Active Contour Segmentation Algorithm Based On The Image

Image segmentation have been extensively used in image processing、object trackingn medical imaging and so on.It’s goal is that dividing the image into a series of subregions with approximately similar properties.Currently,there are a lot of technologies to solve the problem of image segmentation.However, the diversity and complexity of the image make image segmentation be an important and challenging research topic.In all image segmentation methods,the active contour model is a very important method which is greatly concerned. It has a rigorous mathematical theory,and makes full use of the dynamic evolution,can solve some difficult problem which can’t solved by many traditional methods.Those fully embody the advantages of this method.This paper gives a brief introduction about the background、research purpose and significance of the active contour model.It also briefly explains the mathematical theory,and research several classical active contour models. Driven by the imperfection in those models,we propose some improvement and innovation,which can get ideal segmentation results.The primary work and remarks are as follows(1) Driven by the problem of CV model can’t efficiently segment images with intensity inhomogeneous and local binary fitting model is sensitive to heavy noise, we propose a novel local active contour model driven by image intensity and gradient. This model constructs Local binary gray energy and Local binary gradient energy by using Gaussian function.Then Getting the final energy by Linear combining these two energy. Finally, the contour can be driven to the objective boundaries by minimizing the energy functional. Experimental results show that the proposed model can overcome the impact of the intensity inhomogeneity and heavy noise in the image to get satisfied segmentations.(2) We propose a new active contour model based on fuzzy.This model’s energy function has two parts,including local fuzzy energy and global fuzzy energy. The total energy function is automatically combined local fuzzy energy and global fuzzy energy, and it can drive the contour to the ground truth by minimizing the energy function.The local term allows the method to deal with intensity inhomogeneity in image.The global term can avoid unsatisfying results led by unsuitable initial contour position. Experimental results show that the proposed model can overcome the impact of the intensity inhomogeneity and has strong robustness to the initial contour curve.(3) Drived by LIF model is sensitive to initial contour,we propose an automated image segmentation algorithm based on FCM and LIF model.Experimental results show that the proposed algotithm can get the suitable initial level set function automatically.As a result,the sensitivity to the initial contour is solved and it can decrease computational cost.