Extraction of memory colors using Bayesian Networks
Mustafa Musa Jaber, Eli S. Saber, Ferat Sahin · International Conference on System of Systems Engineering · 2009
In this work, a region classification algorithm based on low-level features and probabilistic framework is proposed where skin, sky, and vegetation memory color classes are detected in digital images. A region's low-level features are extracted using a segmentation map of input image. Bayesian Network (BN) is used to classify memory color regions for smart rendering in printing applications. Other applications of the proposed technique include image annotation, indexing, and content retrieval. The algorithm was tested on a large database of color images with 85% classification accuracy.