Level set segmentation method based on local area fitting information and guided filter fuzzy clustering

Image segmentation plays an important role in many aspects of medical image analysis, especially in computer-aided diagnosis. In order to obtain a more accurate target contour, solve the problem that the level set method is sensitive to the initial contour, and make the algorithm more robust to noise, this thesis uses the guided filter fuzzy clustering algorithm to perform image pre-segmentation, and then combines with the level set model based on local area fitting information. First, the fuzzy clustering algorithm is used to obtain the control parameters of the level set evolution. Then, combined with the local neighborhood characteristics of the image points, the regional term coefficients and edge stop functions of the level set model are improved. Finally, the Gaussian Laplacian energy term is introduced to enhance the edge information of the target. Experiments have proved that the algorithm model proposed in this thesis realizes the automation of the initial contour of the level set, and at the same time improves the accuracy of segmentation, and is more robust to noise.