以加拿大Radarsat卫星为例，在简要介绍了SAR图像的产品格式及其多级别，多类型的特点并比较了有关遥感图像的几种定位方法之后，指出距离－多普勒（R－D）原理用于SAR图像像素的绝对定位模型的优点，在分析了直接坐标变换所遇到的困难后，提出了两种基于R－D原理的实用化定位方法，并对算法进行了验证，作定位精度作了比较分析，试验结果表明，相对位置法所取得的定位精度达到了目前国际上有关这一领域的研究所取得的精度，效果良好，迭代法所取得的定位精度也在加拿大RSI公司技术文档的规定范围内，具有一定的实用性。

Synthetic Aperture Radar (SAR), as a kind of active remote sensors, is a system that can scan and image earth surface objects in a high resoution using its EM wave. Ever since the late of 1970s, SAR as a way of earth observation has progressed to the space borne stage. However, to utilize SAR image product adequately and efficiently, we often need to locate accurately an arbitrary pixel, which is corresponding with the target on the earth surface. Taking Radarsat 1 as an example, this paper first briefly describes the multi level, multi type traits as well as the product format of SAR image. After comparison of several location methods of remote sensing image, R D theory is proved to have its advantage as the footstone of absolute pixel location model of SAR image. In developing pixel location model, the dilemma of employing direct coordinates transformation is pointed out and two practical location algorithms based on R D theory are presented. The test results betrary that the relative position location algorithm can achieve a high accuracy no less than the similar research till to now in the world. The iteration location algorithm can also achieve a high accuracy permitted in the technical document of RSI Co. and therefore has a certain practicability.