Abstarct: Estimation of a precision velocity model has been one of the challenges faced by seismic processors. The presence of structural complexity and extreme topography makes it difficult to estimate velocity model and creates unqualified seismic images. Different methods are used to estimate velocity model, which have strengths and weaknesses.Time migration methods are resistant to velocity variations and do not accurately determine dip and position of geological targets. Depth migration methods are very sensitive to the velocity variation, and use of the wrong velocity model creates images that mislead the interpreter. Therefore, it is very important to use a suitable velocity model for depth migration.In this thesis, the semblance method was used to obtain initial velocity model. Then, using the coefficient, new velocity model would be obtained. Using new velocity model, the Kirchhoff depth migration process was performed. Kirchhoff's algorithm has been used in the proposed method due to advantages such as flexibility in calculations and handling of slopes. After creating PSDM sections, the CIG analysis is done. In this way, the coefficient which gives the best continuity in events, would be selected for velocity updating in that CDP gather. After selecting suitable coefficients, the coefficients will apply to initial velocity model. After velocity model has been updated, the PSDM process performed for last time, and final section is achieved.The proposed method was implemented on a synthetic and two field data from Gorgan plain and Anar region. Gorgan plain data has a smooth topography and complex structure, but Anar region data has an extreme topography and complex structure that have a high velocity laxyer named Gachsaran Formation, which makes it difficult to seismic imaging.In the synthetic data, a high velocity laxyer was considered, and the purpose of imaging was determination laxyers beneath the high velocity laxyer, which was well-researched, but also has some artifact that are related to weakness of migration algorithm and the presence of a high velocity laxyer. In the Gorgan field data, the quality of the final section has increased, which can referred to change the slope of the laxyers, increasing the amplitude of the reflectors, and increasing continuity of laxyers. In the Anar field data, which has a complex structure and extreme topography, the seismic image obtained by conventional method was completely inefficient and noisy. The proposed method was implemented on the Anar data. Result from updated velocity model shows improvements in image quality and more horizons were imaged.