Research On Image Segmentation Algorithm Based On Active Contour Model

Image Segmentation is a complex and key technology in the field of digital image processing and analysis.Its purpose is to extract the region of interest from the image,and use this region in the subsequent image processing and analysis,so the image segmentation result Will directly affect the quality of the entire digital image processing.Active contour model is one of the most popular algorithms in the field of image segmentation in recent years.Firstly,the basic theories of variational method,gradient descent flow,curve evolution theory and level set method were described.The advantages and disadvantages of the relevant active contour model are analyzed theoretically.Then,several classical active contour models are introduced,And the advantages and disadvantages of these models are illustrated by the experimental results.Then,two improved models are proposed,namely the Chan-Vese model based on the enhanced image information and the active contour model based on the hybrid image rigions.The Chan-Vese model based on the enhanced image information firstly enhances the gray tmiformity and the target boundary of the segmented image by Isotropic Undecimated Wavelet Transform algorithm,and then uses the information of the original and enhanced image to construct the energy functional of the model,And set a contour curve as the initial value of the energy functional,then minimizes this energy functional,When the energy functional reaches the minimum value,the segmentation result is obtained.The active contour model based on the hybrid region will apply the global information of the enhanced image and the local information of the original image to construct the energy functional of the model.In a similar way,The segmentation result is obtained when the minimum value is reached.The experimental results show that Chan-Vese model based on enhanced image information improves the sensitivity of the traditional Chan-Vese model to the images with intensity inhomogeneity effectively,while the active contour model based on hybrid image rigions is not sensitive to image intensity inhomogeneity and initialization contour.And it has significant effect on solve the weight coefficient problem of other active and contour models based on global and local image information,at the same time,the segmentation result accuracy is improved.