Based on image and elevation data, accurate positioning can be achieved without relying on GPS, which has high practical value in many scenarios. The position process includes image skyline detection, elevation data skyline detection, skyline feature extraction and skyline matching. However, due to the influence of natural environment and the accuracy of elevation data, position matching has many difficulties and challenges. To solve these problems, this paper proposes a pyramid fuzzy slope feature based on sliding window, which consists of average height feature, multi-layer slope feature and segment feature. Experimental results on different precision elevation data show that this feature can mitigate the adverse effects caused by noise and distortion of elevation data in a certain range, accurately distinguish the target elevation data skyline from the non-target elevation data skyline, and has good accuracy and stability as well.