当前地震数据成像处理中，多次波被视为规则噪音而进行压制。表面多次波是海洋地震资料中能量很强的波现象，目前各种表面多次波压制方法几乎都有诸多的假设条件，致使实际资料难以取得满意的表面多次波压制效果。表面多次波携带着能量很强的地下反射界面的有效信息，本项目基于多次波形成的物理原理、双程声波方程的逆时偏移方法，研究基于最小二乘逆时偏移的表面多次波成像方法。对表面多次波高精度分离方法、成像条件、速度模型精度、CPU/GPU并行加速算法、串声干扰压制、成像精度和效果进行深入研究，以形成高精度、优化的最小二乘逆时偏移的表面多次波成像算法。表面多次波成像可有效增强地下介质的成像范围、覆盖次数和照明的均匀程度。表面多次波成像方法的研究，是当前复杂构造油气藏和岩性油气藏勘探所要求的保幅处理技术的丰富和发展，更为重要的是为海洋地震资料中普遍存在的表面多次波的有效应用探索了研究方向。

So far in almost all seismic data imaging, mutiples have been regarded as regular noises which are suppressed in prestack preprocessing. Strong reflection energy is carried by surface-related mutiples in seismic data, especially in offshore seismic data, and almost every kind of algorithm of surface-related multiple suppression has so many assumed conditons which means it is very difficult to get satisfied multiple suppression result for fielddata. It is well known that surface-related mutiples carry high-energy efficient information of the subsurface, based on the theory of physical seismology of surface-related mutiple and prestack reverse time migration(RTM), the least-squares RTM of multiples imaging algorithm is propoed in this project, and further study on separation of surface-related multiple from the field dagta,imaging condition, velocity model precision, CPU/GPU parallel speed up algorithms, suppression of crosstalk artifact,imaging precision and imaging effect of least-squares RTM of multiple imaging, and the high precisional,pratical surface-related multiple imaging algorithm is given when this project finishes.Imgaing of surface-related mutiples can improving the imaging quality,and increasing the extend of imaging and illumination.The research of surface-related multiple imaging aproach, is the richness and development of the true amplitude seismic processing algorithm which is required for complex structural hydrocarbon reservoirs and lithologic reservoirs exploration, and more important, this project explore the research direction for effective application of surface-related multiple which exists in offshore seismic data.