Abstract: Close-range photogrammetry has been widely used in foundation pit monitoring, industrial surveying, cultural relic restoration and other fields. The monitoring accuracy affected by the image greyscale variation remains the solving problem during the long-term close-range photogrammetry. In this paper, the foundation pit monitoring of Yangzijiang River station in Section 2 of Hefei rail transit Line 5 is selected as the engineering background. Firstly, the influence of image grayscale variation on the monitoring accuracy is analyzed. And the optimal grayscale interval is determined. Then, the spectral fusion method is used for the greyscale correcting on the image with non-optimal greyscale interval. The original phase spectrum of the image is fused with the spectrum with optimal greyscale image, leading to the adjustment to the optimal greyscale interval. The matching accuracy before and after the spectrum fusion is analyzed. The experimental results show that the optimal greyscale range of the image of the highest accuracy is 139.0～173.8. After spectrum fusion, the image accuracy is improved on 43.7% to 79.5%. The reduced variation coefficient can lead to the improved data stability. This paper can provide a new solution to the problem of poor image accuracy caused by greyscale variation of long-term close-range photogrammetry.

Key words: close-range photogrammetry, long time series monitoring, image greyscale, optimal grayscale interval, spectrum fusion