The Research Of Face Detection Based On Skin Color Information And Adaboost Algorithm

Human face detection means that for a given image or video, to determine whether it contains face regions, if so, determines the number, the exact location and the size of all the faces.Human face detection is not only a necessary precondition of face recognition, expression recognition technology, face tracking, but also, it plays an important role in applications like in the intelligent human-computer interaction, video conferencing, intelligent surveillance, video retrieval and so on. Therefore, face detection technology attracted widespread attention in pattern recognition, computer vision, human-computer interaction and other fields.This paper discusses an improved technique for detecting faces in multi-pose color images with complex background, the main research work embodied in the following three aspects:(1) The research on the color-based face detection technology is present. First, Gaussian model in the YCbCr chrominance space is built to segment the non-skin-color pixels from the image. Then, the characteristics of the possible face blocks are studied, so we can extract candidate human face regions. This process lays a solid foundation the following detection step. The experiments show that this system obtains good detection performance in complex background, with high robustness for illumination, expression and other changes.(2) A new algorithm based on improved Adaboost for face detection is proposed . Adaboost is a learning algorithm for constructing accurate classifiers. It can obtain a strong learning algorithm by combining a series of weak learning algorithms, the ideal classifier which has high recognition accuracy is obtained by training samples. This paper analyses the issues of overfitting in training process and time-consuming problem in detection process,then improves the weigh-update rules on a certain degree.Experimental results show that the modified weigh-update rules can effectively avoid overfitting and improve the performance of classifier.(3) A novel face detection method combined skin color detection and Adaboost algorithm is proposed in this paper. The new scheme is able to detect faces with high detection rate and low false acceptance rate ,it can performan better than skin color detection and Adaboost algorithm, In this way the rate of false positive is reduced and the processing speed is greatly increased, improving the overall performance of the detection module.The complete system is tested on a variety of color images with different scales, various poses, different expressions, lighting conditions and orientation, it is also compared with some conventional techniques. Experimental results show the proposed system can effectively used in multi-pose and multi-face color images with complex background. It obtains competitive results and good detection performance, which illustrate the effectiveness and availability of the algorithm.