Effective distance measure for unusual facial expression detection of human face images

摘要

In this paper, a simple method for the facial expression detection to identify the normal and unusual of expression variations of the human face is proposed. The approach uses the binary edge image to present the shape of facial expression. The modified Hausdorff distance measure is adapted to our approach to compare between the unusual and normal face expression. In our algorithm, it just needs to use three face images per subject for training, which include two normal facial images and one unusual facial image to obtain individual thresholds for human face detection, respectively. Therefore, the proposed technique does not require a large number of images for training.