The Research Of The Key Technology Of Face Recognition Under Video Scenario

In recent years, video-based face recognition is one of the most excited researchfield in face recognition. The video-based face recognition has much value in patternrecognition, and the latter is widely used in security, criminal investigation, accesscontrol system, attendance system. Face recognition has a lot of benefits because ofits characteristic of high reliability, non-contact and so on, especially comparing withother biological recognition method. But not only the face Images takes lots of dataspace but also face Images is easily affected by pose and illumination and occlusionunder video. These reasons bring a great challenge to face recognition.The recent research results on video based face recognition have been discussed,based on the classification of the relevant methods. This paper also studies theapplication of sparse representation method in the field of face recognition. The sparserepresentation classifier is very robust to the illumination and occlusion. Because ofits low computational speed, most of business application systems do not consideradopting this method. So the face recognition method based on sparse representationis not completely out of the laboratory. Therefore, the speed and complexity of sparserepresentation classifier have been studied and some solutions have been put forward.The main research results are:(1)The speed of using sparse representation classifierfor face recognition is boosted up to real time level.(2)Image sequences in video iscoded to solve single samples problem. After experimental verification, the proposedmethod not only improved the effect on encountering illumination and occlusion, butalso improved greatly in speed.