Abstract: The high resolution remote sensing images can play an important role in the 1:50000 geological mapping in the alpine-gorge area with high altitude, deep cutting and poor crossing conditions.However, the remote sensing image products provided by the current market cannot match the working base maps such as field position information and topographic maps, and their geometric accuracy seriously restricts the field investigation of geological mapping.In order to solve this practical problem, a set of geometric precision correction techniques for high resolution remote sensing images is proposed by using high resolution remote sensing image SPOT7 and exemplifying Saishiteng Mountain in the northern margin of Qaidam Basin in Qinghai-Tibet Plateau as an example.Based on the fusion registration of high-resolution remote sensing images, this method innovates the semi-automatic man-machine interactive ground control point selection method, as well as the techniques of mass selection and spatial uniform distribution of typical feature control points.Through repeatedly checking and improving the geometric accuracy of corrected remote sensing images, the geometric precision correction of high-resolution remote sensing images is realized.The corrected remote sensing images give full play to the leading role of remote sensing technology, greatly improve the efficiency of field geological mapping, and can provide technical reference for geometric precision correction of high resolution remote sensing images in complex terrain areas.