In digital photogrammetry, automatic tie-point matching is necessary for photo triangulation, integrated sensor orientation, epipolar image generation, image registration, image transformation, image stitching, surface modeling, etc. The objective for this research is to find out a cost-effective way to perform automatic tie-point matching in photo triangulation. SURF (Speed-Up Robust Features) is a scale and rotation invariant feature extraction operator that is suitable for feature point matching even a certain degree of geometrical distortion exists. In this paper, based on the SURF image matching scheme, we develop three tie-point matching strategies and a robust estimation filter to remove blunder error. In the meantime, multiple images measurement is considered to increase the redundancy and internal reliability during least-squares bundle adjustment. In the experiment, the performance of the developed tie-point matching scheme is evaluated by the posterior standard error (sigma0) of image coordinate measurement after free network bundle adjustment. Images acquired from different platforms and purposes are test, such as vertical aerial DSLR images in urban area, oblique aerial DSLR images, unmanned aerial vehicle images taken in mountainous area, close-range object images. Experimental results show that the proposed scheme can match abundant of tie points and the sigma0 is less than 1/3 pixels. It demonstrates that the feasibility of the developed algorithm is high and is suitable for most of the photogrammetric applications.