Human mental states recognition under face occlusion
Jayshree Patii, Rachna Patei, Shraddha V. Kothiya · 2017
Recognition of mental state of human is most challenging due to face occlusion. Because of face occlusion, some of facial features points are loss so result will be incorrect at that time. Estimating feature points and removing face occlusion is exceptionally complex and time consuming so alternative is to use hand-over-face gesture to recognized emotions or human mental states. Preprocessing, feature extraction and classification steps are carried out for recognizing mental state of human in the case of face occlusion. In pre-processing step, the three steps will be performed i.e. converting input image into gray scale, resizing and aligning image. In these feature extraction stage, HOG based features are extracting then classification done by SVM classifier. Out of these steps, classification and feature extraction are most important steps because it affects overall accuracy of the system. Selection of relevant methods of feature extraction and classifier are most important factors for achieving good recognition performance. Human mental states under face occlusion are recognized by using MATLAB tool that includes Graphical User Interface Development Environment (GUIDE) for designing graphical GUIs.