Dissertation
Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device

Face Detection Technology Based on Adaboost Algorithm

Author LiSi
Tutor HouJianHua
School Central South University for Nationalities
Course Electronics and Communication Engineering
Keywords Face detection Adaboost algorithm Haar features Classifier
CLC TP391.41
Type Master's thesis
Year 2013
Downloads 13
Quotes 0
Download Dissertation

As an important branch of target detection, face detection technology is a keysubject in the area of artificial intelligence and pattern recognition. It has beenwidely applied in such fields as personal identification, visual communication,human-computer interface, environmental monitoring, video surveillance, educationsystem, etc. With the development of intelligent computing technology and theintroduction of many new methods and new techniques, the effect of face detectionis getting better and the face detection speed is becoming faster.Face detection is the process that for any given image, one can determinewhether a face exists or not by searching the image through certain strategies. If aface is recognized, the face detection system will return the information aboutposition, size and posture of the searched face. Face detection is an important part offace recognition and image compression. This dissertation focuses on the Adaboostbased ace detection algorithm, and the main work and contributions are in thefollowing four aspects.(1) Firstly,the significance and background of the topic is introduced, thecurrent research status of face detection, the face detection methods and theevaluation criteria of face detection system are also provided.(2) Secondly,the theses has analyzed the principle of Viola-Jones detector. Thetheoretical basis and development process of Adaboost learning algorithm has beenexpounded.(3) The training and detection processes of face detection system have beenanalyzed in detail. The theses has elaborated the construction and training process ofHaar features, weak classifier, strong classifier, and cascade classifier. And a facedetection system has been constructed.(4) The traditional detection method are compared with its amplificationcounterpart to illustrate what influence each has on the performance of face detectionsystem. The result shows that the traditional detection method is superior indetection rate and error detection rate. At the same time, regarding with the issueabout degradation phenomenon, we have compared the suppression effect of traditional Discrete Adaboost algorithm with that of he improved Adaboostalgorithm. Experiment results show that the improved algorithm can better avoid thedegeneracy phenomenon’s damage to the detection system.Finally, the theses has proposed that by combining different detectionalgorithms, the speed and accuracy of face detection can be further increased, and itwill help to facilitate face detection along the direction of generalization, automation,real time and accuracy.

Related Dissertations
More Dissertations