| Face detection and recognition technology is one of the most significant and practical topics in fields of pattern recognition and computer vision. On the basis of the deep comparison and analysis of current popular face detection and recognition technologies, this dissertation discussed how to improve the accuracy and speed of the facial detection and recognition system, at the same time guarantee that the system has a strong robustness. Finally the author realizes a facial detection and recognition system.This thesis focused on the follow contents:In the discussion of facial detection: Combine face skin color model with auto-adapted threshold selection method, use this threshold to realize accurate segmentation of the skin color area, after that, the follow processes were carried on in order to improve the image's quality: first using average value filter to delete the image noise, then using some morphological image processing methods such as erode, dilate to cut the bridge areas which connects different skin-color regions. Our experiment shows that it's an efficient way.In the discussion of facial recognition: do the Discrete Cosine Transform to origin image, and then select the low frequency part of Discrete Cosine Transform Coefficient to be the observation vector of Hidden Markov Model for training. Modeling for 2D face image through 2D Hidden Markov Model is more efficient to reflect the individual differences. Our experiment shows that according to this method, we are able to obtain high recognition accuracy.The massive testing and experiments for the facial detection and recognition system completed by us show that, in the field of facial detection and recognition, the system has a very strong robustness to the light condition as well as changing of expression and face attitude. Under various complicate conditions, it maintains more than 96 percent rate of detection and identification, as well as high recognition speed. |
PDF Full Text Request