In the noninvasive imaging of the samples such as biological tissues with various forms of OCT, the dispersion effect would seriously influence the image contrast and resolution. Compared with the invariant dispersion, the sample-induced dispersion is much more difficult to be compensated because it is depth-dependent. In this work, we proposed a time-frequency analysis method to compensate for the depth-dependent dispersion without the pre-knowledge of the dispersive property of the tissue. With the Radon-Ambiguity transform and the fractional Fourier transform the dispersion amount corresponding to the specific depth could be calculated and compensated. The validity of this algorithm was first demonstrated by the simulation of the compensation of the water-induced dispersion at the micron resolution. Furthermore, the TiO2 phantoms and the fingertip of a healthy human volunteer were measured with the common-path micron resolution optical coherence tomography system. The experimental results revealed that the image resolution acquired by the method proposed could be improved by a factor of 2–3. This work can be helpful for dispersion compensation in other optical fields.