随着计算机技术的发展，实时成像处理器技术的成熟，不但处理数据量越来越大，而且对成像质量要求也越来越高，实时成像处理中需要有自动聚焦算法。但是由于实时成像处理器的特殊结构和实时性要求，需要算法的计算量不能太大、不能太不确定。因此并不是所有的算法都适合用于实时处理。\r\n主要介绍Map Drift算法的原理和实现步骤以及在实时成像处理器中的应用。文中对适合于实时成像处理器中应用的公式进行了推导，得到频域二次相位误差系数和相关峰位置的线性关系，并利用该关系在频域进行相位补偿，有效缩减迭代次数。利用一种插值来计算相关峰位置的方法，消除了离散化计算相关峰位置所带来的量化误差的影响，并在迭代过程中改进二次相位误差系数和相关峰位置关系的系数，对减少迭代次数，提高估计精度有一定帮助。最后给出仿真结果，验证算法的正确性。

With the developing of the computing technology,the technology used in the development of the real time SAR processor is mature.Not only the larger and larger data needs processed,but also the better and better image quality is demanded.The auto-focus algorithm must be used in the real-time SAR processor to improve the SAR image quality and adjust the parameters in the image reconstruction.However due to the special architecture and the real time demand of the real-time processor,the computing load cannot be a random quantity and too large.So that it is not all algorithm fit for real-time processor.In this article,the Map Drift Algorithm(MDA) is used for SAR auto-focus processing.The theory and method of the Map Drift Algorithm are introduced,and its application in real-time SAR processor is discussed.A linear relationship between the secondary phase error in the spectrum of the SAR signal and the position of correlation peak can be confirmed.The iteration times of MDA and also the computing quantity can be reduced effectively using this relationship.An interpolation was introduces to calculate the position of correlation peak in our paper,therefore the errors brought by discrete processing is greatly improved.It also make the estimation of the Doppler parameters more precise.The computing quantity and the image quality of our algorithm are shown as a test result with the Radarsat's data.