Authors : M. Sujaritha and S. Annadurai
Abstract: In this study, we introduce a novel level set method for color image segmentation. It is based on the Binary Space Partitioning (BSP) tree technique developed by Pei and Cheng and the multiphase level-set framework developed by T. Chan and L. Vese. We present a new variational formulation for geometric contours that divides the image region in a binary fashion using binary quaternion moment preserving thresholding technique and therefore completely eliminates the need of the costly re-initialization and calculation of number of regions procedure. The sum of square error value determines the required homogeneity of the color in the region and inturn decides the number of regions in the image. Our variational formulation consists of an internal energy term that penalizes the deviation of the level set function from a signed distance function and quaternion moment based external energy term that drives the motion of the zero level set rapidly and discontinuously toward the color boundaries in an image. The resulting evolution of the level set function is the gradient flow that minimizes the overall energy functional. The proposed algorithm has been applied to both synthetic and natural color images with promising results.
M. Sujaritha and S. Annadurai , 2009. Color Image Segmentation Using Binary Level-set Partitioning Approach. International Journal of Soft Computing, 4: 76-84.