Akciğer Bölgesinin Bölütlenmesinde Karmaşık Değerli Sınıflayıcıların Karşılaştırılması

Authors : Murat Ceylan, Yüksel Özbay, Osman Nuri Uçan

Pages : 0-0

View : 43 | Download : 8

Publication Date : 2011-05-01

Article Type : Research Paper

Abstract :Image segmentation is an important step in many computer vision algorithms. The objective of segmentation is to obtain an optimal region of convergence. Error in this stage will impact all higher level activities. In this study, three types of complexvalued classifier were compared to the segmentation of lung region. These classifiers are complex-valued artificial neural network insert ignore into journalissuearticles values(CVANN);, complex-valued wavelet artificial neural network insert ignore into journalissuearticles values(CVWANN); and complex valued artificial neural network with complex wavelet transform insert ignore into journalissuearticles values(CWT-CVANN);. To test the performance of the proposed systems, Lung Image Database Consortium insert ignore into journalissuearticles values(LIDC); dataset was used. Obtained results shown that lung region segmentation done using CVWANN and CVANN with worst accuracy rates as 38.59% and 75.66%, respectively. On the other hand, CWT-CVANN structure segmented lung region with 100% accuracy rate. Moreover, this structure required only 4.5 second per image for segmentation task. Thus, it is concluded that CWT-CVANN is a comprising method in lung region segmentation problem.
Keywords : Lung region segmentation, complex wavelet transform, complex valued artificial neural network

ORIGINAL ARTICLE URL