Aiming at the problem of limited measurement area in digital holography, a splicing algorithm based on optimized Harris corner algorithm was proposed to realize phase bidirectional splicing. In the process of acquiring digital holographic images, it was ensured that adjacent regions had a certain overlapping portion, and the phase images of the sub-apertures of the obtained objects were spliced; the Harris corner algorithm was used in the splicing to select the corner-dense regions as matching templates, which was efficient and accurate to determine the overlapping region, combined the Gaussian scale space and the pyramid matching idea to optimize the algorithm, and the phase splicing of the 3-D surface reconstruction was realized by weighted fusion. Taking the glass template as the experimental object, the bidirectional splicing of the phase of the object reproduction was completed. The experimental results show that the splicing method could effectively enlarge the measurement area of digital holographic objects and ensured high splicing accuracy.