Frontal face images are segmented into 7 regions using only sum and difference histograms as pixel information, without any a priori knowledge. In the training phase, a decision tree is created using a projection pursuit algorithm: in each step, the optimal one-dimensional projection is chosen by a simulated annealing process according to a projection index, and classes are isolated by a decision boundary that maximizes class separability, until the end nodes contain only one class each. Satisfactory qualitative and quantitative results were obtained and presented. © 2012 Springer-Verlag.
CITATION STYLE
Laboreiro, V. R. S., Maia, J. E. B., & De Araujo, T. P. (2012). Face segmentation using projection pursuit for texture classification. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7435 LNCS, pp. 237–244). https://doi.org/10.1007/978-3-642-32639-4_29
Mendeley helps you to discover research relevant for your work.