| Popis: |
Human face states the inner emotions, thoughts and physical disorders. These emotions are expressed on the face via facial muscles. The estimated time through which a facial expression occurs on the face is between 0.5 to 4 seconds, and a micro expression between 0.1 to 0.5 seconds. Obviously, for the purpose of recording micro expressions, obtaining videos frames between 30 up to 200 frame per second is essential. This research uses Kinect V.2 sensor to get the color and depth data in 30 fps. Depth image stores useful 2.5-Dimentional information from skin wrinkles which is the main key to recognize even slightest micro facial expressions. Experiment starts with splitting color and depth images into facial parts, and after applying preprocessing techniques, features extraction out of both type of data in spatial and frequency domain takes place. Some of the features which are used in this study are Histogram of Oriented Gradient (HOG), Gabor Filter, Speeded Up Robust Features (SURF), Local Phase Quantization (LPQ), Local Binary Pattern (LBP). Non dominated Sorting Genetic Algorithm II (NSGA-II) feature selection algorithm applies on extracted features to have faster learning process and finally selected features are sent to neuro-fuzzy and neural network classifiers. Proposed method is evaluated with the benchmark databases such as, Eurecom Kinect Face DB, VAP RGBD-T Face, JAFFE, Face Grabber DB, FEEDB, and CASME. Also, the proposed method is compared with other similar methods and Convolutional Neural Network (CNN) method on mentioned databases. The results are really satisfactory, and it indicates classification accuracy improvement of proposed method versus other methods. |
