An Interactive Algorithm for Blurred Medical Image Segmentation Based on Curve Fitting
Wenming Huang · 2008
This paper presents a user steered segmentation algorithm based on the radial basis function (RBF) curve fitting for blurred medical image caused by the continuity of organism, which results in the difficulties of algorithm based on the radial basis function (RBF) curve fitting for blurred medical image caused by the continuity of organism, which results in the difficulties of satisfactory automatic segmentation. The algorithm begins with the feature knots selection from the user interested area, followed by the reconstruction of an implicit surface, whose iso-line is the segmented areas, in 3D by using RBF to interpolate these knots. Two acceleration methods are also presented for the bounding box of the feature points and mathematical morphology operators. Lots of experiments show that the approach is efficient for dealing with very blurred and noisy medial images. The interpolation knots should be selected adaptively such that the fine details can be represented well. The accuracy of the segmentation is mainly determined by the position of the knots.