Abstarct: The results of post stack processing and imaging of reflection seismic data are strongly dependent to the quality of stack section which is an approximation to zero offset section. To get a more close to zero offset section, the CMP records must be stacked by a proper velocity spectra. Therefore, the velocity analysis is one of the most important issue in exploration seismology. Semblance, the most common coherence measure, has poor resolution. Increase the resolution of the semblance velocity spectra causes the accuracy of estimated velocity for NMO correction and stacking is improved. The reason for low resolution of semblance method is little sensitivity to velocity changes. Thus when we add terms that have more sensitivity to velocity and time changes to semblance, the resolution increases. In this thesis, we present a new weighted semblance method that ensure high resolution velocity spectra. To increase resolution of semblance spectra, we introduced two weighting functions baxsed on the first to second singular values ratio of time window and the position of seismic wavelet in time window to semblance equation. This two weighting functions enhance the resolution of velocity spectra in velocity and time direction, respectively. Semblance, also unable to pick velocities in area with class II AVO which data experience a server polarity reversal. To enhance resolution of velocity spectra in such area, we applied our method on AB semblance. Finally, we test our method on both synthetic and real data. The results indicate that new proposed weighted semblance method provides higher resolution than conventional semblance and can separate the reflectors which are mixed in semblance spectrum. This method also shows good results when our data is contaminated with high noise ratio or in events with high interference and in presence of class II AVO anomalies is able to recognize and determine the velocity of such data.