Research On The Detection And Positioning Of Standing Faces In The Intelligent Recording And Broadcasting System

With the promotion of intelligent education,the intelligent recording system has been widely used.The application of sports face detection in intelligent recording and broadcasting systems has become a new topic in the field of intelligent identification.Because the research on moving face detection involves the extraction of moving objects in video and face detection and positioning two aspects professional knowledge,which make the research more difficult,the current research is still not fully meet the actual needs.Based on the deep analysis of the demand and of intelligent recording and broadcasting system,combined with current research status and needs at home and abroad,this paper proposes a moving face detection and positioning algorithm based on the improved frame difference method.The main work of this article is as follows:(1)Segmentation of the motion area.Firstly,on the basis of analyzing the defects of the conventional algorithm,a two-frame difference circulative difference method is introduced.The segmentation of the motion area is successfully implemented by this algorithm.In the extraction of candidate face area,this paper makes full use of morphological operations,removes image noise and merges scattered small areas,and has achieved satisfactory results.(2)Active area image preprocessing.Firstly,keeping the motion area with the minimum enclosing rectangle.Then,the compensation method based on the reference white is used to compensate the light in the rectangular area.And then,the skin color threshold method based on the RGB color space is used.Detecting skin color in a rectangular area,and delete the area where there is no skin color exists,thereby reducing the calculation amount of subsequent screening operations and improving the execution efficiency of the algorithm.(3)Standing face detection and positioning.Classification detect the candidate's face area according to the size.In the smaller areas,using hair details,face ratios,and proportions of skin area excludes non-face areas and preserves larger areas;human eye detection is added to larger areas.When there are multiple candidate faces,the areas with human eyes are retained.Then recording the coordinates of face area.In addition,due to the different depth of field captured by the single camera,the area is divided according to the actual situation of the classroom,and an auxiliary item of angle is added to analyze the coordinate information of the face area of the candidate,to determine the standing face and position the mark to stand on the face.The "segmentation-search-partition analysis" mode adopted in this paper effectively solves the drawbacks of previous face detection algorithms.Through many experiments,the proposed algorithm has achieved good results.