Dissertation
Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device

Study on Level Set Evolution Without Reinitialization

Author QinJin
Tutor HeChuanJiang
School Chongqing University
Course Computational Mathematics
Keywords Image Segmentation Geometric Active Contour Model Level Set Method Partial Differential Equation
CLC TP391.41
Type Master's thesis
Year 2011
Downloads 11
Quotes 0
Download Dissertation

Image segmentation is one of the most fundamental problems in image processing and is also a basic technique in image analysis. Thus far, a large number of good algorithms and methodologies including geometric active contours using the level-set method have been developed for this task.Geometric active contours models, which are implemented via curve evolution theory and level set method, have been proved to be an efficient framework for image segmentation. Existing geometric models can be roughly categorized into two classes: region-based models and edge-based models. These two types of models both have their pros and cons, and the choice of them in applications depends on different characteristics of images. In this study, we focus on edge-based models.One of the most popular edge-based models is the level set evolution without re-initialization (i.e. Distance preserving level set scheme) for image segmentation. It employs a penalizing energy to force level set function to be close to a signed distance function during evolution, thus re-initialization can be entirely eliminated. Meanwhile, a significantly larger time step and simple finite difference scheme can be used for numerically solving the evolution partial differential equation, and therefore speeds up the curve evolution. However, this method has some limitations in applications, such as, failing to locatig sharp corner, deteting the exterior and interior boundaties, and segmenting images with concavities or multi-objects.In this dissertation, we intergrate the laplacian of binomial distribution filter (LoB) into the distance preserving level set scheme, and proposed a novel method for image segmentation. The proposed method can detect certain object boundaries, for which the original method is not applicable; e.g., it can automatically detect interior and exterior contours of an object and edges of multi-objects. Moreover, active contours can move into boundary concavities and perform better in the presence of sharp corner. Experiments on synthetic and real images demonstrate the advantages of the proposed method over the original one.

Related Dissertations
More Dissertations