Abstarct: Steeply dipping structures in complex geological media have a great capability for petroleum bearing structures. They are becoming more and more important in recent years. Imaging of such complex geological structures is not performed properly using a set of ray-tracing and, wave-field baxsed, one-way wave equation imaging methods. Thus, reverse time migration (RTM), as a two way wave-field baxsed method is used to image these steeply dipping structures. In addition to its superiority, RTM faces to some challenges. In this thesis, it has attempt to improve the RTM imaging method by introducing a new wave-field extrapolation technique named as Leapfrog-rapid expansion method (L-REM) and also to propose a new imaging condition to suppress the low frequencies artifacts of the RTM method. Hence, the possibility of using a symplectic integrator for wave-field extrapolation baxsed on the Leapfrog (L) and rapid expansion method has been discussed and a new scheme (L-REM) was proposed to extrapolate the wave-field and its derivative. The obtained results of this new scheme on numerical examples indicate that using of this method for estimation of wave fields and their derivatives are in high level of accuracy and stability comparing to the similar methods. Furthermore, it is not limited by time steps and dose not faces to the dispersion and stability problems. The results also show that L-REM is not only accurate for small time sampling steps but also is more accurate for large time steps comparing to the other methods. In the next step, it has tried to improve the imaging condition as the heart of RTM method to obtain a high quality image of steep dip structures and to suppress its low frequency noises. The Poynting vectors were then calculated by using the wave-field and its derivative results. These vectors were used to separate the up-going and down-going components of the wave-field. They also used to calculate the reflection angles as a basis for a proposed weighting function. After that, a new imaging condition baxsed on the separated wave-field components was provided. This imaging condition could supress a part of the low frequency noises that start to appear for the reflection angle larger than 60 degree. To supress the remains part of the low frequency noises, the aforementioned weighting function was added to the imaging condition to form a complete new imaging condition. baxsed of the direct relationship of reflection angles and noise production, the reflection angle range 61 to 90 degree was divided to a triplet domain and a specific weight was allocated for each sub-domain. Finally, the improved RTM algorithm was tested and validated by using some synthetic different models and a real dataset. The acquired results showed the efficiency and capability of the new RTM procedure for imaging of the subsurface structures in complex geological media. It has also shown that the proposed method is superior than to the others imaging methods of seismic data in imaging of steep dipping structures and handle more properly for the low frequency noises of RTM method. It was also shown this method produce a final migrated image with high quality in each cases.