Research On Face Detection Algorithm Based On Complex Network Particle Swarm Optimization Algorithm And Adaboost

With the rapid development of computer technology, face detection technology has attracted more and more attention by people as an important branch of computer image processing and artificial intelligence technology. The AdaBoost algorithm based on the statistics is one of the methods with real-time face detection and a high detection rate currently. However, the disadvantages of the basic AdaBoost algorithm are that it takes long time in sample training process and has a high false alarm rate in the detection process. To deal with these problems, the face detection algorithm based on AdaBoost is studied thoroughly. By introducing some characteristics of complex network theory to improve the network topology structure of the particle swarm, an adaptive particle swarm algorithm based on complex network is proposed, and the proposed algorithm will be used to improve the basic AdaBoost algorithm. The simulation results indicate that the proposed algorithm can effectively shorten the sample training time and improve the face detection rate.The primary contributions of this thesis are listed as follows:(1) The disadvantages of particle swarm optimization (PSO) algorithm are that it is easy to fall into local optimum in high-dimensional space and has a low convergence rate in the iterative process. To deal with these problems, an adaptive particle swarm optimization algorithm based on directed weighted complex network (DWCNPSO) is proposed. Particles can be scattered uniformly over the search space by using the topology of small-world network to initialize the particles position. At the same time, an evolutionary mechanism of the directed dynamic network is employed to make the particles evolve into the scale-free network when the in-degree obeying power-law distribution. In the proposed method, not only the diversity of the algorithm was improved, but also particles’ falling into local optimum was avoided.(2) For Haar grayscale characteristics of its own limitation and long training time problems in traditional AdaBoost algorithm, an improved algorithm is proposed and applied in face detection. In the training feature selection, the traditional exhaustive search strategy is taken place by adaptive complex network particle swarm optimization algorithm in sample training process of AdaBoost algorithm. At the same time, the update rules of weights of the samples have been adjusted based on Misclassification cost coefficient, which avoid the phenomenon that weight of the distortion. Then by changing the weighted parameters of weak classifier for solving formula, guaranteeing the improved algorithm can get low false alarm rate. In the end, the correlation concept about classifier is introduced into the design of the fitness function of particle swarm and different complementary weak classifier are combined, And this can increase the diversity of the weak classifiers and improve the detection rate.