With the rapid development of digital pathology, high-resolution and large field stitching of pathological slice images is of great significance in clinical diagnosis, tissue analysis, and research. Through image stitching technology, single-view images can be stitched into full-view digital slice images. However, existing stitching algorithms face problems such as high computational complexity, large stitching errors, and detail losses when processing large-scale pathological slice images, which limits their effectiveness in practical medical applications. To solve the above problems, a fast stitching algorithm for microscopic images of pathological slices is designed. Firstly, the phase correlation method and neighborhood search are employed for registration. Then, graph theory models are used to optimize the stitching path. Finally, an improved trigonometric weighting method is applied to achieve image fusion, obtaining high-quality pathological slice images with a complete field of view. The experimental results show that for test images with a resolution of 2.3 million, the registration accuracy of the proposed algorithm is better than 4 pixels, and the stitching speed exceeds 20 f/s.