Firstly

the methodological framework of the head pose estimation based on interframe information guiding is proposed. Aiming at the problem that histogram of gradient（ HOG）-based face detection algorithm has high accuracy but slow speed

propose a strategy for reducing search range by introducing face frame of the previous frame. Aiming at problem of feature points drifting

a method of pre-rejecting poor points according to attitude angle of the previous frame is proposed. Finally

use the Eye Diap dataset to verify the optimization method.Experimental result shows that the proposed optimization method of head pose estimation based on interframe information guiding can effectively improve the real-time performance and accuracy.