Research And Realization Of Multi-algorithm Merging On Face Detection

Face detection is a hot and difficult research area in computer vision and computer graphics. Human face detection is a challenging project because the face can have varying pose, size, skin color, and facial expression. In addition, other factors, such as wearing of glasses, presence or absence of hair, and occlusion can make the appearance of the face unpredictable. Other effects, such as varying lighting conditions and complexity of the scene containing the face can make the generalization of face detection algorithms difficult. In this thesis, I present a novel face detection framework based on a face detection algorithm merged with some advanced algorithms.I firstly survey basic techniques in face detection and face recognition, then I present a face detection algorithm merged with some advanced algorithms. After analysis of advantages and disadvantages of the face detection algorithm, absorbed some advantages of algorithms,and improved these algorithms certainly and finally merged. The face detection consists of three stages, the first stage is the face detection based on skin region growing, namely to locate facial features of people by skin region growing technology to separate skin image from background image ,the second stage is the the cascade face detector is learnt by using Adaboost algorithm, which finds the number of faces and their approximate positions in the image. For speed up, multi-resolution searching strategies are introduced.; the third stage is the searching of detailed local features using geometric matching models like eye model and mouth model with the feature position in the previously face detection as the initialization, which finally finds the desired feature points. During the final stage, the global searching result serves as the geometry constraints on the local feature detection procedure.The framework proposed in this thesis has many applications like person identification, facial expression synthesis and analysis, multi-modal human computer interaction, etc. Experimental results show that this algorithm has some detection effect on face image on complex background.