An adaptive speed term based on generalized fuzzy operator for level set segmentation

The major problem of level set segmentation is boundary leakage when a weak boundary or missing part of a boundary occurs. Some efforts, e.g., adding extra stopping criteria or region information, have been used to avoid such problem. However, in existing level-set techniques, the gradient information is used as stopping criteria for curve evolution, and also provides the attracting force to the zero level set from a target boundary, which can never fully stop the level set evolution even for ideal edges and thereby makes leakage unavoidable. In this paper, we proposed an adaptive speed term based on generalized fuzzy operator (GFO), as a replacement for traditional gradient-based edge map applied in level set segmentation techniques. The proposed GFO provides more robust edge estimation and more reliable criteria for adaptively changing the advent speed. Therefore, by using this term, the leakage problem can be avoided effectively in most cases compared to traditional level set methods. Additionally, the computation of the adaptive approach proposed in this paper is fast and its extension to the 3-D case is straightforward.