The generation of the phase compensation factor, the range compensation factor and the azimuth compensation factor are the key in the real-time CS (chirp scaling) algorithm of synthetic aperture radar. An existing CS algorithm featured by the invariance of compensation factors in a region is improved. For the unified compensation factor in units inside the update step size, the compensation factor corresponding to the average frequency of all units in a region is used to replace the compensation factor corresponding to the first unit, which makes the final compensation result using the improved algorithm is more uniform than that of the original algorithm. The computation load of the improved algorithm is equivalent to that of the original algorithm. Through Matlab simulation, the imaging performance of the improved algorithm is improved. In order to facilitate the generation of FPGA code, the Simulink tool is used to build a model for generating three compensation factors used in the improved algorithm. The output code of Simulink model was loaded into Vivado software. The compensation factor outputted by Vivado is compared with the accurate compensation factor outputted by Matlab. The output generated by Vivado meets the accuracy requirement, which verifies the correctness of the built Simulink model.