For nearly all atomic force microscopies (AFMs) utilized now, only the signals in trace scanning process are employed to reconstruct the sample surface topography, while the data of retrace scanning process are just used for adjustment. In this paper, a novel trace and retrace scanning combined (TRSC) topography reconstruction strategy, which successfully utilizes data from both trace and retrace processes, is proposed for AFMs to increase the surface reconstruction accuracy. Specifically, two reconstructed topography images are obtained, one is for trace scanning, and the other is for retrace scanning; the hysteresis distortion is successfully compensated with a data fusion-based postprocessing method; the alignment for trace and retrace images is further guaranteed with a feature point-based strategy; then the two images are combined together with confidence levels to reconstruct the final accurate topography image. Compared with the conventional scanning method, the proposed TRSC imaging algorithm provides much more accurate image for the sample surface, and it is especially valid for high-speed scanning tasks. Some simulation and experimental results are included to demonstrate the superior performance of the proposed imaging method.