Skin Detection Based on Combining Kernel-based Classifiers via Game Theory
Paper ID : 1107-IST
1Fahimeh Salimi Koochi *, 2Hadi Farzin
1ICT Research Institute (ITRC)
In this paper, a novel fusion method is proposed to combine different kernel-based classifiers in the structure of skin detection systems. This method is based on the multi-perturbation Shapley analysis, a framework which relies on game theory, to estimate the weight of each classifier in the weighted averaging fusion rule. In this study, six different types of kernel functions are applied in the frame of Generalized Discriminant Analysis (GDA) algorithm in the YCbCr color space. The efficiency of the resulting system is evaluated using Compaq database. We demonstrate that by fusing different classifiers using the proposed method, the performance of the detection system improves.
skin detection; kernel methods; GDA; fusion; Game theory; Shapley value