期刊名称：International Journal of Innovative Research in Computer and Communication Engineering
印刷版ISSN：2320-9798
电子版ISSN：2320-9801
出版年度：2015
卷号：3
期号：2
DOI：10.15680/ijircce.2015.0302102
出版社：S&S Publications
摘要：An innovative method to incorporate prior domain knowledge into normalized cuts for biomedicalimage segmentation has started to play one of the most fundamental vital roles in diagnosis and treatment of diseases asthe novel medical imaging technologies progress .In this paper; we initiated an innovative method to embody precedingknowledge into normalized cuts. The preceding is incorporated into the cost function by maximizing the similarity ofthe preceding to one partition and the dissimilarity to the other. This simple formulation can also be extended tomultiple proceeding to allow the modeling of the shape variations. A shape model obtained by PCA on a training setcan be easily integrated into the new framework. This is in variation to other methods that usually incorporatepreceding knowledge by hard constraints during optimization. An Eigen value problem inferred by spectral relaxationis not sparse, but can still be solved efficiently. We engage this method to biomedical data sets as well as naturalimages of people from a public database and examine it with other normalized cut based segmentation algorithms. Wedemonstrate that our method gives promising results and can still give a good segmentation even when the prior is notaccurate.
关键词：Image segmentation; normalized cuts; normalized cuts with shape prior; shape model; spectral;relaxation and medical segmentation
