Research And Implementation Of Face Mosaic Algorithm

Face mosaic refers to mosaic the existent face part after performing the face detection on a video or an image. Face detection is the primary key aspect of face mosaic,and its task is to confirm whether there are human faces in the image. In recent years,with the extensive application of face detection and the development of computer technology, the face detection algorithm is endless, the cascade Adaboost algorithm stands out in many face detection algorithm with its high efficiency and fastness. The cascade Adaboost algorithm is a fast face detection algorithm proposed by Paul Viola and Michal Jones in 2001. Its presentation plays an important role in the development of face detection technology.In this paper, the cascade Adaboost algorithm is analyzed in detail, the main research is as follows:(1) In this paper, several common face detection methods are classified, and the methods and steps of face image preprocessing are summarized: Gaussian denoising,gray scale processing, scale transformation and histogram equalization.(2) In this paper,the principle and training process of cascade Adaboost algorithm are described in detail. At the same time, this paper the concept of Haar feature, weak classifier and strong classifier etc. of Adaboost algorithm is introduced. In short,Adaboost algorithm is a weak order lifting algorithm. It is possible to reduce the false detection rate of the trained classifier to infinity close to zero by training enough samples and enough features.(3) Aiming at the problem that the processing speed of the existing face mosaic method is slow,this paper proposes a dynamic adaptive adjustment of Haar detection function threshold method.(4) In this paper, efficient and fast face detection and mosaic processing becomes a reality, selecting the extended Haar feature combined with OpenCV open source computer vision library and AdaBoost face detection algorithm.In this paper, the dynamic adaptive adjustment of Haar detection function threshold method increases the face detection rate to 95.47%.Its picture processing time is about 2.4ms / frame. The method achieves a fast and efficient face detection and mosaic processing. At the same time, the algorithm has low computational complexity and good stability, and can be widely used in face mosaic processing.