Research And Implementation On Key Algorithms Of Automatic Face Recognition

The current face recognition algorithms has been relatively mature. Through most kindsof recognition algorithms has been relatively effective on accuracy, they have low timeefficiency. This article tries to improve the time efficiency of the algorithms on the basis ofnot reducing the accuracy of the existing algorithms. The main research contents andachievements of this dissertation are as follows:1. The dissertation researches on the basic theory of image processing, including cannyedge detection, hough transform, image rotation transformation, which are related with facerecognition. At the part of eye location by the method of face image hough transform,semi-circle detection is used by this algorithm other than circle detection, and good resultshave been achieved.2. The dissertation researches the cascade classifier based on adaboost algorithmstructure, using face-haar-features to train the cascade classifier and to detect face regions. Forthe non-face-region which has similar face-haar-features, this algorithm is prone to classify itto face region, this dissertation proposes a method to eliminate the non-face-region by skincolor detection based on YCbCr pattern, and puts forward rapid skin detection to save runtime, to ensure the system in real-time.3. In the process of extracting face image features, the Gabor wavelet features of eightdirection and five dimensions have been extraction. For the problem of so many features, thisdissertation has adopt down-sampling method and PCA algorithm. Experimental results showthat this method has eliminated great many uncorrelated elements and greatly simplified thestructure of recognition system.4. Artificial neural network based on back propagation algorithm(BP-ANN) has beenstructured, by the way, sigmoid unit as perception has been adopted, delta rule has been usedto training network, and the method of adding momentum factor has been employed toimproves the learning speed. In the test of yale face library, the structure achieved fastlearning speed, high recognition accuracy and speed.