An Active Contour Model Based on Retinex and Pre-Fitting Reflectance for Fast Image Segmentation

In the present article, this paper provides a method for fast image segmentation computer vision, which is based on level set method. One dominating challenge in uneven illumination and inhomogeneous intensity, are caused by position of light source or convex surface. This proposes variational model Retinex theory. To be specific, firstly, figures out pre-fitting reflectance using an algorithm whole domain before iterations; secondly, it reconstructs additive model; thirdly, uses deviation between global low-frequency component to approximate reflectance, significant part energy function. addition, new regularization term has been put forward extract vanishing gradients. Furthermore, capable accelerating process. Symmetry plays essential role constructing function figuring gradient flow set.