Research On Image Segmentation Algorithm Based On PDE And Improved Variational Level Set

During the process of capturing,transmitting,and acquiring images,images are easily disturbed by various types of noise.For pixels in an image,the input of noise tends to change the intensity values of some pixels,resulting in significant differences in the intensity values of neighboring pixels.This thesis focuses on PDE and improved variational level set model,and deeply explores the image segmentation algorithm based on PDE and improved variational level set,the main contents are as follows:Firstly,it introduces several traditional image segmentation methods,the basic theoretical knowledge of curve evolution,and two basic methods of image processing using PDE in detail.It focuses on the theoretical knowledge of the improved variational level set method applied in this thesis,clearly introduces the performance of each segmentation method and its application in practice,and compares the actual segmentation results of several classical models.Finally,the evaluation criteria for image segmentation methods are introduced,laying a solid foundation for further experimental analysis.Secondly,a new level set image segmentation algorithm is proposed for the improved PDE-based GAC model.In this algorithm,’v is used instead of v in the traditional GAC model,while adjusting the evolution speed of corner points and weak boundaries.The experimental results show that compared with the traditional PDE-based GAC model,this algorithm can obtain better segmentation results in a short time,reduce boundary leakage to a certain extent,and shorten the experimental time.Thirdly,for the variational level set model,a new algorithm combining PDE and improving the variational level set noise image segmentation algorithm is proposed.In this algorithm,the symbolic distance penalty term and the double well potential function penalty term are introduced to avoid the reinitialization of the variational level set,improve the robustness of the algorithm in terms of initial contour and noise,and increase the segmentation accuracy of grayscale uneven images,to obtain images with better segmentation effect and high segmentation efficiency,and finally obtain a noisy image segmentation model combining PDE and improved variational level set method.Experimental results show that compared with the traditional level set model,this algorithm can obtain better segmentation results in a short time,reduce boundary leakage to a certain extent,and shorten the experimental time.