Abstract: For complex spatial distributed high and steep rock slopes, traditional surveying methods still face challenges in efficiency and accuracy. To obtain rock mass structural parameters quickly and accurately for evaluating the stability of rock slopes, a multi-attitude close-range photogrammetry method using unmanned aerial vehicles (UAVs) is proposed. Firstly, the conceptual slope model was established rapidly by measuring key points on site, and the flight route planning and attitude control model were combined to quickly plan the flight routes and camera orientations. Then, by implementing software on the remote controller instead of a PC, the method achieves real-time, on-site, coarse-flight-free, rapid multi-attitude close-range photogrammetry with high-quality image acquisition of high and steep slopes at millimeter resolution. Finally, the method is validated through a multi-attitude close-range photogrammetry experiment on a high and steep slope in Jingyue district, Changchun City, Jilin Province. The feasibility and reliability of the method are assessed based on model accuracy, texture clarity, and orientation interpretation accuracy. The results show that the accuracy of the model in x, y and z directions is 0.018, 0.016, and 0.033 m, respectively. The accuracy of dip direction and dip angle interpretation is 3.1° and 2.6°, respectively. Compared with conventional close-range photogrammetry, the multi-attitude close-range photogrammetry significantly enhances texture details and greatly reduces surface distortion effects. The Chinese PCQI (perceived clarity quality index) values are higher, while the NIQE (natural image quality evaluator) values are lower, indicating better image quality. The result of PCQI and NIQE metrics both showed that the image captured with 45° lateral tilt and 30° vertical tilt provide higher quality than those from other tilt angles. Multi-attitude close-range photogrammetry outperforms conventional close-range photogrammetry, effectively reducing dead angles and mitigating texture distortion, thereby substantially improving model image quality. The model was verified to be reliable by  point accuracy, distance, and orientation interpretation accuracy assessments. The rock mass structural parameters, such as trace lines, orientations, dominant joint sets, and RQD (rock quality designation), interpreted by this method provide reliable foundational data for slope stability evaluation.

Key words: drone, nap-of-the-object photogrammetry, multi-pose, steep slope, flight path planning