In computer vision head pose measurement system, large space is needed for the installation of coded feature points. To solve the problem, a new method is put forward for the measurement of the Line-of-Sight (LOS) with a simple structure as well as smaller installation space. The method uses the positioning cameras to capture the picture of the coded feature points on the helmet, and a computer for image calculating and analyzing. The head pose is obtained finally with the Orthogonal Iterative (OI) algorithm. Based on the characteristic of the feature points, we conducted a series of simulation analysis. The simulation results proved the rationality and the validity of the method, and showed that a high-precision LOS can be obtained by using the coded feature points distributed within a smaller space.