Medical Image Segmentation Based on the SVM & K-NN based Edge Stop Function

Abstract: Edge-based active contour models are successful in segmenting images with intensity in homogeneity but it fail when images are poorly defined boundaries. Such as medical images. The conventional Edge stop function (ESF) use only gradient information it fails to stop gradient magnitude to rectify this problem, we formulate a group of ESFs for edge based active contour models to segment the images with poorly defined boundaries. In our proposed work, which includes gradient information and probability scores from a standard classifier. The distance regularized level set, and k-nearest neighbour’s and SVM are used to experimenting the medical images.

Keywords: Index Terms—Edge-based active contour, edge-stop function, gradient information, image segmentation, probability score.