Accurate detection of human face position in the environmental images using Gabor wavelet transformations
Vahid Ghods, Mohammad Karim Sohrabi, Mehdi Joulaeian · 2017 Computing Conference · 2017
Face Detection means to determine the presence or absence as well as positions and range of all humans' faces in the image. In general, face detection methods can be decomposed into moving or stationary detections in color images or with gray levels. In this paper, a new approach is developed for face detection in stationary images with gray levels including a certain size, and maximum angle of 30 degrees and resistant against light variations in desirable time. The image features are extracted using Gabor wavelets algorithm and will be sent for testing into Perceptron Neural Network (PNN). In the training phase, 72 face images and 72 non-face images are used through the face-rec database. In addition, for each face or non-face image, its mirror image, rotated image with the angles of 5, 10 and 15 degrees in positive and negative directions and images with one pixel shift in all four directions are deployed within training set in order to reduce sensitivity of the network. Entirely, the results of simulation experiments demonstrated the effectiveness of the proposed method.