Research On Algorithms Of Face Detection And Recognition Across Cameras

Nowadays,the international situation is not stable,and countries pay more and more attention to security issues.In September of 2016,the G20 Summit is held in China,which is called “The most tightly controlled G20 Summit in the history”.The G20 Summit used face recognition technology particularly to do security surveillance in key areas so as to ensure the smooth progress of the summit meeting.In daily life,people also pay more and more attention to security issues,and all streets and lanes are equipped with cameras.The “Skynet monitoring system” created by public security organs is the powerful safeguard of the city security.With the advent of the intelligent era,the traditional manual processing of monitoring video mode is difficult to adapt to the needs of the people,and people are gradually realizing the real intelligent monitoring system.In this paper,a cross-camera face detection and recognition system is proposed,which realizes the detection and recognition of the same face in different cameras as well as the dynamic face tracking.The system mainly contains three parts: face recognition by independent camera,face tracking by independent camera and face recognition by cross camera.The article focuses on the improvement of the algorithm in the system.In the face detection part of the independent camera,the method of gradually narrowing the detection range to achieve the final detection is adopted.The first step is to do a motion detection of the target and to detect the area of human body.Next step is to locate the human skin area through the skin color feature.The third step is to realize the face detection by AdaBoost algorithm.This method can not only improve the correct detection rate,but also effectively reduce the missed detection rate.In the face tracking of the independent camera,the Camshift algorithm is optimized for real-time tracking of the human face,and Kalman prediction mechanism is applied in the algorithm,thus improving the tracking effect of the obscured target.Experiments show that the tracking of the target can be more accurate by the combination of the two methods.As for the part of target recognition by cross camera,the paper combines Gabor wavelet and PCA to fulfil the recognition task of face image.This algorithm uses the Gabor filter to extract the features of the tracked face image.Althouth Gabor algorithm can effectively extract the face feature information,it results in a lot of redundant features,which increases the amount of calculation greatly.The PCA algorithm can effectively reduce the dimension of the data.The combination of the two algorithms can effectively extract the features of the face image and reduce the dimension of the feature data as well as simplify calculation significantly.The cross-camera face detection and recognition system in this paper is used to obtain human face information and to complete face recognition in other cameras in complex circumstances,realizing the function of face detection,tracking and recognition under different dynamic backgrounds.The system function test and the experiment results show that the cross-camera face detection and recognition system in this paper is with complete function,high accuracy,and better in real time,and it can meet the requirements of the system.