17 March 2008 Adaptive Markov feature estimation and categorization using the projection-slice theorem
Author Affiliations +
Classification of features extracted by use of the projection-slice theorem and the representation of data streams through a generalized random filed model is investigated. The approach taken here is to generate probability density functions from the data that can be utilized for the generation of transition and emission probabilities allowing an adaptive progression for the Markov Random Field , MRF, model. Because different image variants of the same image are generally collinear, the images are orthogonalized using eigen vectors that correspond to the largest eigenvalues of the covariance matrix representing the image variations. This helps to reduce the dimensionality of the data as well as ensures maximally independent data in the feature selection process. The projection-slice synthetic discriminant functions are utilized to combine the features selected by use of the MRF to reduce a significant amount of data in the generation of the PSDF, the ensuing results are compared to the original data set combined in the PSDF, showing no significant loss in the peak-to-correlation energy performance metric while significant data is removed in the generation of the PSDF.
© (2008) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Vahid R. Riasati, "Adaptive Markov feature estimation and categorization using the projection-slice theorem", Proc. SPIE 6973, Data Mining, Intrusion Detection, Information Assurance, and Data Networks Security 2008, 69730J (17 March 2008); doi: 10.1117/12.777411; https://doi.org/10.1117/12.777411

Back to Top