Abstract: Camera calibration is one of the fundamental issues in computer vision.At present,camera self calibration techniques based on active vision fall into two species: the first one requires camera to translate along three dimensional directions many times,so that the camera intrinsic parameters can be solved.The second one requires camera to rotate.Because the second requires camera to have no translation while it rotates,it is very difficult to fulfil this requirement in practice.In this paper,a new camera self calibration technique is proposed.It belongs to the first method.Unlike the former methods,our approach doesn′t require camera to translate along two orthogonal directions many times.If the camera is controlled to translate three times and the translation vectors are linear independent,the camera intrinsic parameters can be solved through pseudo inverse matrix theory.The analysis shows that the solution is existing and unique,and the important result has been proved.Tests with synthetic data and real images indicate that the algorithm is robust and practical.