Research On Image Segmentation Model Based On Wavelet Analysis And C-V Level Set

Image denoising and image segmentation are two important processes in image processing, so far, about these two questions, researchers have proposed a number of methods and models. In this paper, it discusses these two issues by the use of wavelet analysis and curve evolution theory of level set, and improves some of these methods and models. Then by numerical experiments, it verifies the proposed methods and models than the traditional methods and models are better or more efficient.In the first chapter, it mainly introduces the relevant theoretical knowledge of wavelet analysis. Firstly, it introduces development background of the wavelet analysis and some typical wavelet transform. Then introduces the wavelet multi-resolution analysis and Mallat algorithm which occupies a very important position in the development history of the wavelet. Finally, the related content of one-dimensional wavelet analysis is extended to the two-dimensional wavelet analysis, because image processing needs to use two-dimensional wavelet transform and two-dimensional wavelet multi-resolution analysis and two-dimensional wavelet Mallat algorithm.In the second chapter, it mainly introduces the knowledge of image denoising based on wavelet analysis. First, it introduces the knowledge points of image noise and evaluation standard of denoising performance. Then it reviews these common denoising methods. Finally, it discusses the wavelet threshold denoising theory, and constructs a new threshold function, and improves the threshold setting method. And then through the numerical experiment, it proves that the presented method is better than the traditional method in denoising effect.In the third chapter, it mainly introduces the knowledge about the curve evolution theory and the level set method. Firstly, it introduces the related knowledge points of plane curve, including its characteristics of differential geometry, level set method of plane closed curve and so on. Then it introduces the knowledge related to partial differential equation. Finally, it introduces the contents of curve geometric evolution, including the general expressions and level set method of the curve evolution, and so on.In the fourth chapter, it mainly introduces the active contour model for image segmentation and its development. Firstly, it introduces the definition of image segmentation and image segmentation methods and current situation of research. Then it introduces the parametric active contour model and geometric active contour model and the process of their development. through the numerical experiments, it compares the advantages and disadvantages of the two models and part of the segmentation performance indicators. And then it focuses on the C-V model of image segmentation based on level set, including the principle of C-V model, level set representation of the C-V model, the value implementation and algorithm flow of C-V level set model. And through numerical experiments, it verifies the advantage of C-V level set image segmentation model. Finally, based on the shortcomings of C-V level set image segmentation model, it puts forward some improvement measures and advice.In the fifth chapter, it mainly discusses improvement measures and solutions of the C-V level set image segmentation model and its algorithm implementation scheme. Firstly, According to the shortcoming that the C-V level set image segmentation model needs to constantly initialize the level set function to keep it for the signed distance function, it introduces a type of C-V level set image segmentation model which need not to re-initialize the level set function, and it verifies through numerical experiments that the model has certain advantages on segmentation when the size of a image is large. Then it describes the C-V level set image segmentation model of the added edge detection function, and proposes a new edge detection function, the numerical experiments show that the edge detection function proposed has obvious advantages on the image segmentation when the image has a strong edge or the initial contour of the image is far from the object edge. Then it proposes an improved C-V level image segmentation model based on wavelet multi-scale decomposition, first of all, it constructs the edge detection function which uses the high frequency information of wavelet multi-scale decomposition, and then using interpolation method, from coarse scale to fine scale, makes in turn image segmentation, the numerical experiments show that the segmentation efficiency and the segmentation accuracy of this method is obviously superior to other methods. Finally, it improves narrow band method which has the one of the level set fast algorithm, proposes a rapid narrow band algorithm, and through numerical experiments verify its efficiency is higher than the average narrow band method.Finally, it summarizes and prospects for the paper, first of all, it reviews all the contents of this paper, and then it points out the advantages and disadvantages of this paper, at last, it looks forward to the follow-up research direction and research content.