Enhancement and Segmentation of Scar Color Images after a Scoliosis Surgery

In this paper, a color-based segmentation scheme applied to dermatoscopic scar images is proposed. The image is first preprocessed via a partial differential equations (PDE)-based filter which combines in the RGB color space coherent anisotropic diffusion and shock filters. Then the scar image is segmented using a segmentation algorithm based on color distributions. A post-processing algorithm was furthermore developed to extract the scar contour, based on spline interpolations, smoothed with the generalized-cross validation method. Experiments on real patient images show the performance of the proposed method.

• Type: Conference
• Year: 2003
• Where: DICTA
• Authors: Thomas Hurtut, Farida Cheriet, Julie Joncas, Jean Dansereau
• Keywords: Color-based Segmentation Scheme, Dermatoscopic Scar Images, DICTA 2003, DICTA 2007, Scar Image