Research on Image Segmentation Combined G-L Fractional Differential and LIF Mode
Guimei Zhang, Ke Xu, Bingbing Chen · 2019
Local Image Fitting (LIF) model is sensitive to the initial location of evolving curve when segmenting images with intensity inhomogeneous, weak texture and edges, it usually falls into local optimal. To address this issue, a new method for image segmentation based on fractional differential and LIF model is proposed. It introduces Grünwald-Letnikov (G-L) fractional gradient into LIF model, and adds a new global fractional fitting term. So the gradient of the intensity inhomogeneous and weak texture regions is enhanced. Evolution curve is driven by both the local intensity fitting term and the global fractional differential gradient fitting term, As a result, both robustness to curve's initial location and segmentation efficiency are improved. The proposed model is a improvement of LIF model. It can be used for segmenting the intensity inhomogeneous, weak edge and weak texture images.