This work gives an efficient and fast approach to obtain an accurate velocity model for seismic imaging. An accurate velocity model is prerequisite for obtaining the image that represents the true underground geological structure: major evidence to explore for potential reservoirs and indicate hydrocarbons. This method optimizes a differential measure of the image gather flatness by comparing the residual moveout between neighboring traces. It stands out as a more accurate and automated inversion method among a good number of existing approaches. In this thesis, two versions of differential semblance velocity analysis are introduced: NMO-based and Kirchhoff­based. For each version, both the efficiency and effectiveness are discussed. For NMO­based differential semblance, the application is illustrated on both synthetic data set and field data set to invert both ID and 2D velocity models. For the Kirchhoff-based differential semblance velocity analysis, the application is applied on the synthetic data set to invert a constant velocity model, providing the preliminary evidence of the effectiveness and insight for inverting depth varying or 2D velocity models for further research. Overall, this automated inversion method for velocity analysis is a potential technique for the velocity inversion for exploration.