Improvement Of Face Detection Algorithm For Video Image And Application

Compared with other organs in the human body,facial information has good biological characteristics and advantages such as safety and stability,which become a hot research topic for domestic and foreign scholars.Face detection as a key part of face image processing,It is of great research significance and value to quickly and accurately detect the number and location of face information.When capturing video images and pictures,there are numerous environmental factors that introduce noise,which affects the accuracy and stability of the face detection system.Therefore,based on the deep research of face detection algorithm,in this paper,we focus on improve and optimize the relevant face detection algorithm.The main contents are as follows:1.Due to the face masking and distortion in video images and pictures.And based on the analysis of traditional skin color model and Adaboost algorithm.this paper proposes a skin color segmentation face detection method based on K-means clustering for face image.The first is the preprocessing of video images and picture,and then convert the RGB color space to YCb Cr color space.Because of the clustering characteristics of skin color information in YCb Cr space,the skin color segmentation of K-means clustering is performed in this space,use the elliptical skin color model to determine the The skin color region information,and the human face is determined by the binary morphological processing and the skin color connected component area in the skin color region;then candidate human face regions are secondarily verified by the Adaboost algorithm.The experimental results show that the skin color face detection through the integration of clustering information has a higher detection effect and good robustness.2.Due to the continuous and gradual process of the video image sequence,the change between adjacent frames is very small,if the detection is performed on each frame of the video images,the complexity and redundancy is increased.Therefore,this paper combines face detection and tracking to reduce the redundancy between video frames and improve detection efficiency.First,an improved face detection system is used to detect face in video images,then track the detected face.When the tracking offset is greater than a given threshold,the face detection system is activated.Relative to face detection,tracking requires less time complexity,the results of tracking affects the efficiency of the test results seriously.Therefore,the feature points are integrated into the tracking system to Improve tracking effect.Experiments show that the integration of feature points is more efficient.Experimental analysis shows that face detection with tracking information has less time complexity.