The traditional vision measurement model has difficulty in guaranteeing the accuracy of measurement in the depth of field. And in this way, high-precision measurement of large components parts in three-dimensional large scale space can hardly be realized. To solve this problem, a binocular measuring method based on 3D image distortion compensation is proposed. Considering the rule of image distortion in the three-dimensional space, and combining with binocular vision measurement principle, a new binocular vision measurement model, based on 3D image distortion compensation, is established in the paper. And the model is based on the rule of image distortion in the three-dimensional space, and combined with binocular vision measurement principle. Besides, a new calibration method is proposed. This method is for the distortion parameters of the model and the intrinsic parameters of the cameras. Experimental results show that the proposed binocular vision measurement method in this paper is much more effectively than the traditional method. The results indicate that the proposed method largely improves the measurement accuracy under the condition of large depth of field. Meanwhile, this method also significantly improves the measurement accuracy in the three-dimensional space.