Abstract：To accomplish spatial rectangular coordinate transformation with big rotation angles, the improved coordinate transformation model is proposed. This model is based on scale parameter estimation using the matrix singular value decomposition (SVD). Three estimation methods, including direction cosine, unit quaternion and Rodrigues are presented for the improved coordinate transformation model. Finally, simulated and practical experiments of coordinate transformation with big angle are tested to compare and analyze the adaptability of the three methods using the total number of singular points, the condition number of normal matrix, the number of iterations and the RMS, etc. Experimental results show that the improved model is better than the extant seven parameters model, that the SVD solution of rotation matrix is superior to unit matrix as an initial value, and that the direction cosine method is preferable to the other two methods.
