Detection And Tracking Of The Main Symmetry Axis Of The Face

In the current society,moving object detection and tracking in video is one of the key topics in computer visual field.It plays an important role in many aspects via combining image processing and pattern recognition.With the increasing requirements from complex environmental applications,how to detect and track the moving targets accurately and intelligently has become an urgent proposition of our time to get a breakthrough.The research on tracking of facial features is usually divided into two main parts:one is the detection and extraction of facial features;another is the tracking of facial features.To track the target effectively,it is a must to be able to extract the facial features effectively;otherwise there may be problems such as loss of target,tracking failure and so on.This paper first introduces the background and significance of facial feature research,and the research status in this field at home and abroad,expounds the main problems in this field,and based on it,develops the research of face feature detection and tracking.In the aspect of facial scope detection,this article introduces two commonly used methods,namely facial feature detection based on skin color and the AdaBoost method.Firstly,the principle of the two algorithms is introduced in details,and then the experiments are done to compare the results.Experiments show that facial detection based on AdaBoost is better than based on skin color.Based on this,a method of main symmetry axis detection based on clustering analysis is used for detecting the direction of the extracted faces.In the face tracking part,the CamShift algorithm is adopted to realize the tracking,which is a tracking method based on histogram model.In order to track the target better,the basic CamShift algorithm is improved,and puts forward a tracking method based on the integration of texture and chromaticity,creating histogram model about the joint of chromaticity and texture,and designs a tracking algorithm based on chromaticity-joint texture feature.The algorithm builds a model with the chromaticity and texture features of the face,and has certain robustness;making full use of the target facial detection results,realize automatic face tracking.Finally,the improved CamShift algorithm is introduced into the tracking system to verify its effectiveness.The results show that the system can meet the real-time tracking requirements and is robust against background interference.