A camera self-calibration algorithm based on image transformation is presented in the paper. The differences among all the elements of the image of absolute conic in magnitude are so large that solution is extremely sensitive to noise. To challenge the problem,the algorithm is proposed which includes two steps: Firstly,the camera intrinsic matrix is estimated. Secondly,with the estimated intrinsic matrix of the first step,all the elements of the image of absolute conic become in the same magnitude by image transformation. The theory and experiments with both simulation and real data demonstrate that the algorithm can lead to an enormous improvement in the stability and robustness of the result.