Inhomogeneity-embedded active contour for natural image segmentation

作者：

Highlights：

• A novel inhomogeneity-embedded active contour model is built for segmenting images.

• A saliency-inspired framework is built to automatically initialize the contour.

• Our model can produce reliable segmentation especially for inhomogeneous images.

• An image database is proposed in which objects are mostly intrinsic inhomogeneous.

摘要

Highlights•A novel inhomogeneity-embedded active contour model is built for segmenting images.•A saliency-inspired framework is built to automatically initialize the contour.•Our model can produce reliable segmentation especially for inhomogeneous images.•An image database is proposed in which objects are mostly intrinsic inhomogeneous.

论文关键词：Active contour,Level set,Natural image segmentation,Intensity inhomogeneity,Saliency-inspired initial contour

论文评审过程：Received 21 July 2014, Revised 28 December 2014, Accepted 1 March 2015, Available online 11 March 2015.

论文官网地址：https://doi.org/10.1016/j.patcog.2015.03.001