To analyze the stability and determine the support form of high steep slopes, it is important to con? firm the occurrence and location information of rock discontinuities. In order to solve high risk and low efficiency ex? isting in the traditional survey method, this paper proposed the new survey method to semi-automatically identify the discontinuities in slopes based on nap-of-the-object photogrammetry and a clustering algorithm. Firstly, M210-RTK was used to obtain high resolution digital images; the structure from motion algorithm was used to generate the de? tailed 3D models and 3D point clouds. Then, through the K-Nearest neighbor and principal component analysis algo? rithm, the set of coplanar point clouds was selected to determine the boundary range of discontinuities. Finally, the best planar equation of coplanar point cloud is fitted based on the least-squares method, and the normal vector direction of the planar equation is the basis for calculating the occurrence parameters. Validation experiments show that the accuracy of the digital surface model is better than 2cm in any direction, and the error in dip and dip direction are better than 3° and 2°, respectively. This method identified rock preferred discontinuities in the abandoned mining area of Changsha, and the influence of preferred discontinuities on slope stability was analyzed, providing the vital foundation for the slope evaluation and treatment.