In swept source optical coherence tomography,the maintenance of imagi ng resolution in a wide range of depth is affected by the dispersion which is brought by the system and the samples.In this manuscript,a depth dependent compensation algorithm is reported,which is used to accurately compensate the dispersion mismatch at different depths of the sample.The main i dea of this algorithm is that segments the sample image according to different depths in the frequency domain of interference spectrum,and then uses the iterative compensation method based on parabola to c ompensate the dispersion at all depths of the sample.The algorithm is independent of the prio r knowledge about samples and,as such,can be directly applied to the spectrum obtained by swept source optical coherence tomography.Through the experiments on different samples such as cover glass heaps and damaged KDP crystal in a swept source optical coherence tomography system with t he detected depth of 5.720mm,it is proved that the depth dependent algorithm is able to effectiv ely improve the axial resolution within the full depth of detection and make it close to the theoretic al value,at the same time,it has good adaptability for different types of samples.