|
|
DOI Prefix 10.20431 |
Information
Journal Policies
Image Segmentation Using Truncated Compound Normal with Gamma Mixture Model
Viziananda Row Sanapala1, Sreenivasa Rao Kraleti2, Srinivasa Rao Peri3
2.Professor, Dept of Statistics, Andhra University, Visakhapatnam, India
3.Professor, Dept of Computer Science & Systems Engg, Andhra University, Visakhapatnam, India
Citation : Viziananda Row Sanapala,Sreenivasa Rao Kraleti,Srinivasa Rao Peri, Image Segmentation Using Truncated Compound Normal with Gamma Mixture Model International Journal of Research Studies in Computer Science and Engineering 2016, 3(5) : 1-13
In this paper, we formally present the truncated compound normal with gamma distribution model and define a mixture model(TCNGM) based on this as an extension work to the proposed compound normal with gamma mixture(CNGM) model introduced by us in our earlier work on image segmentation. We present update equations for this model for maximum likelihood estimation (MLE) procedure under Expectation Maximization (EM) framework, construct EM algorithm, and test the feasibility of the model to solve mixture density estimation problem in general and image segmentation in particular. We have found this model to be a competing one in the context of variations in data distributions within probabilistic framework.