Research On Multi-Pose Face Location And Recognition Method

Though there are a lot of difficulties in the field of automatic face recognition, potentialvalue is always encouraging researchers untiring efforts all the time in theoretical researchand practical application, at the same time, it prompts the research of other relevantdisciplines too. This paper studies face location and recognition on the basis of face detection.The contributions in this paper are described below.Firstly, in order to see the number of pixel in each grey level, the curve of the gray image,the distribution situation of the valley and peak, this paper introduces the concept of theapproximate histogram.Secondly, propose an improved and rough facial feature location strategy. This method ismainly based on changing total mark projection order and direction of the facial characteristic,and makes use of valley bottom characteristic in the curve graph to look for and locate theposition where the characteristic of appearing department is lighted. After rough location, itcombines the variable template and priori knowledge of face morphology, and directs us tothe accurate localization of characteristic.Finally,introduce the sub space of characteristic to face recognition, further strengthen itsapplication environment, and use this principle in the course of face recognition correlatingwith posture.