A New Medical Image Sequence Segmentation Edge Function Of Gac Model

As a research hotspot in the field of image processing, medical image segmentation is widely used in computer-aided diagnosis and treatment of medical. As the object of study, medical images, especially medical image sequences, have unique characteristics that general images do not have, which may provide new opportunity for image processing worker to think, analyze and solve problem from a new perspective. It has important practical significance to dig out these features unique to the medical image sequences, especially when they are quantified and applied to many existing models’improvement, as well as guidance for clinical health care work.Traditional image segmentation model used in the field of medical image segmentation is roughly divided into region-based segmentation algorithm, edge-based segmentation algorithm, and segmentation algorithm combining region and edge. In the paper we study the partial differential equations based Geodesic Active Contour (GAC) Model, and add Local mean and local variance to the edge function, which is a characterization of the image area information. On this basis, we qualify the correlation between medical images of the same sequences, and propose an adaptive parameter, thus construct a new edge function which is to be applied to medical image segmentation. This enhances the robustness of the model, make it better resistance to noise. In order to make the model to maintain a fast convergence rate in the smooth region, reduce the time of iterations, and improve the segmentation efficiency, we express the edge function as a piecewise function finally. At the end of the thesis, we make some comparison of the two edge function previously mentioned through experiment, to prove the improvement.