Medical Image Segmentation Based On Region-Scalable Fitting Energy

The rapid development of disciplines such as artificial intelligence and computer vision has provided extensive support for digital image processing.As an important research topic in the fields of digital image processing,digital image segmentation aims to divide the image into several sub-regions and segment the target region from the background.However,in the existing image segmentation models,there are problems such as poor generality,inaccurate segmentation,and low efficiency.Therefore,the study of image segmentation is of great significance.In image segmentation,there are many models and algorithms.The active contour model based on the level set method has received extensive attention and is a research hotspot in the field.Among them,in the region-based level set active contour model,most models only consider the regional information of the image,such as the gray average,while ignoring the local information,which makes it difficult to accurately segment the intensity inhomogeneity image and cannot provide accurate segmentation result.'In fact,the pixels in the image are closely related to their neighborhood.Therefore,the relationship between adjacent pixels is also an important feature of an image and can play an important role in image segmentation.Based on the variational level set method and partial differential equations,in this paper,in order to segment the image with a complex background,intensity inhomogeneity and noise pollution,the classic active contour model based on variational theory is studied deeply.The research work on contour sensitivity,noise robustness,and segmentation accuracy has been started.The area-based extended fitting energy model is an important part of the active contour model,and it can be applied to image segmentation with intensity inhomogeneity.However,the segmentation of complex vessels with many branches does not perform well.To overcome this problem,this paper proposes a new energy model.It consists of length terms,area terms,and punitive.Since the length term and the area term in the original region expansion and fitting energy model both have the effect of smoothing the evolution contour,this causes the contour evolution to stop early,and further,the target segmentation is incomplete.Therefore,this paper uses the function property to reduce the impact of length and area terms.Noise reduction is achieved by improving area terms.The experimental results show that the model in this paper is better at segmenting complex blood vessels,more insensitive to initialization contours,and robust to illumination.