorithm with existing algorithms.