Key measure frame techniques based on image difference

In the real-time video environment,what camera collected is image serial of facial expression.Face recognition needs to perform expression recognition of every-frame image collected which requires a lot of count so as that it can not meet the real-time needs.Therefore,key frame techniques based on image difference is put forward,which uses small wave to convert the face recognition and makes the serials of neutral expression image frame and measure image frame construct neutral expression image and measure image re-spectively,then the Euclidean distance expression spring image and neutral expression spring image is calculated.According to the change tendency of Euclidean distance,the system can check the key frame image which is in the lagest state,from video serials,which will be used in later feature measure.The experiment indicates combining key frame measure calculation and matching recognition cal-culation based on spring pattern can satisfy the real-time requirement,and get the desired recognition rate.