Improving the performance of skin segmentation in quasi-skin regions via multiple classifier system

This paper presents a skin segmentation methodbased on multiple classifier system strategy in order toimprove the performance of classification especially in quasiskinregions. Quasi-skin regions in digital images are non-skinpatches which have characteristics like the human skin and areknown as a basic origin of misclassification error in skinsegmentation. To cope with this problem, we have designed analgorithmic architecture by combining four prominentclassifiers to construct a synergy to conceal their weaknessesand amplify their strengths. Participant classifiers in ourapproach include cellular learning automaton, likelihood,Gaussian and Support Vector Machines in which decisionmaking performs via a conditional voting step. The accuracyand specificity were employed to evaluate the performance.Experiments on a collected test-set database including 142challenging images demonstrate that the woposed skindetector is able to improve the accuracy and specificity up to1.92% and 0.83%, respectively, than the best of individualclassifier.